diff --git a/apps/website/i18n/course-ladder.ru.json b/apps/website/i18n/course-ladder.ru.json
index 81b54a0876..24df093478 100644
--- a/apps/website/i18n/course-ladder.ru.json
+++ b/apps/website/i18n/course-ladder.ru.json
@@ -1,5 +1,5 @@
 {
-  "SAY_LADDER": "Лестница: первые три курса бесплатные, четвёртый и все следующие открываются с клубом. Какие три — выбираете вы.",
+  "SAY_LADDER": "Лестница: первые три курса бесплатные и без регистрации, четвёртый и все следующие открываются с клубом. Какие три — выбираете вы.",
   "SAY_RUNG": "Ступень {0} вашей лестницы",
   "SAY_FREE_TITLE": "Этот курс для вас бесплатный",
   "SAY_FREE_LINE": "Откройте одним нажатием. Ничего не спрашиваем, из браузера ничего не уходит.",
@@ -19,7 +19,7 @@
   "SAY_SHARED": "Ваши пройденные уроки видит автор курса.",
   "SAY_COURSES": "Все курсы",
   "SAY_CATALOG_TITLE": "Курсы",
-  "SAY_CATALOG_LEAD": "Каждый курс — 27 коротких уроков в браузере, в каждом живой виджет или спека, которую можно запустить. Любые три — бесплатно, остальные открывает клуб.",
+  "SAY_CATALOG_LEAD": "Каждый курс — 27 коротких уроков в браузере, в каждом живой виджет или спека, которую можно запустить. Любые три — бесплатно, без регистрации и почты; остальные открывает клуб.",
   "SAY_FREE_LEFT": "Осталось бесплатных курсов: {0} из 3",
   "SAY_FREE_NONE": "Три бесплатных курса выбраны. Все остальные открываются с клубом.",
   "SAY_IN_CLUB": "Вы в клубе: открыты все курсы.",
@@ -41,5 +41,11 @@
   "SAY_PAY_WAITING": "Ждём подтверждения оплаты...",
   "SAY_PAY_SLOW": "Оплата ещё не подтверждена. Если вы оплатили, страница продолжает проверять; можно обновить её позже.",
   "SAY_PAY_DONE": "Оплата получена. Добро пожаловать в клуб.",
-  "SAY_PAY_OR_BOT": "Или оплатить звёздами Telegram в нашем боте"
+  "SAY_PAY_OR_BOT": "Или оплатить звёздами Telegram в нашем боте",
+  "SAY_MORE_TITLE": "Другие курсы",
+  "SAY_PROMO_TITLE": "Три курса бесплатно, без регистрации",
+  "SAY_PROMO_LINE": "Каждый урок идёт прямо в браузере, с живым виджетом или спецификацией, которую можно запустить. Любые три курса бесплатно, без регистрации.",
+  "SAY_PROMO_PRESS": "Смотреть курсы",
+  "SAY_SIGN_IN": "Уже в клубе? Войти через Telegram",
+  "SAY_SIGNED_IN": "Вы вошли: ключ клуба сохранён в этом браузере."
 }
diff --git a/apps/website/public/learn/a-bar-detector/index.html b/apps/website/public/learn/a-bar-detector/index.html
index 7d21e38238..35b01a1f8a 100644
--- a/apps/website/public/learn/a-bar-detector/index.html
+++ b/apps/website/public/learn/a-bar-detector/index.html
@@ -123,5 +123,6 @@ <h1>A bar detector</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Three values</h3><p>Why three, how balanced ternary writes every number without a sign, and what flipping and cutting trits do.</p><ol><li><a href="/learn/why-three/">01 Why three</a></li><li><a href="/learn/balanced-ternary/">02 Balanced ternary</a></li><li><a href="/learn/flip-and-cut/">03 Flip to negate, cut to round</a></li></ol></li><li><h3>Module 2 · Logic with unknown</h3><p>Kleene's three-valued gates, two gates binary has no twin for, and addition as a pair of tables.</p><ol><li><a href="/learn/kleene-logic/">04 Three-valued logic</a></li><li><a href="/learn/consensus-and-any/">05 Consensus and any</a></li><li><a href="/learn/sum-and-carry/">06 Sum and carry</a></li></ol></li><li><h3>Module 3 · Adding trits</h3><p>The half adder, the full adder and a carry that ripples left, one place at a time.</p><ol><li><a href="/learn/half-adder/">07 The half adder</a></li><li><a href="/learn/full-adder/">08 The full adder</a></li><li><a href="/learn/ripple-carry/">09 Ripple carry</a></li></ol></li><li><h3>Module 4 · Multiplying without a multiplier</h3><p>Copy, drop or flip: a product by one trit, long multiplication, and a MAC that only adds.</p><ol><li><a href="/learn/times-a-trit/">10 Times one trit</a></li><li><a href="/learn/long-multiplication/">11 Long multiplication</a></li><li><a href="/learn/ternary-mac/">12 A ternary MAC</a></li></ol></li><li><h3>Module 5 · Trits in binary memory</h3><p>Two bits per trit, five trits per byte, and the bits a 27-trit word needs.</p><ol><li><a href="/learn/two-bits-per-trit/">13 Two bits per trit</a></li><li><a href="/learn/five-trits-a-byte/">14 Five trits in a byte</a></li><li><a href="/learn/a-word-of-27/">15 A word of 27 trits</a></li></ol></li><li><h3>Module 6 · The TRI-27 instruction word</h3><p>The 32-bit word the Trinity emulator decodes: its fields, its 47 opcodes and its 15-bit immediate.</p><ol><li><a href="/learn/the-instruction-word/">16 The instruction word</a></li><li><a href="/learn/the-opcode-map/">17 The opcode map</a></li><li><a href="/learn/the-immediate/">18 The immediate</a></li></ol></li><li><h3>Module 7 · A ternary machine</h3><p>An ALU built from this course's adders, a three-way jump, and a program you can step.</p><ol><li><a href="/learn/a-ternary-alu/">19 A ternary ALU</a></li><li><a href="/learn/three-way-jump/">20 A three-way jump</a></li><li><a href="/learn/run-a-program/">21 Run a program</a></li></ol></li><li><h3>Module 8 · Three answers</h3><p>Compare at the highest differing trit, find a number in thirds, sort with three-way compares.</p><ol><li><a href="/learn/compare-at-the-top/">22 Compare at the top</a></li><li><a href="/learn/questions-in-thirds/">23 Questions in thirds</a></li><li><a href="/learn/sort-by-three/">24 Sort with three-way compares</a></li></ol></li><li><h3>Module 9 · Ternary neurons</h3><p>Weights of -1, 0, +1: one neuron, a detector and a small layer, with no multiplier anywhere.</p><ol><li><a href="/learn/one-ternary-neuron/">25 One ternary neuron</a></li><li class="on" aria-current="page"><a href="/learn/a-bar-detector/">26 A bar detector</a></li><li><a href="/learn/a-tiny-layer/">27 A tiny layer</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/a-bend-at-zero/index.html b/apps/website/public/learn/a-bend-at-zero/index.html
index 3cbdbe9ad4..1d2141bf1b 100644
--- a/apps/website/public/learn/a-bend-at-zero/index.html
+++ b/apps/website/public/learn/a-bend-at-zero/index.html
@@ -94,5 +94,6 @@ <h1>A bend at zero</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Lab: our own research</h3><p>A number format of our own, an honest scoreboard, and a model's tables multiplied on the board.</p><ol><li><a href="/learn/t27-number-format/">01 T27: a number format of our own</a></li><li><a href="/learn/an-honest-scoreboard/">02 An honest scoreboard</a></li><li><a href="/learn/a-model-on-the-board/">03 A model's tables on the board</a></li></ol></li><li><h3>Module 2 · AI numbers: the MX block</h3><p>How AI chips keep weights in a few bits: one shared scale per block, the scale byte itself, and what one outlier does to its neighbours.</p><ol><li><a href="/learn/one-scale-per-block/">04 One scale for a block of weights</a></li><li><a href="/learn/scale-byte-native/">05 The scale byte on a real machine</a></li><li><a href="/learn/one-outlier/">06 One outlier, many zeros</a></li></ol></li><li><h3>Module 3 · Ternary weights</h3><p>Weights that are only minus, zero or plus a scale, the five rules a ternary alphabet must pass, and a test pass that checked nothing.</p><ol><li><a href="/learn/phi-scaled-ternary/">07 Weights of minus phi, zero and plus phi</a></li><li><a href="/learn/a-pass-that-checked-nothing/">08 A pass that checked nothing</a></li><li><a href="/learn/five-rules-for-weights/">09 Five rules for a weight alphabet</a></li></ol></li><li><h3>Module 4 · The Ternary Network Float</h3><p>A rule the compiler enforces before any test runs, and a 17-bit float whose exponent is four balanced trits.</p><ol><li><a href="/learn/blocked-is-not-failed/">10 Blocked is not failed</a></li><li><a href="/learn/four-trit-exponent/">11 A float whose exponent is four trits</a></li><li><a href="/learn/negate-twice/">12 Negate twice, get the number back</a></li></ol></li><li><h3>Module 5 · Arithmetic on signed numbers</h3><p>Multiply two signed numbers, add them when their signs differ, and do both at once in a multiply-accumulate.</p><ol><li><a href="/learn/a-sign-is-one-bit/">13 A sign is one bit</a></li><li><a href="/learn/opposite-signs-subtract/">14 Opposite signs subtract</a></li><li><a href="/learn/multiply-then-add/">15 Multiply, then add</a></li></ol></li><li><h3>Module 6 · Parts of a neuron</h3><p>A ReLU that bends at zero, a power of two for softmax, and an argmax that names the answer.</p><ol><li class="on" aria-current="page"><a href="/learn/a-bend-at-zero/">16 A bend at zero</a></li><li><a href="/learn/two-to-the-x/">17 Two to the x</a></li><li><a href="/learn/pick-the-largest/">18 Pick the largest</a></li></ol></li><li><h3>Module 7 · Learning from a mistake</h3><p>A loss that prices a wrong guess in bits, one step that moves a weight against its gradient, and the hidden layer that XOR needs.</p><ol><li><a href="/learn/a-loss-in-bits/">19 A loss in bits</a></li><li><a href="/learn/one-step-downhill/">20 One step downhill</a></li><li><a href="/learn/xor-needs-a-bend/">21 XOR needs a bend</a></li></ol></li><li><h3>Module 8 · BitNet: ternary networks</h3><p>A threshold that squeezes a sum back to three values, one neuron that becomes a different function when its weights change, and a neuron that reads its inputs 27 trits at a time.</p><ol><li><a href="/learn/back-to-three-values/">22 Back to three values</a></li><li><a href="/learn/same-neuron-new-weights/">23 Same neuron, new weights</a></li><li><a href="/learn/a-neuron-in-chunks/">24 A neuron in chunks</a></li></ol></li><li><h3>Module 9 · The ternary MAC as a chip</h3><p>The 27-trit dot product as wires with no register, the same sum added into a register on every clock, and a small whole network to close the course.</p><ol><li><a href="/learn/a-dot-product-in-wires/">25 A dot product made of wires</a></li><li><a href="/learn/add-it-up-every-clock/">26 Add it up on every clock</a></li><li><a href="/learn/a-whole-network/">27 A whole network</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/a-build-receipt/index.html b/apps/website/public/learn/a-build-receipt/index.html
index 2802ea8a70..706b7795dd 100644
--- a/apps/website/public/learn/a-build-receipt/index.html
+++ b/apps/website/public/learn/a-build-receipt/index.html
@@ -170,5 +170,6 @@ <h1>A build, itemised</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The chip</h3><p>What an FPGA is, which chip we use, and how its pins meet the board.</p><ol><li><a href="/learn/fpga-is-a-grid/">01 An FPGA is a grid you configure</a></li><li><a href="/learn/which-chip/">02 Which chip is this?</a></li><li><a href="/learn/pins-and-package/">03 Pins, balls and the package</a></li></ol></li><li><h3>Module 2 · Numbers in hardware</h3><p>Bits, trits and number formats, and what arithmetic costs in logic.</p><ol><li><a href="/learn/bits-and-trits/">04 Bits, trits and GF16</a></li><li><a href="/learn/what-arithmetic-costs/">05 What arithmetic costs</a></li><li><a href="/learn/how-much-fits/">06 How much fits on the chip</a></li></ol></li><li><h3>Module 3 · Your t27 program</h3><p>Write a spec, test it, and see why compiling is not the same as being right.</p><ol><li><a href="/learn/hello-t27/">07 Hello, t27</a></li><li><a href="/learn/tests-are-the-spec/">08 Tests are part of the spec</a></li><li><a href="/learn/compiles-is-not-correct/">09 Compiles is not the same as correct</a></li></ol></li><li><h3>Module 4 · Inside t27c</h3><p>How the compiler reads a spec and what it writes, including the native t27b.</p><ol><li><a href="/learn/compiler-stages/">10 The stages of t27c</a></li><li><a href="/learn/t27c-on-a-real-machine/">11 t27c on a real machine</a></li><li><a href="/learn/seven-backends-and-t27b/">12 Seven backends, and t27b</a></li></ol></li><li><h3>Module 5 · From spec to hardware</h3><p>The Verilog t27c writes, the cells it becomes, and what one LUT does.</p><ol><li><a href="/learn/spec-to-verilog/">13 From a spec to Verilog</a></li><li><a href="/learn/count-the-cells/">14 Count the cells</a></li><li><a href="/learn/one-lut-up-close/">15 One LUT, up close</a></li></ol></li><li><h3>Module 6 · Reading synthesis</h3><p>What yosys reports about your design, and which warnings matter.</p><ol><li><a href="/learn/synthesis-warnings/">16 Synthesis warnings that matter</a></li><li><a href="/learn/state-machines/">17 State machines</a></li><li><a href="/learn/where-the-luts-go/">18 Where the LUTs go</a></li></ol></li><li><h3>Module 7 · Place, route, timing</h3><p>Where the cells land on the die, and whether the clock is met.</p><ol><li><a href="/learn/placement/">19 Placement: where the cells land</a></li><li><a href="/learn/logic-depth/">20 Logic depth</a></li><li><a href="/learn/timing-slack/">21 Timing and slack</a></li></ol></li><li><h3>Module 8 · The bitstream</h3><p>How a routed design becomes the bits the chip loads, with no Vivado.</p><ol><li><a href="/learn/fasm-to-frames/">22 From FASM to frames</a></li><li class="on" aria-current="page"><a href="/learn/a-build-receipt/">23 A build, itemised</a></li><li><a href="/learn/read-a-bitstream/">24 Read a bitstream</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>Ask the chip who it is, load the bits, and check them.</p><ol><li><a href="/learn/jtag-asks-the-chip/">25 JTAG asks the chip</a></li><li><a href="/learn/same-bits-on-the-board/">26 The same bits, on the board</a></li><li><a href="/learn/capstone/">27 Capstone: out-guess yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/a-correlator-finds-the-frame/index.html b/apps/website/public/learn/a-correlator-finds-the-frame/index.html
index 40d6364d95..2f73933559 100644
--- a/apps/website/public/learn/a-correlator-finds-the-frame/index.html
+++ b/apps/website/public/learn/a-correlator-finds-the-frame/index.html
@@ -284,5 +284,6 @@ <h1>A correlator finds the frame</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Adding numbers</h3><p>Why addition is slow: the carry that walks up the word, the dedicated chain in every slice, prefix networks and carry-save trees.</p><ol><li><a href="/learn/the-carry-chain/">01 The carry chain</a></li><li><a href="/learn/prefix-adders/">02 Looking ahead: prefix adders</a></li><li><a href="/learn/carry-save/">03 Three numbers at once</a></li></ol></li><li><h3>Module 2 · Multiplying</h3><p>A product is a sum of shifted copies; Booth recoding halves them, and the DSP48E1 slice does the rest in one block.</p><ol><li><a href="/learn/shift-and-add/">04 Shift and add</a></li><li><a href="/learn/booths-trick/">05 Booth's trick</a></li><li><a href="/learn/the-dsp48e1-slice/">06 The DSP48E1 slice</a></li></ol></li><li><h3>Module 3 · Fixed point</h3><p>Where the binary point sits, what rounding does to a value and to its average, and what each bit of a quantizer buys.</p><ol><li><a href="/learn/q-formats/">07 Q formats</a></li><li><a href="/learn/rounding-and-saturation/">08 Rounding and saturation</a></li><li><a href="/learn/quantization-noise/">09 Quantization noise</a></li></ol></li><li><h3>Module 4 · Functions in hardware</h3><p>Sine, cosine, angle and length from shifts and adds, and when a table is the better answer.</p><ol><li><a href="/learn/rotating-by-shifts/">10 Rotating by shifts</a></li><li><a href="/learn/angle-and-magnitude/">11 Angle and magnitude</a></li><li><a href="/learn/tables-and-interpolation/">12 Tables and interpolation</a></li></ol></li><li><h3>Module 5 · Signals and sampling</h3><p>What sampling does to a frequency, how an oscillator is built from an adder, and what a DFT bin measures.</p><ol><li><a href="/learn/sampling-and-aliasing/">13 Sampling and aliasing</a></li><li><a href="/learn/the-nco/">14 A numerically controlled oscillator</a></li><li><a href="/learn/bins-of-a-spectrum/">15 Bins of a spectrum</a></li></ol></li><li><h3>Module 6 · FIR filters</h3><p>The moving average, a windowed-sinc design with integer taps, and folding the taps onto DSP slices.</p><ol><li><a href="/learn/the-moving-average/">16 The moving average</a></li><li><a href="/learn/designing-taps/">17 Designing taps</a></li><li><a href="/learn/folding-taps-onto-slices/">18 Folding taps onto slices</a></li></ol></li><li><h3>Module 7 · Multirate</h3><p>Lowering the sample rate without folding noise in: decimation, the CIC filter and the polyphase form.</p><ol><li><a href="/learn/lowering-the-rate/">19 Lowering the rate</a></li><li><a href="/learn/the-cic-filter/">20 The CIC filter</a></li><li><a href="/learn/polyphase/">21 Polyphase</a></li></ol></li><li><h3>Module 8 · The FFT</h3><p>N log N instead of N^2: butterflies, the bit-reversed input order and the bits each stage adds.</p><ol><li><a href="/learn/the-butterfly/">22 The butterfly</a></li><li><a href="/learn/twiddles-and-bit-reversal/">23 Twiddles and bit reversal</a></li><li><a href="/learn/growth-through-the-stages/">24 Growth through the stages</a></li></ol></li><li><h3>Module 9 · On the bench</h3><p>From the arithmetic to the board: a correlator from the modem spec, a budget of slices and timing, and the capstone filter.</p><ol><li class="on" aria-current="page"><a href="/learn/a-correlator-finds-the-frame/">25 A correlator finds the frame</a></li><li><a href="/learn/does-it-fit-and-close-timing/">26 Does it fit and close timing?</a></li><li><a href="/learn/capstone-a-filter-on-the-board/">27 Capstone: a filter on the board</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/a-dot-product-in-wires/index.html b/apps/website/public/learn/a-dot-product-in-wires/index.html
index 0f04be8584..fc1684a800 100644
--- a/apps/website/public/learn/a-dot-product-in-wires/index.html
+++ b/apps/website/public/learn/a-dot-product-in-wires/index.html
@@ -113,5 +113,6 @@ <h1>A dot product made of wires</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Lab: our own research</h3><p>A number format of our own, an honest scoreboard, and a model's tables multiplied on the board.</p><ol><li><a href="/learn/t27-number-format/">01 T27: a number format of our own</a></li><li><a href="/learn/an-honest-scoreboard/">02 An honest scoreboard</a></li><li><a href="/learn/a-model-on-the-board/">03 A model's tables on the board</a></li></ol></li><li><h3>Module 2 · AI numbers: the MX block</h3><p>How AI chips keep weights in a few bits: one shared scale per block, the scale byte itself, and what one outlier does to its neighbours.</p><ol><li><a href="/learn/one-scale-per-block/">04 One scale for a block of weights</a></li><li><a href="/learn/scale-byte-native/">05 The scale byte on a real machine</a></li><li><a href="/learn/one-outlier/">06 One outlier, many zeros</a></li></ol></li><li><h3>Module 3 · Ternary weights</h3><p>Weights that are only minus, zero or plus a scale, the five rules a ternary alphabet must pass, and a test pass that checked nothing.</p><ol><li><a href="/learn/phi-scaled-ternary/">07 Weights of minus phi, zero and plus phi</a></li><li><a href="/learn/a-pass-that-checked-nothing/">08 A pass that checked nothing</a></li><li><a href="/learn/five-rules-for-weights/">09 Five rules for a weight alphabet</a></li></ol></li><li><h3>Module 4 · The Ternary Network Float</h3><p>A rule the compiler enforces before any test runs, and a 17-bit float whose exponent is four balanced trits.</p><ol><li><a href="/learn/blocked-is-not-failed/">10 Blocked is not failed</a></li><li><a href="/learn/four-trit-exponent/">11 A float whose exponent is four trits</a></li><li><a href="/learn/negate-twice/">12 Negate twice, get the number back</a></li></ol></li><li><h3>Module 5 · Arithmetic on signed numbers</h3><p>Multiply two signed numbers, add them when their signs differ, and do both at once in a multiply-accumulate.</p><ol><li><a href="/learn/a-sign-is-one-bit/">13 A sign is one bit</a></li><li><a href="/learn/opposite-signs-subtract/">14 Opposite signs subtract</a></li><li><a href="/learn/multiply-then-add/">15 Multiply, then add</a></li></ol></li><li><h3>Module 6 · Parts of a neuron</h3><p>A ReLU that bends at zero, a power of two for softmax, and an argmax that names the answer.</p><ol><li><a href="/learn/a-bend-at-zero/">16 A bend at zero</a></li><li><a href="/learn/two-to-the-x/">17 Two to the x</a></li><li><a href="/learn/pick-the-largest/">18 Pick the largest</a></li></ol></li><li><h3>Module 7 · Learning from a mistake</h3><p>A loss that prices a wrong guess in bits, one step that moves a weight against its gradient, and the hidden layer that XOR needs.</p><ol><li><a href="/learn/a-loss-in-bits/">19 A loss in bits</a></li><li><a href="/learn/one-step-downhill/">20 One step downhill</a></li><li><a href="/learn/xor-needs-a-bend/">21 XOR needs a bend</a></li></ol></li><li><h3>Module 8 · BitNet: ternary networks</h3><p>A threshold that squeezes a sum back to three values, one neuron that becomes a different function when its weights change, and a neuron that reads its inputs 27 trits at a time.</p><ol><li><a href="/learn/back-to-three-values/">22 Back to three values</a></li><li><a href="/learn/same-neuron-new-weights/">23 Same neuron, new weights</a></li><li><a href="/learn/a-neuron-in-chunks/">24 A neuron in chunks</a></li></ol></li><li><h3>Module 9 · The ternary MAC as a chip</h3><p>The 27-trit dot product as wires with no register, the same sum added into a register on every clock, and a small whole network to close the course.</p><ol><li class="on" aria-current="page"><a href="/learn/a-dot-product-in-wires/">25 A dot product made of wires</a></li><li><a href="/learn/add-it-up-every-clock/">26 Add it up on every clock</a></li><li><a href="/learn/a-whole-network/">27 A whole network</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/a-float-cut-by-phi/index.html b/apps/website/public/learn/a-float-cut-by-phi/index.html
index d647cff31d..0ad865ec78 100644
--- a/apps/website/public/learn/a-float-cut-by-phi/index.html
+++ b/apps/website/public/learn/a-float-cut-by-phi/index.html
@@ -629,5 +629,6 @@ <h1>A float cut by phi</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The rule and its numbers</h3><p>One rule splits every width, the ratio it aims at, and the Lucas numbers behind the 3.</p><ol><li class="on" aria-current="page"><a href="/learn/a-float-cut-by-phi/">01 A float cut by phi</a></li><li><a href="/learn/phi-as-a-ratio/">02 Phi as a ratio</a></li><li><a href="/learn/lucas-numbers-stay-whole/">03 Lucas numbers stay whole</a></li></ol></li><li><h3>Module 2 · Why phi, why three</h3><p>Why the split is phi, why base three, and how a spec checks GF16 keeps phi.</p><ol><li><a href="/learn/why-the-split-is-phi/">04 Why the split is phi</a></li><li><a href="/learn/why-base-three/">05 Why base three</a></li><li><a href="/learn/phi-in-sixteen-bits/">06 Phi in sixteen bits</a></li></ol></li><li><h3>Module 3 · The small rungs: GF4 to GF8</h3><p>GF4, GF6 and GF8, the fewest bits, where rounding to whole bits costs the most.</p><ol><li><a href="/learn/gf4-four-bits/">07 GF4: four bits</a></li><li><a href="/learn/gf6-six-bits/">08 GF6: six bits</a></li><li><a href="/learn/gf8-one-byte/">09 GF8: one byte</a></li></ol></li><li><h3>Module 4 · Ten to fourteen bits</h3><p>GF10, GF12 and GF14, and how the distance from 1 / phi moves as the word grows.</p><ol><li><a href="/learn/gf10-ten-bits/">10 GF10: ten bits</a></li><li><a href="/learn/gf12-twelve-bits/">11 GF12: twelve bits</a></li><li><a href="/learn/gf14-fourteen-bits/">12 GF14: fourteen bits</a></li></ol></li><li><h3>Module 5 · GF16 at work</h3><p>The primary 16-bit format, a two-term dot product in GF-T16, then GF20 and GF24.</p><ol><li><a href="/learn/gf16-the-primary/">13 GF16: the primary format</a></li><li><a href="/learn/two-products-one-sum/">14 Two products, one sum</a></li><li><a href="/learn/gf24-twenty-four-bits/">15 GF20 and GF24</a></li></ol></li><li><h3>Module 6 · GF32 to GF64</h3><p>GF32 beside IEEE single, GF48 with no IEEE twin, GF64 beside IEEE double.</p><ol><li><a href="/learn/gf32-a-single/">16 GF32: a single</a></li><li><a href="/learn/gf48-forty-eight-bits/">17 GF48: forty-eight bits</a></li><li><a href="/learn/gf64-a-double/">18 GF64: a double</a></li></ol></li><li><h3>Module 7 · GF96 to GF256</h3><p>GF96, GF128 and GF256, where the specs hold the layout with invariants.</p><ol><li><a href="/learn/gf96-ninety-six-bits/">19 GF96: ninety-six bits</a></li><li><a href="/learn/gf128-a-quad/">20 GF128: a quad</a></li><li><a href="/learn/gf256-two-five-six/">21 GF256: two hundred fifty-six bits</a></li></ol></li><li><h3>Module 8 · The widest rungs, then trits</h3><p>GF512 and GF1024, the two widest rungs, then GF-T8, where the exponent moves to trits.</p><ol><li><a href="/learn/gf512-five-twelve/">22 GF512: five hundred twelve bits</a></li><li><a href="/learn/gf1024-one-kilobit/">23 GF1024: one kilobit</a></li><li><a href="/learn/gft8-exponent-in-trits/">24 GF-T8: an exponent in trits</a></li></ol></li><li><h3>Module 9 · More trits, then the decode</h3><p>GF-T16 and GF-T32, then why fixed fields decode in parallel and a posit does not.</p><ol><li><a href="/learn/gft16-six-trits/">25 GF-T16: six trits</a></li><li><a href="/learn/gft32-twelve-trits/">26 GF-T32: twelve trits</a></li><li><a href="/learn/decode-in-one-step/">27 Decode in one step</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/a-loss-in-bits/index.html b/apps/website/public/learn/a-loss-in-bits/index.html
index 38763cc920..db43b3a6ea 100644
--- a/apps/website/public/learn/a-loss-in-bits/index.html
+++ b/apps/website/public/learn/a-loss-in-bits/index.html
@@ -172,5 +172,6 @@ <h1>A loss in bits</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Lab: our own research</h3><p>A number format of our own, an honest scoreboard, and a model's tables multiplied on the board.</p><ol><li><a href="/learn/t27-number-format/">01 T27: a number format of our own</a></li><li><a href="/learn/an-honest-scoreboard/">02 An honest scoreboard</a></li><li><a href="/learn/a-model-on-the-board/">03 A model's tables on the board</a></li></ol></li><li><h3>Module 2 · AI numbers: the MX block</h3><p>How AI chips keep weights in a few bits: one shared scale per block, the scale byte itself, and what one outlier does to its neighbours.</p><ol><li><a href="/learn/one-scale-per-block/">04 One scale for a block of weights</a></li><li><a href="/learn/scale-byte-native/">05 The scale byte on a real machine</a></li><li><a href="/learn/one-outlier/">06 One outlier, many zeros</a></li></ol></li><li><h3>Module 3 · Ternary weights</h3><p>Weights that are only minus, zero or plus a scale, the five rules a ternary alphabet must pass, and a test pass that checked nothing.</p><ol><li><a href="/learn/phi-scaled-ternary/">07 Weights of minus phi, zero and plus phi</a></li><li><a href="/learn/a-pass-that-checked-nothing/">08 A pass that checked nothing</a></li><li><a href="/learn/five-rules-for-weights/">09 Five rules for a weight alphabet</a></li></ol></li><li><h3>Module 4 · The Ternary Network Float</h3><p>A rule the compiler enforces before any test runs, and a 17-bit float whose exponent is four balanced trits.</p><ol><li><a href="/learn/blocked-is-not-failed/">10 Blocked is not failed</a></li><li><a href="/learn/four-trit-exponent/">11 A float whose exponent is four trits</a></li><li><a href="/learn/negate-twice/">12 Negate twice, get the number back</a></li></ol></li><li><h3>Module 5 · Arithmetic on signed numbers</h3><p>Multiply two signed numbers, add them when their signs differ, and do both at once in a multiply-accumulate.</p><ol><li><a href="/learn/a-sign-is-one-bit/">13 A sign is one bit</a></li><li><a href="/learn/opposite-signs-subtract/">14 Opposite signs subtract</a></li><li><a href="/learn/multiply-then-add/">15 Multiply, then add</a></li></ol></li><li><h3>Module 6 · Parts of a neuron</h3><p>A ReLU that bends at zero, a power of two for softmax, and an argmax that names the answer.</p><ol><li><a href="/learn/a-bend-at-zero/">16 A bend at zero</a></li><li><a href="/learn/two-to-the-x/">17 Two to the x</a></li><li><a href="/learn/pick-the-largest/">18 Pick the largest</a></li></ol></li><li><h3>Module 7 · Learning from a mistake</h3><p>A loss that prices a wrong guess in bits, one step that moves a weight against its gradient, and the hidden layer that XOR needs.</p><ol><li class="on" aria-current="page"><a href="/learn/a-loss-in-bits/">19 A loss in bits</a></li><li><a href="/learn/one-step-downhill/">20 One step downhill</a></li><li><a href="/learn/xor-needs-a-bend/">21 XOR needs a bend</a></li></ol></li><li><h3>Module 8 · BitNet: ternary networks</h3><p>A threshold that squeezes a sum back to three values, one neuron that becomes a different function when its weights change, and a neuron that reads its inputs 27 trits at a time.</p><ol><li><a href="/learn/back-to-three-values/">22 Back to three values</a></li><li><a href="/learn/same-neuron-new-weights/">23 Same neuron, new weights</a></li><li><a href="/learn/a-neuron-in-chunks/">24 A neuron in chunks</a></li></ol></li><li><h3>Module 9 · The ternary MAC as a chip</h3><p>The 27-trit dot product as wires with no register, the same sum added into a register on every clock, and a small whole network to close the course.</p><ol><li><a href="/learn/a-dot-product-in-wires/">25 A dot product made of wires</a></li><li><a href="/learn/add-it-up-every-clock/">26 Add it up on every clock</a></li><li><a href="/learn/a-whole-network/">27 A whole network</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/a-model-on-the-board/index.html b/apps/website/public/learn/a-model-on-the-board/index.html
index fd9d13108b..b9caad40ea 100644
--- a/apps/website/public/learn/a-model-on-the-board/index.html
+++ b/apps/website/public/learn/a-model-on-the-board/index.html
@@ -316,5 +316,6 @@ <h1>A model's tables on the board</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Lab: our own research</h3><p>A number format of our own, an honest scoreboard, and a model's tables multiplied on the board.</p><ol><li><a href="/learn/t27-number-format/">01 T27: a number format of our own</a></li><li><a href="/learn/an-honest-scoreboard/">02 An honest scoreboard</a></li><li class="on" aria-current="page"><a href="/learn/a-model-on-the-board/">03 A model's tables on the board</a></li></ol></li><li><h3>Module 2 · AI numbers: the MX block</h3><p>How AI chips keep weights in a few bits: one shared scale per block, the scale byte itself, and what one outlier does to its neighbours.</p><ol><li><a href="/learn/one-scale-per-block/">04 One scale for a block of weights</a></li><li><a href="/learn/scale-byte-native/">05 The scale byte on a real machine</a></li><li><a href="/learn/one-outlier/">06 One outlier, many zeros</a></li></ol></li><li><h3>Module 3 · Ternary weights</h3><p>Weights that are only minus, zero or plus a scale, the five rules a ternary alphabet must pass, and a test pass that checked nothing.</p><ol><li><a href="/learn/phi-scaled-ternary/">07 Weights of minus phi, zero and plus phi</a></li><li><a href="/learn/a-pass-that-checked-nothing/">08 A pass that checked nothing</a></li><li><a href="/learn/five-rules-for-weights/">09 Five rules for a weight alphabet</a></li></ol></li><li><h3>Module 4 · The Ternary Network Float</h3><p>A rule the compiler enforces before any test runs, and a 17-bit float whose exponent is four balanced trits.</p><ol><li><a href="/learn/blocked-is-not-failed/">10 Blocked is not failed</a></li><li><a href="/learn/four-trit-exponent/">11 A float whose exponent is four trits</a></li><li><a href="/learn/negate-twice/">12 Negate twice, get the number back</a></li></ol></li><li><h3>Module 5 · Arithmetic on signed numbers</h3><p>Multiply two signed numbers, add them when their signs differ, and do both at once in a multiply-accumulate.</p><ol><li><a href="/learn/a-sign-is-one-bit/">13 A sign is one bit</a></li><li><a href="/learn/opposite-signs-subtract/">14 Opposite signs subtract</a></li><li><a href="/learn/multiply-then-add/">15 Multiply, then add</a></li></ol></li><li><h3>Module 6 · Parts of a neuron</h3><p>A ReLU that bends at zero, a power of two for softmax, and an argmax that names the answer.</p><ol><li><a href="/learn/a-bend-at-zero/">16 A bend at zero</a></li><li><a href="/learn/two-to-the-x/">17 Two to the x</a></li><li><a href="/learn/pick-the-largest/">18 Pick the largest</a></li></ol></li><li><h3>Module 7 · Learning from a mistake</h3><p>A loss that prices a wrong guess in bits, one step that moves a weight against its gradient, and the hidden layer that XOR needs.</p><ol><li><a href="/learn/a-loss-in-bits/">19 A loss in bits</a></li><li><a href="/learn/one-step-downhill/">20 One step downhill</a></li><li><a href="/learn/xor-needs-a-bend/">21 XOR needs a bend</a></li></ol></li><li><h3>Module 8 · BitNet: ternary networks</h3><p>A threshold that squeezes a sum back to three values, one neuron that becomes a different function when its weights change, and a neuron that reads its inputs 27 trits at a time.</p><ol><li><a href="/learn/back-to-three-values/">22 Back to three values</a></li><li><a href="/learn/same-neuron-new-weights/">23 Same neuron, new weights</a></li><li><a href="/learn/a-neuron-in-chunks/">24 A neuron in chunks</a></li></ol></li><li><h3>Module 9 · The ternary MAC as a chip</h3><p>The 27-trit dot product as wires with no register, the same sum added into a register on every clock, and a small whole network to close the course.</p><ol><li><a href="/learn/a-dot-product-in-wires/">25 A dot product made of wires</a></li><li><a href="/learn/add-it-up-every-clock/">26 Add it up on every clock</a></li><li><a href="/learn/a-whole-network/">27 A whole network</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/a-neuron-in-chunks/index.html b/apps/website/public/learn/a-neuron-in-chunks/index.html
index f0c8395fe7..b755833d35 100644
--- a/apps/website/public/learn/a-neuron-in-chunks/index.html
+++ b/apps/website/public/learn/a-neuron-in-chunks/index.html
@@ -141,5 +141,6 @@ <h1>A neuron in chunks</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Lab: our own research</h3><p>A number format of our own, an honest scoreboard, and a model's tables multiplied on the board.</p><ol><li><a href="/learn/t27-number-format/">01 T27: a number format of our own</a></li><li><a href="/learn/an-honest-scoreboard/">02 An honest scoreboard</a></li><li><a href="/learn/a-model-on-the-board/">03 A model's tables on the board</a></li></ol></li><li><h3>Module 2 · AI numbers: the MX block</h3><p>How AI chips keep weights in a few bits: one shared scale per block, the scale byte itself, and what one outlier does to its neighbours.</p><ol><li><a href="/learn/one-scale-per-block/">04 One scale for a block of weights</a></li><li><a href="/learn/scale-byte-native/">05 The scale byte on a real machine</a></li><li><a href="/learn/one-outlier/">06 One outlier, many zeros</a></li></ol></li><li><h3>Module 3 · Ternary weights</h3><p>Weights that are only minus, zero or plus a scale, the five rules a ternary alphabet must pass, and a test pass that checked nothing.</p><ol><li><a href="/learn/phi-scaled-ternary/">07 Weights of minus phi, zero and plus phi</a></li><li><a href="/learn/a-pass-that-checked-nothing/">08 A pass that checked nothing</a></li><li><a href="/learn/five-rules-for-weights/">09 Five rules for a weight alphabet</a></li></ol></li><li><h3>Module 4 · The Ternary Network Float</h3><p>A rule the compiler enforces before any test runs, and a 17-bit float whose exponent is four balanced trits.</p><ol><li><a href="/learn/blocked-is-not-failed/">10 Blocked is not failed</a></li><li><a href="/learn/four-trit-exponent/">11 A float whose exponent is four trits</a></li><li><a href="/learn/negate-twice/">12 Negate twice, get the number back</a></li></ol></li><li><h3>Module 5 · Arithmetic on signed numbers</h3><p>Multiply two signed numbers, add them when their signs differ, and do both at once in a multiply-accumulate.</p><ol><li><a href="/learn/a-sign-is-one-bit/">13 A sign is one bit</a></li><li><a href="/learn/opposite-signs-subtract/">14 Opposite signs subtract</a></li><li><a href="/learn/multiply-then-add/">15 Multiply, then add</a></li></ol></li><li><h3>Module 6 · Parts of a neuron</h3><p>A ReLU that bends at zero, a power of two for softmax, and an argmax that names the answer.</p><ol><li><a href="/learn/a-bend-at-zero/">16 A bend at zero</a></li><li><a href="/learn/two-to-the-x/">17 Two to the x</a></li><li><a href="/learn/pick-the-largest/">18 Pick the largest</a></li></ol></li><li><h3>Module 7 · Learning from a mistake</h3><p>A loss that prices a wrong guess in bits, one step that moves a weight against its gradient, and the hidden layer that XOR needs.</p><ol><li><a href="/learn/a-loss-in-bits/">19 A loss in bits</a></li><li><a href="/learn/one-step-downhill/">20 One step downhill</a></li><li><a href="/learn/xor-needs-a-bend/">21 XOR needs a bend</a></li></ol></li><li><h3>Module 8 · BitNet: ternary networks</h3><p>A threshold that squeezes a sum back to three values, one neuron that becomes a different function when its weights change, and a neuron that reads its inputs 27 trits at a time.</p><ol><li><a href="/learn/back-to-three-values/">22 Back to three values</a></li><li><a href="/learn/same-neuron-new-weights/">23 Same neuron, new weights</a></li><li class="on" aria-current="page"><a href="/learn/a-neuron-in-chunks/">24 A neuron in chunks</a></li></ol></li><li><h3>Module 9 · The ternary MAC as a chip</h3><p>The 27-trit dot product as wires with no register, the same sum added into a register on every clock, and a small whole network to close the course.</p><ol><li><a href="/learn/a-dot-product-in-wires/">25 A dot product made of wires</a></li><li><a href="/learn/add-it-up-every-clock/">26 Add it up on every clock</a></li><li><a href="/learn/a-whole-network/">27 A whole network</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/a-pass-that-checked-nothing/index.html b/apps/website/public/learn/a-pass-that-checked-nothing/index.html
index e3af584b7a..9cec7764bd 100644
--- a/apps/website/public/learn/a-pass-that-checked-nothing/index.html
+++ b/apps/website/public/learn/a-pass-that-checked-nothing/index.html
@@ -454,5 +454,6 @@ <h1>A pass that checked nothing</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Lab: our own research</h3><p>A number format of our own, an honest scoreboard, and a model's tables multiplied on the board.</p><ol><li><a href="/learn/t27-number-format/">01 T27: a number format of our own</a></li><li><a href="/learn/an-honest-scoreboard/">02 An honest scoreboard</a></li><li><a href="/learn/a-model-on-the-board/">03 A model's tables on the board</a></li></ol></li><li><h3>Module 2 · AI numbers: the MX block</h3><p>How AI chips keep weights in a few bits: one shared scale per block, the scale byte itself, and what one outlier does to its neighbours.</p><ol><li><a href="/learn/one-scale-per-block/">04 One scale for a block of weights</a></li><li><a href="/learn/scale-byte-native/">05 The scale byte on a real machine</a></li><li><a href="/learn/one-outlier/">06 One outlier, many zeros</a></li></ol></li><li><h3>Module 3 · Ternary weights</h3><p>Weights that are only minus, zero or plus a scale, the five rules a ternary alphabet must pass, and a test pass that checked nothing.</p><ol><li><a href="/learn/phi-scaled-ternary/">07 Weights of minus phi, zero and plus phi</a></li><li class="on" aria-current="page"><a href="/learn/a-pass-that-checked-nothing/">08 A pass that checked nothing</a></li><li><a href="/learn/five-rules-for-weights/">09 Five rules for a weight alphabet</a></li></ol></li><li><h3>Module 4 · The Ternary Network Float</h3><p>A rule the compiler enforces before any test runs, and a 17-bit float whose exponent is four balanced trits.</p><ol><li><a href="/learn/blocked-is-not-failed/">10 Blocked is not failed</a></li><li><a href="/learn/four-trit-exponent/">11 A float whose exponent is four trits</a></li><li><a href="/learn/negate-twice/">12 Negate twice, get the number back</a></li></ol></li><li><h3>Module 5 · Arithmetic on signed numbers</h3><p>Multiply two signed numbers, add them when their signs differ, and do both at once in a multiply-accumulate.</p><ol><li><a href="/learn/a-sign-is-one-bit/">13 A sign is one bit</a></li><li><a href="/learn/opposite-signs-subtract/">14 Opposite signs subtract</a></li><li><a href="/learn/multiply-then-add/">15 Multiply, then add</a></li></ol></li><li><h3>Module 6 · Parts of a neuron</h3><p>A ReLU that bends at zero, a power of two for softmax, and an argmax that names the answer.</p><ol><li><a href="/learn/a-bend-at-zero/">16 A bend at zero</a></li><li><a href="/learn/two-to-the-x/">17 Two to the x</a></li><li><a href="/learn/pick-the-largest/">18 Pick the largest</a></li></ol></li><li><h3>Module 7 · Learning from a mistake</h3><p>A loss that prices a wrong guess in bits, one step that moves a weight against its gradient, and the hidden layer that XOR needs.</p><ol><li><a href="/learn/a-loss-in-bits/">19 A loss in bits</a></li><li><a href="/learn/one-step-downhill/">20 One step downhill</a></li><li><a href="/learn/xor-needs-a-bend/">21 XOR needs a bend</a></li></ol></li><li><h3>Module 8 · BitNet: ternary networks</h3><p>A threshold that squeezes a sum back to three values, one neuron that becomes a different function when its weights change, and a neuron that reads its inputs 27 trits at a time.</p><ol><li><a href="/learn/back-to-three-values/">22 Back to three values</a></li><li><a href="/learn/same-neuron-new-weights/">23 Same neuron, new weights</a></li><li><a href="/learn/a-neuron-in-chunks/">24 A neuron in chunks</a></li></ol></li><li><h3>Module 9 · The ternary MAC as a chip</h3><p>The 27-trit dot product as wires with no register, the same sum added into a register on every clock, and a small whole network to close the course.</p><ol><li><a href="/learn/a-dot-product-in-wires/">25 A dot product made of wires</a></li><li><a href="/learn/add-it-up-every-clock/">26 Add it up on every clock</a></li><li><a href="/learn/a-whole-network/">27 A whole network</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/a-sign-is-one-bit/index.html b/apps/website/public/learn/a-sign-is-one-bit/index.html
index f2575a1ea7..b66ceedbd4 100644
--- a/apps/website/public/learn/a-sign-is-one-bit/index.html
+++ b/apps/website/public/learn/a-sign-is-one-bit/index.html
@@ -192,5 +192,6 @@ <h1>A sign is one bit</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Lab: our own research</h3><p>A number format of our own, an honest scoreboard, and a model's tables multiplied on the board.</p><ol><li><a href="/learn/t27-number-format/">01 T27: a number format of our own</a></li><li><a href="/learn/an-honest-scoreboard/">02 An honest scoreboard</a></li><li><a href="/learn/a-model-on-the-board/">03 A model's tables on the board</a></li></ol></li><li><h3>Module 2 · AI numbers: the MX block</h3><p>How AI chips keep weights in a few bits: one shared scale per block, the scale byte itself, and what one outlier does to its neighbours.</p><ol><li><a href="/learn/one-scale-per-block/">04 One scale for a block of weights</a></li><li><a href="/learn/scale-byte-native/">05 The scale byte on a real machine</a></li><li><a href="/learn/one-outlier/">06 One outlier, many zeros</a></li></ol></li><li><h3>Module 3 · Ternary weights</h3><p>Weights that are only minus, zero or plus a scale, the five rules a ternary alphabet must pass, and a test pass that checked nothing.</p><ol><li><a href="/learn/phi-scaled-ternary/">07 Weights of minus phi, zero and plus phi</a></li><li><a href="/learn/a-pass-that-checked-nothing/">08 A pass that checked nothing</a></li><li><a href="/learn/five-rules-for-weights/">09 Five rules for a weight alphabet</a></li></ol></li><li><h3>Module 4 · The Ternary Network Float</h3><p>A rule the compiler enforces before any test runs, and a 17-bit float whose exponent is four balanced trits.</p><ol><li><a href="/learn/blocked-is-not-failed/">10 Blocked is not failed</a></li><li><a href="/learn/four-trit-exponent/">11 A float whose exponent is four trits</a></li><li><a href="/learn/negate-twice/">12 Negate twice, get the number back</a></li></ol></li><li><h3>Module 5 · Arithmetic on signed numbers</h3><p>Multiply two signed numbers, add them when their signs differ, and do both at once in a multiply-accumulate.</p><ol><li class="on" aria-current="page"><a href="/learn/a-sign-is-one-bit/">13 A sign is one bit</a></li><li><a href="/learn/opposite-signs-subtract/">14 Opposite signs subtract</a></li><li><a href="/learn/multiply-then-add/">15 Multiply, then add</a></li></ol></li><li><h3>Module 6 · Parts of a neuron</h3><p>A ReLU that bends at zero, a power of two for softmax, and an argmax that names the answer.</p><ol><li><a href="/learn/a-bend-at-zero/">16 A bend at zero</a></li><li><a href="/learn/two-to-the-x/">17 Two to the x</a></li><li><a href="/learn/pick-the-largest/">18 Pick the largest</a></li></ol></li><li><h3>Module 7 · Learning from a mistake</h3><p>A loss that prices a wrong guess in bits, one step that moves a weight against its gradient, and the hidden layer that XOR needs.</p><ol><li><a href="/learn/a-loss-in-bits/">19 A loss in bits</a></li><li><a href="/learn/one-step-downhill/">20 One step downhill</a></li><li><a href="/learn/xor-needs-a-bend/">21 XOR needs a bend</a></li></ol></li><li><h3>Module 8 · BitNet: ternary networks</h3><p>A threshold that squeezes a sum back to three values, one neuron that becomes a different function when its weights change, and a neuron that reads its inputs 27 trits at a time.</p><ol><li><a href="/learn/back-to-three-values/">22 Back to three values</a></li><li><a href="/learn/same-neuron-new-weights/">23 Same neuron, new weights</a></li><li><a href="/learn/a-neuron-in-chunks/">24 A neuron in chunks</a></li></ol></li><li><h3>Module 9 · The ternary MAC as a chip</h3><p>The 27-trit dot product as wires with no register, the same sum added into a register on every clock, and a small whole network to close the course.</p><ol><li><a href="/learn/a-dot-product-in-wires/">25 A dot product made of wires</a></li><li><a href="/learn/add-it-up-every-clock/">26 Add it up on every clock</a></li><li><a href="/learn/a-whole-network/">27 A whole network</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/a-small-program/index.html b/apps/website/public/learn/a-small-program/index.html
index 3988253c49..ea305f335b 100644
--- a/apps/website/public/learn/a-small-program/index.html
+++ b/apps/website/public/learn/a-small-program/index.html
@@ -89,5 +89,6 @@ <h1>A small complete program</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · A spec file</h3><p>What a .t27 file holds: a module line, prose lines and comments.</p><ol><li><a href="/learn/what-a-spec-is/">01 What a spec is</a></li><li><a href="/learn/the-module-line/">02 The module line</a></li><li><a href="/learn/comments-and-prose/">03 Comments and prose lines</a></li></ol></li><li><h3>Module 2 · Values and types</h3><p>Constants, the width of a whole number, and true, false and text.</p><ol><li><a href="/learn/constants/">04 Constants</a></li><li><a href="/learn/whole-numbers/">05 Whole numbers and their widths</a></li><li><a href="/learn/true-false-and-text/">06 True, false and text</a></li></ol></li><li><h3>Module 3 · Arrays, trits and expressions</h3><p>Lists of values, the three values of a trit, and what operators compute.</p><ol><li><a href="/learn/arrays/">07 Arrays</a></li><li><a href="/learn/trits/">08 Trits: minus one, zero, one</a></li><li><a href="/learn/expressions/">09 Expressions</a></li></ol></li><li><h3>Module 4 · Tests and functions</h3><p>A test block, several small tests in one spec, and a function with typed inputs.</p><ol><li><a href="/learn/a-test-block/">10 A test block</a></li><li><a href="/learn/many-tests/">11 Many tests in one spec</a></li><li><a href="/learn/functions/">12 Functions</a></li></ol></li><li><h3>Module 5 · Inside a function</h3><p>Names with let and var, choices with if and switch, and loops.</p><ol><li><a href="/learn/local-names/">13 Local names: let and var</a></li><li><a href="/learn/if-and-else/">14 Choices: if, else and switch</a></li><li><a href="/learn/loops/">15 Loops: while and for</a></li></ol></li><li><h3>Module 6 · Visibility and your own types</h3><p>What pub shares with other modules, and the structs and enums you declare.</p><ol><li><a href="/learn/pub-or-private/">16 pub or private</a></li><li><a href="/learn/structs/">17 Structs</a></li><li><a href="/learn/enums/">18 Enums</a></li></ol></li><li><h3>Module 7 · Modules, rules and gen-ts</h3><p>Reading another module, stating a rule that always holds, and a spec turned into TypeScript.</p><ol><li><a href="/learn/use-other-modules/">19 Using other modules</a></li><li><a href="/learn/invariants/">20 Invariants and benches</a></li><li><a href="/learn/gen-ts/">21 gen-ts: a spec becomes TypeScript</a></li></ol></li><li><h3>Module 8 · The compiler and tri</h3><p>The backends t27c writes, the stages it reads a spec in, and the tri command.</p><ol><li><a href="/learn/seven-backends/">22 Seven backends</a></li><li><a href="/learn/stages-of-t27c/">23 The stages of t27c</a></li><li><a href="/learn/tri-commands/">24 tri: the command line</a></li></ol></li><li><h3>Module 9 · Errors, a program, what next</h3><p>Reading a compiler error, one small complete program, and the course to take next.</p><ol><li><a href="/learn/reading-errors/">25 Reading compiler errors</a></li><li class="on" aria-current="page"><a href="/learn/a-small-program/">26 A small complete program</a></li><li><a href="/learn/where-next/">27 Where to go next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/a-ternary-alu/index.html b/apps/website/public/learn/a-ternary-alu/index.html
index 49bbac9358..7853b2ed50 100644
--- a/apps/website/public/learn/a-ternary-alu/index.html
+++ b/apps/website/public/learn/a-ternary-alu/index.html
@@ -535,5 +535,6 @@ <h1>A ternary ALU</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Three values</h3><p>Why three, how balanced ternary writes every number without a sign, and what flipping and cutting trits do.</p><ol><li><a href="/learn/why-three/">01 Why three</a></li><li><a href="/learn/balanced-ternary/">02 Balanced ternary</a></li><li><a href="/learn/flip-and-cut/">03 Flip to negate, cut to round</a></li></ol></li><li><h3>Module 2 · Logic with unknown</h3><p>Kleene's three-valued gates, two gates binary has no twin for, and addition as a pair of tables.</p><ol><li><a href="/learn/kleene-logic/">04 Three-valued logic</a></li><li><a href="/learn/consensus-and-any/">05 Consensus and any</a></li><li><a href="/learn/sum-and-carry/">06 Sum and carry</a></li></ol></li><li><h3>Module 3 · Adding trits</h3><p>The half adder, the full adder and a carry that ripples left, one place at a time.</p><ol><li><a href="/learn/half-adder/">07 The half adder</a></li><li><a href="/learn/full-adder/">08 The full adder</a></li><li><a href="/learn/ripple-carry/">09 Ripple carry</a></li></ol></li><li><h3>Module 4 · Multiplying without a multiplier</h3><p>Copy, drop or flip: a product by one trit, long multiplication, and a MAC that only adds.</p><ol><li><a href="/learn/times-a-trit/">10 Times one trit</a></li><li><a href="/learn/long-multiplication/">11 Long multiplication</a></li><li><a href="/learn/ternary-mac/">12 A ternary MAC</a></li></ol></li><li><h3>Module 5 · Trits in binary memory</h3><p>Two bits per trit, five trits per byte, and the bits a 27-trit word needs.</p><ol><li><a href="/learn/two-bits-per-trit/">13 Two bits per trit</a></li><li><a href="/learn/five-trits-a-byte/">14 Five trits in a byte</a></li><li><a href="/learn/a-word-of-27/">15 A word of 27 trits</a></li></ol></li><li><h3>Module 6 · The TRI-27 instruction word</h3><p>The 32-bit word the Trinity emulator decodes: its fields, its 47 opcodes and its 15-bit immediate.</p><ol><li><a href="/learn/the-instruction-word/">16 The instruction word</a></li><li><a href="/learn/the-opcode-map/">17 The opcode map</a></li><li><a href="/learn/the-immediate/">18 The immediate</a></li></ol></li><li><h3>Module 7 · A ternary machine</h3><p>An ALU built from this course's adders, a three-way jump, and a program you can step.</p><ol><li class="on" aria-current="page"><a href="/learn/a-ternary-alu/">19 A ternary ALU</a></li><li><a href="/learn/three-way-jump/">20 A three-way jump</a></li><li><a href="/learn/run-a-program/">21 Run a program</a></li></ol></li><li><h3>Module 8 · Three answers</h3><p>Compare at the highest differing trit, find a number in thirds, sort with three-way compares.</p><ol><li><a href="/learn/compare-at-the-top/">22 Compare at the top</a></li><li><a href="/learn/questions-in-thirds/">23 Questions in thirds</a></li><li><a href="/learn/sort-by-three/">24 Sort with three-way compares</a></li></ol></li><li><h3>Module 9 · Ternary neurons</h3><p>Weights of -1, 0, +1: one neuron, a detector and a small layer, with no multiplier anywhere.</p><ol><li><a href="/learn/one-ternary-neuron/">25 One ternary neuron</a></li><li><a href="/learn/a-bar-detector/">26 A bar detector</a></li><li><a href="/learn/a-tiny-layer/">27 A tiny layer</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/a-test-block/index.html b/apps/website/public/learn/a-test-block/index.html
index 018ce09cbf..050ac75495 100644
--- a/apps/website/public/learn/a-test-block/index.html
+++ b/apps/website/public/learn/a-test-block/index.html
@@ -82,5 +82,6 @@ <h1>A test block</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · A spec file</h3><p>What a .t27 file holds: a module line, prose lines and comments.</p><ol><li><a href="/learn/what-a-spec-is/">01 What a spec is</a></li><li><a href="/learn/the-module-line/">02 The module line</a></li><li><a href="/learn/comments-and-prose/">03 Comments and prose lines</a></li></ol></li><li><h3>Module 2 · Values and types</h3><p>Constants, the width of a whole number, and true, false and text.</p><ol><li><a href="/learn/constants/">04 Constants</a></li><li><a href="/learn/whole-numbers/">05 Whole numbers and their widths</a></li><li><a href="/learn/true-false-and-text/">06 True, false and text</a></li></ol></li><li><h3>Module 3 · Arrays, trits and expressions</h3><p>Lists of values, the three values of a trit, and what operators compute.</p><ol><li><a href="/learn/arrays/">07 Arrays</a></li><li><a href="/learn/trits/">08 Trits: minus one, zero, one</a></li><li><a href="/learn/expressions/">09 Expressions</a></li></ol></li><li><h3>Module 4 · Tests and functions</h3><p>A test block, several small tests in one spec, and a function with typed inputs.</p><ol><li class="on" aria-current="page"><a href="/learn/a-test-block/">10 A test block</a></li><li><a href="/learn/many-tests/">11 Many tests in one spec</a></li><li><a href="/learn/functions/">12 Functions</a></li></ol></li><li><h3>Module 5 · Inside a function</h3><p>Names with let and var, choices with if and switch, and loops.</p><ol><li><a href="/learn/local-names/">13 Local names: let and var</a></li><li><a href="/learn/if-and-else/">14 Choices: if, else and switch</a></li><li><a href="/learn/loops/">15 Loops: while and for</a></li></ol></li><li><h3>Module 6 · Visibility and your own types</h3><p>What pub shares with other modules, and the structs and enums you declare.</p><ol><li><a href="/learn/pub-or-private/">16 pub or private</a></li><li><a href="/learn/structs/">17 Structs</a></li><li><a href="/learn/enums/">18 Enums</a></li></ol></li><li><h3>Module 7 · Modules, rules and gen-ts</h3><p>Reading another module, stating a rule that always holds, and a spec turned into TypeScript.</p><ol><li><a href="/learn/use-other-modules/">19 Using other modules</a></li><li><a href="/learn/invariants/">20 Invariants and benches</a></li><li><a href="/learn/gen-ts/">21 gen-ts: a spec becomes TypeScript</a></li></ol></li><li><h3>Module 8 · The compiler and tri</h3><p>The backends t27c writes, the stages it reads a spec in, and the tri command.</p><ol><li><a href="/learn/seven-backends/">22 Seven backends</a></li><li><a href="/learn/stages-of-t27c/">23 The stages of t27c</a></li><li><a href="/learn/tri-commands/">24 tri: the command line</a></li></ol></li><li><h3>Module 9 · Errors, a program, what next</h3><p>Reading a compiler error, one small complete program, and the course to take next.</p><ol><li><a href="/learn/reading-errors/">25 Reading compiler errors</a></li><li><a href="/learn/a-small-program/">26 A small complete program</a></li><li><a href="/learn/where-next/">27 Where to go next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/a-tiny-layer/index.html b/apps/website/public/learn/a-tiny-layer/index.html
index 7a183a4198..acdf023068 100644
--- a/apps/website/public/learn/a-tiny-layer/index.html
+++ b/apps/website/public/learn/a-tiny-layer/index.html
@@ -132,5 +132,6 @@ <h1>A tiny layer</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Three values</h3><p>Why three, how balanced ternary writes every number without a sign, and what flipping and cutting trits do.</p><ol><li><a href="/learn/why-three/">01 Why three</a></li><li><a href="/learn/balanced-ternary/">02 Balanced ternary</a></li><li><a href="/learn/flip-and-cut/">03 Flip to negate, cut to round</a></li></ol></li><li><h3>Module 2 · Logic with unknown</h3><p>Kleene's three-valued gates, two gates binary has no twin for, and addition as a pair of tables.</p><ol><li><a href="/learn/kleene-logic/">04 Three-valued logic</a></li><li><a href="/learn/consensus-and-any/">05 Consensus and any</a></li><li><a href="/learn/sum-and-carry/">06 Sum and carry</a></li></ol></li><li><h3>Module 3 · Adding trits</h3><p>The half adder, the full adder and a carry that ripples left, one place at a time.</p><ol><li><a href="/learn/half-adder/">07 The half adder</a></li><li><a href="/learn/full-adder/">08 The full adder</a></li><li><a href="/learn/ripple-carry/">09 Ripple carry</a></li></ol></li><li><h3>Module 4 · Multiplying without a multiplier</h3><p>Copy, drop or flip: a product by one trit, long multiplication, and a MAC that only adds.</p><ol><li><a href="/learn/times-a-trit/">10 Times one trit</a></li><li><a href="/learn/long-multiplication/">11 Long multiplication</a></li><li><a href="/learn/ternary-mac/">12 A ternary MAC</a></li></ol></li><li><h3>Module 5 · Trits in binary memory</h3><p>Two bits per trit, five trits per byte, and the bits a 27-trit word needs.</p><ol><li><a href="/learn/two-bits-per-trit/">13 Two bits per trit</a></li><li><a href="/learn/five-trits-a-byte/">14 Five trits in a byte</a></li><li><a href="/learn/a-word-of-27/">15 A word of 27 trits</a></li></ol></li><li><h3>Module 6 · The TRI-27 instruction word</h3><p>The 32-bit word the Trinity emulator decodes: its fields, its 47 opcodes and its 15-bit immediate.</p><ol><li><a href="/learn/the-instruction-word/">16 The instruction word</a></li><li><a href="/learn/the-opcode-map/">17 The opcode map</a></li><li><a href="/learn/the-immediate/">18 The immediate</a></li></ol></li><li><h3>Module 7 · A ternary machine</h3><p>An ALU built from this course's adders, a three-way jump, and a program you can step.</p><ol><li><a href="/learn/a-ternary-alu/">19 A ternary ALU</a></li><li><a href="/learn/three-way-jump/">20 A three-way jump</a></li><li><a href="/learn/run-a-program/">21 Run a program</a></li></ol></li><li><h3>Module 8 · Three answers</h3><p>Compare at the highest differing trit, find a number in thirds, sort with three-way compares.</p><ol><li><a href="/learn/compare-at-the-top/">22 Compare at the top</a></li><li><a href="/learn/questions-in-thirds/">23 Questions in thirds</a></li><li><a href="/learn/sort-by-three/">24 Sort with three-way compares</a></li></ol></li><li><h3>Module 9 · Ternary neurons</h3><p>Weights of -1, 0, +1: one neuron, a detector and a small layer, with no multiplier anywhere.</p><ol><li><a href="/learn/one-ternary-neuron/">25 One ternary neuron</a></li><li><a href="/learn/a-bar-detector/">26 A bar detector</a></li><li class="on" aria-current="page"><a href="/learn/a-tiny-layer/">27 A tiny layer</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/a-whole-network/index.html b/apps/website/public/learn/a-whole-network/index.html
index b21b8c2bf6..37ea6e419f 100644
--- a/apps/website/public/learn/a-whole-network/index.html
+++ b/apps/website/public/learn/a-whole-network/index.html
@@ -142,5 +142,6 @@ <h1>A whole network</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Lab: our own research</h3><p>A number format of our own, an honest scoreboard, and a model's tables multiplied on the board.</p><ol><li><a href="/learn/t27-number-format/">01 T27: a number format of our own</a></li><li><a href="/learn/an-honest-scoreboard/">02 An honest scoreboard</a></li><li><a href="/learn/a-model-on-the-board/">03 A model's tables on the board</a></li></ol></li><li><h3>Module 2 · AI numbers: the MX block</h3><p>How AI chips keep weights in a few bits: one shared scale per block, the scale byte itself, and what one outlier does to its neighbours.</p><ol><li><a href="/learn/one-scale-per-block/">04 One scale for a block of weights</a></li><li><a href="/learn/scale-byte-native/">05 The scale byte on a real machine</a></li><li><a href="/learn/one-outlier/">06 One outlier, many zeros</a></li></ol></li><li><h3>Module 3 · Ternary weights</h3><p>Weights that are only minus, zero or plus a scale, the five rules a ternary alphabet must pass, and a test pass that checked nothing.</p><ol><li><a href="/learn/phi-scaled-ternary/">07 Weights of minus phi, zero and plus phi</a></li><li><a href="/learn/a-pass-that-checked-nothing/">08 A pass that checked nothing</a></li><li><a href="/learn/five-rules-for-weights/">09 Five rules for a weight alphabet</a></li></ol></li><li><h3>Module 4 · The Ternary Network Float</h3><p>A rule the compiler enforces before any test runs, and a 17-bit float whose exponent is four balanced trits.</p><ol><li><a href="/learn/blocked-is-not-failed/">10 Blocked is not failed</a></li><li><a href="/learn/four-trit-exponent/">11 A float whose exponent is four trits</a></li><li><a href="/learn/negate-twice/">12 Negate twice, get the number back</a></li></ol></li><li><h3>Module 5 · Arithmetic on signed numbers</h3><p>Multiply two signed numbers, add them when their signs differ, and do both at once in a multiply-accumulate.</p><ol><li><a href="/learn/a-sign-is-one-bit/">13 A sign is one bit</a></li><li><a href="/learn/opposite-signs-subtract/">14 Opposite signs subtract</a></li><li><a href="/learn/multiply-then-add/">15 Multiply, then add</a></li></ol></li><li><h3>Module 6 · Parts of a neuron</h3><p>A ReLU that bends at zero, a power of two for softmax, and an argmax that names the answer.</p><ol><li><a href="/learn/a-bend-at-zero/">16 A bend at zero</a></li><li><a href="/learn/two-to-the-x/">17 Two to the x</a></li><li><a href="/learn/pick-the-largest/">18 Pick the largest</a></li></ol></li><li><h3>Module 7 · Learning from a mistake</h3><p>A loss that prices a wrong guess in bits, one step that moves a weight against its gradient, and the hidden layer that XOR needs.</p><ol><li><a href="/learn/a-loss-in-bits/">19 A loss in bits</a></li><li><a href="/learn/one-step-downhill/">20 One step downhill</a></li><li><a href="/learn/xor-needs-a-bend/">21 XOR needs a bend</a></li></ol></li><li><h3>Module 8 · BitNet: ternary networks</h3><p>A threshold that squeezes a sum back to three values, one neuron that becomes a different function when its weights change, and a neuron that reads its inputs 27 trits at a time.</p><ol><li><a href="/learn/back-to-three-values/">22 Back to three values</a></li><li><a href="/learn/same-neuron-new-weights/">23 Same neuron, new weights</a></li><li><a href="/learn/a-neuron-in-chunks/">24 A neuron in chunks</a></li></ol></li><li><h3>Module 9 · The ternary MAC as a chip</h3><p>The 27-trit dot product as wires with no register, the same sum added into a register on every clock, and a small whole network to close the course.</p><ol><li><a href="/learn/a-dot-product-in-wires/">25 A dot product made of wires</a></li><li><a href="/learn/add-it-up-every-clock/">26 Add it up on every clock</a></li><li class="on" aria-current="page"><a href="/learn/a-whole-network/">27 A whole network</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/a-word-of-27/index.html b/apps/website/public/learn/a-word-of-27/index.html
index b35e1d62ab..736d891675 100644
--- a/apps/website/public/learn/a-word-of-27/index.html
+++ b/apps/website/public/learn/a-word-of-27/index.html
@@ -769,5 +769,6 @@ <h1>A word of 27 trits</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Three values</h3><p>Why three, how balanced ternary writes every number without a sign, and what flipping and cutting trits do.</p><ol><li><a href="/learn/why-three/">01 Why three</a></li><li><a href="/learn/balanced-ternary/">02 Balanced ternary</a></li><li><a href="/learn/flip-and-cut/">03 Flip to negate, cut to round</a></li></ol></li><li><h3>Module 2 · Logic with unknown</h3><p>Kleene's three-valued gates, two gates binary has no twin for, and addition as a pair of tables.</p><ol><li><a href="/learn/kleene-logic/">04 Three-valued logic</a></li><li><a href="/learn/consensus-and-any/">05 Consensus and any</a></li><li><a href="/learn/sum-and-carry/">06 Sum and carry</a></li></ol></li><li><h3>Module 3 · Adding trits</h3><p>The half adder, the full adder and a carry that ripples left, one place at a time.</p><ol><li><a href="/learn/half-adder/">07 The half adder</a></li><li><a href="/learn/full-adder/">08 The full adder</a></li><li><a href="/learn/ripple-carry/">09 Ripple carry</a></li></ol></li><li><h3>Module 4 · Multiplying without a multiplier</h3><p>Copy, drop or flip: a product by one trit, long multiplication, and a MAC that only adds.</p><ol><li><a href="/learn/times-a-trit/">10 Times one trit</a></li><li><a href="/learn/long-multiplication/">11 Long multiplication</a></li><li><a href="/learn/ternary-mac/">12 A ternary MAC</a></li></ol></li><li><h3>Module 5 · Trits in binary memory</h3><p>Two bits per trit, five trits per byte, and the bits a 27-trit word needs.</p><ol><li><a href="/learn/two-bits-per-trit/">13 Two bits per trit</a></li><li><a href="/learn/five-trits-a-byte/">14 Five trits in a byte</a></li><li class="on" aria-current="page"><a href="/learn/a-word-of-27/">15 A word of 27 trits</a></li></ol></li><li><h3>Module 6 · The TRI-27 instruction word</h3><p>The 32-bit word the Trinity emulator decodes: its fields, its 47 opcodes and its 15-bit immediate.</p><ol><li><a href="/learn/the-instruction-word/">16 The instruction word</a></li><li><a href="/learn/the-opcode-map/">17 The opcode map</a></li><li><a href="/learn/the-immediate/">18 The immediate</a></li></ol></li><li><h3>Module 7 · A ternary machine</h3><p>An ALU built from this course's adders, a three-way jump, and a program you can step.</p><ol><li><a href="/learn/a-ternary-alu/">19 A ternary ALU</a></li><li><a href="/learn/three-way-jump/">20 A three-way jump</a></li><li><a href="/learn/run-a-program/">21 Run a program</a></li></ol></li><li><h3>Module 8 · Three answers</h3><p>Compare at the highest differing trit, find a number in thirds, sort with three-way compares.</p><ol><li><a href="/learn/compare-at-the-top/">22 Compare at the top</a></li><li><a href="/learn/questions-in-thirds/">23 Questions in thirds</a></li><li><a href="/learn/sort-by-three/">24 Sort with three-way compares</a></li></ol></li><li><h3>Module 9 · Ternary neurons</h3><p>Weights of -1, 0, +1: one neuron, a detector and a small layer, with no multiplier anywhere.</p><ol><li><a href="/learn/one-ternary-neuron/">25 One ternary neuron</a></li><li><a href="/learn/a-bar-detector/">26 A bar detector</a></li><li><a href="/learn/a-tiny-layer/">27 A tiny layer</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/add-it-up-every-clock/index.html b/apps/website/public/learn/add-it-up-every-clock/index.html
index 30aa8bedc7..085552be4d 100644
--- a/apps/website/public/learn/add-it-up-every-clock/index.html
+++ b/apps/website/public/learn/add-it-up-every-clock/index.html
@@ -122,5 +122,6 @@ <h1>Add it up on every clock</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Lab: our own research</h3><p>A number format of our own, an honest scoreboard, and a model's tables multiplied on the board.</p><ol><li><a href="/learn/t27-number-format/">01 T27: a number format of our own</a></li><li><a href="/learn/an-honest-scoreboard/">02 An honest scoreboard</a></li><li><a href="/learn/a-model-on-the-board/">03 A model's tables on the board</a></li></ol></li><li><h3>Module 2 · AI numbers: the MX block</h3><p>How AI chips keep weights in a few bits: one shared scale per block, the scale byte itself, and what one outlier does to its neighbours.</p><ol><li><a href="/learn/one-scale-per-block/">04 One scale for a block of weights</a></li><li><a href="/learn/scale-byte-native/">05 The scale byte on a real machine</a></li><li><a href="/learn/one-outlier/">06 One outlier, many zeros</a></li></ol></li><li><h3>Module 3 · Ternary weights</h3><p>Weights that are only minus, zero or plus a scale, the five rules a ternary alphabet must pass, and a test pass that checked nothing.</p><ol><li><a href="/learn/phi-scaled-ternary/">07 Weights of minus phi, zero and plus phi</a></li><li><a href="/learn/a-pass-that-checked-nothing/">08 A pass that checked nothing</a></li><li><a href="/learn/five-rules-for-weights/">09 Five rules for a weight alphabet</a></li></ol></li><li><h3>Module 4 · The Ternary Network Float</h3><p>A rule the compiler enforces before any test runs, and a 17-bit float whose exponent is four balanced trits.</p><ol><li><a href="/learn/blocked-is-not-failed/">10 Blocked is not failed</a></li><li><a href="/learn/four-trit-exponent/">11 A float whose exponent is four trits</a></li><li><a href="/learn/negate-twice/">12 Negate twice, get the number back</a></li></ol></li><li><h3>Module 5 · Arithmetic on signed numbers</h3><p>Multiply two signed numbers, add them when their signs differ, and do both at once in a multiply-accumulate.</p><ol><li><a href="/learn/a-sign-is-one-bit/">13 A sign is one bit</a></li><li><a href="/learn/opposite-signs-subtract/">14 Opposite signs subtract</a></li><li><a href="/learn/multiply-then-add/">15 Multiply, then add</a></li></ol></li><li><h3>Module 6 · Parts of a neuron</h3><p>A ReLU that bends at zero, a power of two for softmax, and an argmax that names the answer.</p><ol><li><a href="/learn/a-bend-at-zero/">16 A bend at zero</a></li><li><a href="/learn/two-to-the-x/">17 Two to the x</a></li><li><a href="/learn/pick-the-largest/">18 Pick the largest</a></li></ol></li><li><h3>Module 7 · Learning from a mistake</h3><p>A loss that prices a wrong guess in bits, one step that moves a weight against its gradient, and the hidden layer that XOR needs.</p><ol><li><a href="/learn/a-loss-in-bits/">19 A loss in bits</a></li><li><a href="/learn/one-step-downhill/">20 One step downhill</a></li><li><a href="/learn/xor-needs-a-bend/">21 XOR needs a bend</a></li></ol></li><li><h3>Module 8 · BitNet: ternary networks</h3><p>A threshold that squeezes a sum back to three values, one neuron that becomes a different function when its weights change, and a neuron that reads its inputs 27 trits at a time.</p><ol><li><a href="/learn/back-to-three-values/">22 Back to three values</a></li><li><a href="/learn/same-neuron-new-weights/">23 Same neuron, new weights</a></li><li><a href="/learn/a-neuron-in-chunks/">24 A neuron in chunks</a></li></ol></li><li><h3>Module 9 · The ternary MAC as a chip</h3><p>The 27-trit dot product as wires with no register, the same sum added into a register on every clock, and a small whole network to close the course.</p><ol><li><a href="/learn/a-dot-product-in-wires/">25 A dot product made of wires</a></li><li class="on" aria-current="page"><a href="/learn/add-it-up-every-clock/">26 Add it up on every clock</a></li><li><a href="/learn/a-whole-network/">27 A whole network</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/address-decode/index.html b/apps/website/public/learn/address-decode/index.html
index 540bdb3f48..9e1ca990a6 100644
--- a/apps/website/public/learn/address-decode/index.html
+++ b/apps/website/public/learn/address-decode/index.html
@@ -228,5 +228,6 @@ <h1>Address decode</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a bus is</h3><p>Why a bus exists at all: a conversation on wires, framed and addressed, and who is allowed to talk.</p><ol><li><a href="/learn/wires-that-talk/">01 Wires that talk</a></li><li><a href="/learn/framing-a-conversation/">02 Framing a conversation</a></li><li><a href="/learn/master-and-servant/">03 Master and servant</a></li></ol></li><li><h3>Module 2 · UART</h3><p>The two-wire bus with no clock: the frame, the divisor that sets the rate, and the status a driver polls.</p><ol><li><a href="/learn/the-frame/">04 The frame: start, data, stop</a></li><li><a href="/learn/baud-and-the-divisor/">05 Baud and the divisor</a></li><li><a href="/learn/status-and-fifos/">06 Status and FIFOs</a></li></ol></li><li><h3>Module 3 · SPI</h3><p>The clocked conversation: four modes, a prescaler ladder, and a chip select per servant.</p><ol><li><a href="/learn/modes-cpol-cpha/">07 Modes: CPOL and CPHA</a></li><li><a href="/learn/the-prescaler-ladder/">08 The prescaler ladder</a></li><li><a href="/learn/chip-select-and-widths/">09 Chip select and widths</a></li></ol></li><li><h3>Module 4 · APB</h3><p>The register bus: PSEL and PENABLE, strobes and wait states, and how many address bits a peripheral count costs.</p><ol><li><a href="/learn/psel-and-penable/">10 PSEL and PENABLE</a></li><li><a href="/learn/strobes-and-waits/">11 Strobes and waits</a></li><li class="on" aria-current="page"><a href="/learn/address-decode/">12 Address decode</a></li></ol></li><li><h3>Module 5 · AXI4</h3><p>The five channels: address, data and response in both directions, lite or full, bursts and IDs.</p><ol><li><a href="/learn/five-channels/">13 Five channels</a></li><li><a href="/learn/lite-or-full/">14 Lite or full</a></li><li><a href="/learn/bursts-and-ids/">15 Bursts and IDs</a></li></ol></li><li><h3>Module 6 · Memory</h3><p>What sits on the far side of every bus: memory maps, port kinds, and latency that a wait state must cover.</p><ol><li><a href="/learn/memory-maps/">16 Memory maps</a></li><li><a href="/learn/ports-and-kinds/">17 Ports and kinds</a></li><li><a href="/learn/latency/">18 Latency</a></li></ol></li><li><h3>Module 7 · Bridges</h3><p>Why designs grow more than one bus, and the packet bridge that moves work between them.</p><ol><li><a href="/learn/why-bridges/">19 Why bridges</a></li><li><a href="/learn/the-packet-bridge/">20 The packet bridge</a></li><li><a href="/learn/everything-in-one-bench/">21 Everything in one bench</a></li></ol></li><li><h3>Module 8 · Ethernet</h3><p>Frames, the frame check sequence, RGMII timing, and the pre-registered steps of a real bring-up.</p><ol><li><a href="/learn/frames-and-the-fcs/">22 Frames and the FCS</a></li><li><a href="/learn/rgmii-timing/">23 RGMII timing</a></li><li><a href="/learn/bring-up-on-the-bench/">24 Bring-up on the bench</a></li></ol></li><li><h3>Module 9 · The bench</h3><p>The discipline that guards real hardware: who holds the IO, taking and giving the lock, and what runs next.</p><ol><li><a href="/learn/who-holds-the-io/">25 Who holds the IO</a></li><li><a href="/learn/claim-and-release/">26 Claim and release</a></li><li><a href="/learn/capstone-what-runs-next/">27 Capstone: what runs next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/ai-numbers/index.html b/apps/website/public/learn/ai-numbers/index.html
index edbfa7563a..76f0039c7d 100644
--- a/apps/website/public/learn/ai-numbers/index.html
+++ b/apps/website/public/learn/ai-numbers/index.html
@@ -67,5 +67,6 @@ <h1>AI numbers with t27: from number formats to a ternary network</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Lab: our own research</h3><p>A number format of our own, an honest scoreboard, and a model's tables multiplied on the board.</p><ol><li><a href="/learn/t27-number-format/">01 T27: a number format of our own</a></li><li><a href="/learn/an-honest-scoreboard/">02 An honest scoreboard</a></li><li><a href="/learn/a-model-on-the-board/">03 A model's tables on the board</a></li></ol></li><li><h3>Module 2 · AI numbers: the MX block</h3><p>How AI chips keep weights in a few bits: one shared scale per block, the scale byte itself, and what one outlier does to its neighbours.</p><ol><li><a href="/learn/one-scale-per-block/">04 One scale for a block of weights</a></li><li><a href="/learn/scale-byte-native/">05 The scale byte on a real machine</a></li><li><a href="/learn/one-outlier/">06 One outlier, many zeros</a></li></ol></li><li><h3>Module 3 · Ternary weights</h3><p>Weights that are only minus, zero or plus a scale, the five rules a ternary alphabet must pass, and a test pass that checked nothing.</p><ol><li><a href="/learn/phi-scaled-ternary/">07 Weights of minus phi, zero and plus phi</a></li><li><a href="/learn/a-pass-that-checked-nothing/">08 A pass that checked nothing</a></li><li><a href="/learn/five-rules-for-weights/">09 Five rules for a weight alphabet</a></li></ol></li><li><h3>Module 4 · The Ternary Network Float</h3><p>A rule the compiler enforces before any test runs, and a 17-bit float whose exponent is four balanced trits.</p><ol><li><a href="/learn/blocked-is-not-failed/">10 Blocked is not failed</a></li><li><a href="/learn/four-trit-exponent/">11 A float whose exponent is four trits</a></li><li><a href="/learn/negate-twice/">12 Negate twice, get the number back</a></li></ol></li><li><h3>Module 5 · Arithmetic on signed numbers</h3><p>Multiply two signed numbers, add them when their signs differ, and do both at once in a multiply-accumulate.</p><ol><li><a href="/learn/a-sign-is-one-bit/">13 A sign is one bit</a></li><li><a href="/learn/opposite-signs-subtract/">14 Opposite signs subtract</a></li><li><a href="/learn/multiply-then-add/">15 Multiply, then add</a></li></ol></li><li><h3>Module 6 · Parts of a neuron</h3><p>A ReLU that bends at zero, a power of two for softmax, and an argmax that names the answer.</p><ol><li><a href="/learn/a-bend-at-zero/">16 A bend at zero</a></li><li><a href="/learn/two-to-the-x/">17 Two to the x</a></li><li><a href="/learn/pick-the-largest/">18 Pick the largest</a></li></ol></li><li><h3>Module 7 · Learning from a mistake</h3><p>A loss that prices a wrong guess in bits, one step that moves a weight against its gradient, and the hidden layer that XOR needs.</p><ol><li><a href="/learn/a-loss-in-bits/">19 A loss in bits</a></li><li><a href="/learn/one-step-downhill/">20 One step downhill</a></li><li><a href="/learn/xor-needs-a-bend/">21 XOR needs a bend</a></li></ol></li><li><h3>Module 8 · BitNet: ternary networks</h3><p>A threshold that squeezes a sum back to three values, one neuron that becomes a different function when its weights change, and a neuron that reads its inputs 27 trits at a time.</p><ol><li><a href="/learn/back-to-three-values/">22 Back to three values</a></li><li><a href="/learn/same-neuron-new-weights/">23 Same neuron, new weights</a></li><li><a href="/learn/a-neuron-in-chunks/">24 A neuron in chunks</a></li></ol></li><li><h3>Module 9 · The ternary MAC as a chip</h3><p>The 27-trit dot product as wires with no register, the same sum added into a register on every clock, and a small whole network to close the course.</p><ol><li><a href="/learn/a-dot-product-in-wires/">25 A dot product made of wires</a></li><li><a href="/learn/add-it-up-every-clock/">26 Add it up on every clock</a></li><li><a href="/learn/a-whole-network/">27 A whole network</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/an-honest-scoreboard/index.html b/apps/website/public/learn/an-honest-scoreboard/index.html
index 4a7443f569..f34ebabfd5 100644
--- a/apps/website/public/learn/an-honest-scoreboard/index.html
+++ b/apps/website/public/learn/an-honest-scoreboard/index.html
@@ -210,5 +210,6 @@ <h1>An honest scoreboard</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Lab: our own research</h3><p>A number format of our own, an honest scoreboard, and a model's tables multiplied on the board.</p><ol><li><a href="/learn/t27-number-format/">01 T27: a number format of our own</a></li><li class="on" aria-current="page"><a href="/learn/an-honest-scoreboard/">02 An honest scoreboard</a></li><li><a href="/learn/a-model-on-the-board/">03 A model's tables on the board</a></li></ol></li><li><h3>Module 2 · AI numbers: the MX block</h3><p>How AI chips keep weights in a few bits: one shared scale per block, the scale byte itself, and what one outlier does to its neighbours.</p><ol><li><a href="/learn/one-scale-per-block/">04 One scale for a block of weights</a></li><li><a href="/learn/scale-byte-native/">05 The scale byte on a real machine</a></li><li><a href="/learn/one-outlier/">06 One outlier, many zeros</a></li></ol></li><li><h3>Module 3 · Ternary weights</h3><p>Weights that are only minus, zero or plus a scale, the five rules a ternary alphabet must pass, and a test pass that checked nothing.</p><ol><li><a href="/learn/phi-scaled-ternary/">07 Weights of minus phi, zero and plus phi</a></li><li><a href="/learn/a-pass-that-checked-nothing/">08 A pass that checked nothing</a></li><li><a href="/learn/five-rules-for-weights/">09 Five rules for a weight alphabet</a></li></ol></li><li><h3>Module 4 · The Ternary Network Float</h3><p>A rule the compiler enforces before any test runs, and a 17-bit float whose exponent is four balanced trits.</p><ol><li><a href="/learn/blocked-is-not-failed/">10 Blocked is not failed</a></li><li><a href="/learn/four-trit-exponent/">11 A float whose exponent is four trits</a></li><li><a href="/learn/negate-twice/">12 Negate twice, get the number back</a></li></ol></li><li><h3>Module 5 · Arithmetic on signed numbers</h3><p>Multiply two signed numbers, add them when their signs differ, and do both at once in a multiply-accumulate.</p><ol><li><a href="/learn/a-sign-is-one-bit/">13 A sign is one bit</a></li><li><a href="/learn/opposite-signs-subtract/">14 Opposite signs subtract</a></li><li><a href="/learn/multiply-then-add/">15 Multiply, then add</a></li></ol></li><li><h3>Module 6 · Parts of a neuron</h3><p>A ReLU that bends at zero, a power of two for softmax, and an argmax that names the answer.</p><ol><li><a href="/learn/a-bend-at-zero/">16 A bend at zero</a></li><li><a href="/learn/two-to-the-x/">17 Two to the x</a></li><li><a href="/learn/pick-the-largest/">18 Pick the largest</a></li></ol></li><li><h3>Module 7 · Learning from a mistake</h3><p>A loss that prices a wrong guess in bits, one step that moves a weight against its gradient, and the hidden layer that XOR needs.</p><ol><li><a href="/learn/a-loss-in-bits/">19 A loss in bits</a></li><li><a href="/learn/one-step-downhill/">20 One step downhill</a></li><li><a href="/learn/xor-needs-a-bend/">21 XOR needs a bend</a></li></ol></li><li><h3>Module 8 · BitNet: ternary networks</h3><p>A threshold that squeezes a sum back to three values, one neuron that becomes a different function when its weights change, and a neuron that reads its inputs 27 trits at a time.</p><ol><li><a href="/learn/back-to-three-values/">22 Back to three values</a></li><li><a href="/learn/same-neuron-new-weights/">23 Same neuron, new weights</a></li><li><a href="/learn/a-neuron-in-chunks/">24 A neuron in chunks</a></li></ol></li><li><h3>Module 9 · The ternary MAC as a chip</h3><p>The 27-trit dot product as wires with no register, the same sum added into a register on every clock, and a small whole network to close the course.</p><ol><li><a href="/learn/a-dot-product-in-wires/">25 A dot product made of wires</a></li><li><a href="/learn/add-it-up-every-clock/">26 Add it up on every clock</a></li><li><a href="/learn/a-whole-network/">27 A whole network</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/anatomy-of-a-tb/index.html b/apps/website/public/learn/anatomy-of-a-tb/index.html
index 8cae13e5c0..b00e33c38c 100644
--- a/apps/website/public/learn/anatomy-of-a-tb/index.html
+++ b/apps/website/public/learn/anatomy-of-a-tb/index.html
@@ -304,5 +304,6 @@ <h1>Anatomy of a testbench</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Why verify</h3><p>Designs that compile and are wrong, the model that decides, and the plan written before the code.</p><ol><li><a href="/learn/bugs-that-compile/">01 Bugs that compile</a></li><li><a href="/learn/golden-model/">02 The golden model</a></li><li><a href="/learn/plan-before-code/">03 Plan before code</a></li></ol></li><li><h3>Module 2 · Testbenches</h3><p>Stimulus, checks and a verdict, written as one spec beside the design it judges.</p><ol><li class="on" aria-current="page"><a href="/learn/anatomy-of-a-tb/">04 Anatomy of a testbench</a></li><li><a href="/learn/self-checking/">05 Self-checking, not staring</a></li><li><a href="/learn/directed-vs-random/">06 Directed versus random</a></li></ol></li><li><h3>Module 3 · Waveforms</h3><p>A trace of every signal, read the way a hardware engineer reads it, and two runs compared.</p><ol><li><a href="/learn/vcd-format/">07 The VCD format</a></li><li><a href="/learn/debug-with-waves/">08 Debug with waves</a></li><li><a href="/learn/compare-traces/">09 Compare two traces</a></li></ol></li><li><h3>Module 4 · Conformance vectors</h3><p>Cases with the answer written beside them, kept where the compiler can reach them.</p><ol><li><a href="/learn/what-a-vector-is/">10 What a vector is</a></li><li><a href="/learn/golden-ruler-vectors/">11 Golden ruler vectors</a></li><li><a href="/learn/end-to-end-scenarios/">12 End-to-end scenarios</a></li></ol></li><li><h3>Module 5 · Cosimulation</h3><p>Spec, simulator and board agreeing on the bench Artix-7 XC7A200T, and what to do when they do not.</p><ol><li><a href="/learn/three-way-agreement/">13 Three-way agreement</a></li><li><a href="/learn/simulator-internals/">14 Inside the simulator</a></li><li><a href="/learn/when-they-disagree/">15 When they disagree</a></li></ol></li><li><h3>Module 6 · Coverage</h3><p>What the tests touched: lines, toggles, states, and what that number hides.</p><ol><li><a href="/learn/line-and-toggle/">16 Line and toggle</a></li><li><a href="/learn/fsm-coverage/">17 FSM coverage</a></li><li><a href="/learn/coverage-lies/">18 Coverage lies</a></li></ol></li><li><h3>Module 7 · Formal</h3><p>Assertions that hold every cycle, bounded search for a counterexample, and why a proof needs induction.</p><ol><li><a href="/learn/assertions/">19 Assertions</a></li><li><a href="/learn/bounded-model-checking/">20 Bounded model checking</a></li><li><a href="/learn/induction/">21 Induction</a></li></ol></li><li><h3>Module 8 · Mutation</h3><p>Break the design on purpose and count what the tests catch.</p><ol><li><a href="/learn/kill-the-mutant/">22 Kill the mutant</a></li><li><a href="/learn/surviving-mutants/">23 Surviving mutants</a></li><li><a href="/learn/test-strength/">24 Test strength</a></li></ol></li><li><h3>Module 9 · Sign-off</h3><p>One command, every receipt, a clean verdict you can show.</p><ol><li><a href="/learn/build-verify/">25 Build and verify</a></li><li><a href="/learn/board-selftest/">26 The board self-tests</a></li><li><a href="/learn/verify-capstone/">27 Capstone: verify a module whole</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/angle-and-magnitude/index.html b/apps/website/public/learn/angle-and-magnitude/index.html
index e1cb0b2a25..d88a2abb55 100644
--- a/apps/website/public/learn/angle-and-magnitude/index.html
+++ b/apps/website/public/learn/angle-and-magnitude/index.html
@@ -439,5 +439,6 @@ <h1>Angle and magnitude</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Adding numbers</h3><p>Why addition is slow: the carry that walks up the word, the dedicated chain in every slice, prefix networks and carry-save trees.</p><ol><li><a href="/learn/the-carry-chain/">01 The carry chain</a></li><li><a href="/learn/prefix-adders/">02 Looking ahead: prefix adders</a></li><li><a href="/learn/carry-save/">03 Three numbers at once</a></li></ol></li><li><h3>Module 2 · Multiplying</h3><p>A product is a sum of shifted copies; Booth recoding halves them, and the DSP48E1 slice does the rest in one block.</p><ol><li><a href="/learn/shift-and-add/">04 Shift and add</a></li><li><a href="/learn/booths-trick/">05 Booth's trick</a></li><li><a href="/learn/the-dsp48e1-slice/">06 The DSP48E1 slice</a></li></ol></li><li><h3>Module 3 · Fixed point</h3><p>Where the binary point sits, what rounding does to a value and to its average, and what each bit of a quantizer buys.</p><ol><li><a href="/learn/q-formats/">07 Q formats</a></li><li><a href="/learn/rounding-and-saturation/">08 Rounding and saturation</a></li><li><a href="/learn/quantization-noise/">09 Quantization noise</a></li></ol></li><li><h3>Module 4 · Functions in hardware</h3><p>Sine, cosine, angle and length from shifts and adds, and when a table is the better answer.</p><ol><li><a href="/learn/rotating-by-shifts/">10 Rotating by shifts</a></li><li class="on" aria-current="page"><a href="/learn/angle-and-magnitude/">11 Angle and magnitude</a></li><li><a href="/learn/tables-and-interpolation/">12 Tables and interpolation</a></li></ol></li><li><h3>Module 5 · Signals and sampling</h3><p>What sampling does to a frequency, how an oscillator is built from an adder, and what a DFT bin measures.</p><ol><li><a href="/learn/sampling-and-aliasing/">13 Sampling and aliasing</a></li><li><a href="/learn/the-nco/">14 A numerically controlled oscillator</a></li><li><a href="/learn/bins-of-a-spectrum/">15 Bins of a spectrum</a></li></ol></li><li><h3>Module 6 · FIR filters</h3><p>The moving average, a windowed-sinc design with integer taps, and folding the taps onto DSP slices.</p><ol><li><a href="/learn/the-moving-average/">16 The moving average</a></li><li><a href="/learn/designing-taps/">17 Designing taps</a></li><li><a href="/learn/folding-taps-onto-slices/">18 Folding taps onto slices</a></li></ol></li><li><h3>Module 7 · Multirate</h3><p>Lowering the sample rate without folding noise in: decimation, the CIC filter and the polyphase form.</p><ol><li><a href="/learn/lowering-the-rate/">19 Lowering the rate</a></li><li><a href="/learn/the-cic-filter/">20 The CIC filter</a></li><li><a href="/learn/polyphase/">21 Polyphase</a></li></ol></li><li><h3>Module 8 · The FFT</h3><p>N log N instead of N^2: butterflies, the bit-reversed input order and the bits each stage adds.</p><ol><li><a href="/learn/the-butterfly/">22 The butterfly</a></li><li><a href="/learn/twiddles-and-bit-reversal/">23 Twiddles and bit reversal</a></li><li><a href="/learn/growth-through-the-stages/">24 Growth through the stages</a></li></ol></li><li><h3>Module 9 · On the bench</h3><p>From the arithmetic to the board: a correlator from the modem spec, a budget of slices and timing, and the capstone filter.</p><ol><li><a href="/learn/a-correlator-finds-the-frame/">25 A correlator finds the frame</a></li><li><a href="/learn/does-it-fit-and-close-timing/">26 Does it fit and close timing?</a></li><li><a href="/learn/capstone-a-filter-on-the-board/">27 Capstone: a filter on the board</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/arithmetic-and-dsp/index.html b/apps/website/public/learn/arithmetic-and-dsp/index.html
index d1dd70ba9d..8256bbce67 100644
--- a/apps/website/public/learn/arithmetic-and-dsp/index.html
+++ b/apps/website/public/learn/arithmetic-and-dsp/index.html
@@ -67,5 +67,6 @@ <h1>Arithmetic and DSP on the FPGA: from the carry chain to a filter</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Adding numbers</h3><p>Why addition is slow: the carry that walks up the word, the dedicated chain in every slice, prefix networks and carry-save trees.</p><ol><li><a href="/learn/the-carry-chain/">01 The carry chain</a></li><li><a href="/learn/prefix-adders/">02 Looking ahead: prefix adders</a></li><li><a href="/learn/carry-save/">03 Three numbers at once</a></li></ol></li><li><h3>Module 2 · Multiplying</h3><p>A product is a sum of shifted copies; Booth recoding halves them, and the DSP48E1 slice does the rest in one block.</p><ol><li><a href="/learn/shift-and-add/">04 Shift and add</a></li><li><a href="/learn/booths-trick/">05 Booth's trick</a></li><li><a href="/learn/the-dsp48e1-slice/">06 The DSP48E1 slice</a></li></ol></li><li><h3>Module 3 · Fixed point</h3><p>Where the binary point sits, what rounding does to a value and to its average, and what each bit of a quantizer buys.</p><ol><li><a href="/learn/q-formats/">07 Q formats</a></li><li><a href="/learn/rounding-and-saturation/">08 Rounding and saturation</a></li><li><a href="/learn/quantization-noise/">09 Quantization noise</a></li></ol></li><li><h3>Module 4 · Functions in hardware</h3><p>Sine, cosine, angle and length from shifts and adds, and when a table is the better answer.</p><ol><li><a href="/learn/rotating-by-shifts/">10 Rotating by shifts</a></li><li><a href="/learn/angle-and-magnitude/">11 Angle and magnitude</a></li><li><a href="/learn/tables-and-interpolation/">12 Tables and interpolation</a></li></ol></li><li><h3>Module 5 · Signals and sampling</h3><p>What sampling does to a frequency, how an oscillator is built from an adder, and what a DFT bin measures.</p><ol><li><a href="/learn/sampling-and-aliasing/">13 Sampling and aliasing</a></li><li><a href="/learn/the-nco/">14 A numerically controlled oscillator</a></li><li><a href="/learn/bins-of-a-spectrum/">15 Bins of a spectrum</a></li></ol></li><li><h3>Module 6 · FIR filters</h3><p>The moving average, a windowed-sinc design with integer taps, and folding the taps onto DSP slices.</p><ol><li><a href="/learn/the-moving-average/">16 The moving average</a></li><li><a href="/learn/designing-taps/">17 Designing taps</a></li><li><a href="/learn/folding-taps-onto-slices/">18 Folding taps onto slices</a></li></ol></li><li><h3>Module 7 · Multirate</h3><p>Lowering the sample rate without folding noise in: decimation, the CIC filter and the polyphase form.</p><ol><li><a href="/learn/lowering-the-rate/">19 Lowering the rate</a></li><li><a href="/learn/the-cic-filter/">20 The CIC filter</a></li><li><a href="/learn/polyphase/">21 Polyphase</a></li></ol></li><li><h3>Module 8 · The FFT</h3><p>N log N instead of N^2: butterflies, the bit-reversed input order and the bits each stage adds.</p><ol><li><a href="/learn/the-butterfly/">22 The butterfly</a></li><li><a href="/learn/twiddles-and-bit-reversal/">23 Twiddles and bit reversal</a></li><li><a href="/learn/growth-through-the-stages/">24 Growth through the stages</a></li></ol></li><li><h3>Module 9 · On the bench</h3><p>From the arithmetic to the board: a correlator from the modem spec, a budget of slices and timing, and the capstone filter.</p><ol><li><a href="/learn/a-correlator-finds-the-frame/">25 A correlator finds the frame</a></li><li><a href="/learn/does-it-fit-and-close-timing/">26 Does it fit and close timing?</a></li><li><a href="/learn/capstone-a-filter-on-the-board/">27 Capstone: a filter on the board</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/arrays/index.html b/apps/website/public/learn/arrays/index.html
index 300fb37f4b..ec5d0db8c5 100644
--- a/apps/website/public/learn/arrays/index.html
+++ b/apps/website/public/learn/arrays/index.html
@@ -84,5 +84,6 @@ <h1>Arrays</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · A spec file</h3><p>What a .t27 file holds: a module line, prose lines and comments.</p><ol><li><a href="/learn/what-a-spec-is/">01 What a spec is</a></li><li><a href="/learn/the-module-line/">02 The module line</a></li><li><a href="/learn/comments-and-prose/">03 Comments and prose lines</a></li></ol></li><li><h3>Module 2 · Values and types</h3><p>Constants, the width of a whole number, and true, false and text.</p><ol><li><a href="/learn/constants/">04 Constants</a></li><li><a href="/learn/whole-numbers/">05 Whole numbers and their widths</a></li><li><a href="/learn/true-false-and-text/">06 True, false and text</a></li></ol></li><li><h3>Module 3 · Arrays, trits and expressions</h3><p>Lists of values, the three values of a trit, and what operators compute.</p><ol><li class="on" aria-current="page"><a href="/learn/arrays/">07 Arrays</a></li><li><a href="/learn/trits/">08 Trits: minus one, zero, one</a></li><li><a href="/learn/expressions/">09 Expressions</a></li></ol></li><li><h3>Module 4 · Tests and functions</h3><p>A test block, several small tests in one spec, and a function with typed inputs.</p><ol><li><a href="/learn/a-test-block/">10 A test block</a></li><li><a href="/learn/many-tests/">11 Many tests in one spec</a></li><li><a href="/learn/functions/">12 Functions</a></li></ol></li><li><h3>Module 5 · Inside a function</h3><p>Names with let and var, choices with if and switch, and loops.</p><ol><li><a href="/learn/local-names/">13 Local names: let and var</a></li><li><a href="/learn/if-and-else/">14 Choices: if, else and switch</a></li><li><a href="/learn/loops/">15 Loops: while and for</a></li></ol></li><li><h3>Module 6 · Visibility and your own types</h3><p>What pub shares with other modules, and the structs and enums you declare.</p><ol><li><a href="/learn/pub-or-private/">16 pub or private</a></li><li><a href="/learn/structs/">17 Structs</a></li><li><a href="/learn/enums/">18 Enums</a></li></ol></li><li><h3>Module 7 · Modules, rules and gen-ts</h3><p>Reading another module, stating a rule that always holds, and a spec turned into TypeScript.</p><ol><li><a href="/learn/use-other-modules/">19 Using other modules</a></li><li><a href="/learn/invariants/">20 Invariants and benches</a></li><li><a href="/learn/gen-ts/">21 gen-ts: a spec becomes TypeScript</a></li></ol></li><li><h3>Module 8 · The compiler and tri</h3><p>The backends t27c writes, the stages it reads a spec in, and the tri command.</p><ol><li><a href="/learn/seven-backends/">22 Seven backends</a></li><li><a href="/learn/stages-of-t27c/">23 The stages of t27c</a></li><li><a href="/learn/tri-commands/">24 tri: the command line</a></li></ol></li><li><h3>Module 9 · Errors, a program, what next</h3><p>Reading a compiler error, one small complete program, and the course to take next.</p><ol><li><a href="/learn/reading-errors/">25 Reading compiler errors</a></li><li><a href="/learn/a-small-program/">26 A small complete program</a></li><li><a href="/learn/where-next/">27 Where to go next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/assertions/index.html b/apps/website/public/learn/assertions/index.html
index bf3d772f3e..7963bbeaa7 100644
--- a/apps/website/public/learn/assertions/index.html
+++ b/apps/website/public/learn/assertions/index.html
@@ -424,5 +424,6 @@ <h1>Assertions</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Why verify</h3><p>Designs that compile and are wrong, the model that decides, and the plan written before the code.</p><ol><li><a href="/learn/bugs-that-compile/">01 Bugs that compile</a></li><li><a href="/learn/golden-model/">02 The golden model</a></li><li><a href="/learn/plan-before-code/">03 Plan before code</a></li></ol></li><li><h3>Module 2 · Testbenches</h3><p>Stimulus, checks and a verdict, written as one spec beside the design it judges.</p><ol><li><a href="/learn/anatomy-of-a-tb/">04 Anatomy of a testbench</a></li><li><a href="/learn/self-checking/">05 Self-checking, not staring</a></li><li><a href="/learn/directed-vs-random/">06 Directed versus random</a></li></ol></li><li><h3>Module 3 · Waveforms</h3><p>A trace of every signal, read the way a hardware engineer reads it, and two runs compared.</p><ol><li><a href="/learn/vcd-format/">07 The VCD format</a></li><li><a href="/learn/debug-with-waves/">08 Debug with waves</a></li><li><a href="/learn/compare-traces/">09 Compare two traces</a></li></ol></li><li><h3>Module 4 · Conformance vectors</h3><p>Cases with the answer written beside them, kept where the compiler can reach them.</p><ol><li><a href="/learn/what-a-vector-is/">10 What a vector is</a></li><li><a href="/learn/golden-ruler-vectors/">11 Golden ruler vectors</a></li><li><a href="/learn/end-to-end-scenarios/">12 End-to-end scenarios</a></li></ol></li><li><h3>Module 5 · Cosimulation</h3><p>Spec, simulator and board agreeing on the bench Artix-7 XC7A200T, and what to do when they do not.</p><ol><li><a href="/learn/three-way-agreement/">13 Three-way agreement</a></li><li><a href="/learn/simulator-internals/">14 Inside the simulator</a></li><li><a href="/learn/when-they-disagree/">15 When they disagree</a></li></ol></li><li><h3>Module 6 · Coverage</h3><p>What the tests touched: lines, toggles, states, and what that number hides.</p><ol><li><a href="/learn/line-and-toggle/">16 Line and toggle</a></li><li><a href="/learn/fsm-coverage/">17 FSM coverage</a></li><li><a href="/learn/coverage-lies/">18 Coverage lies</a></li></ol></li><li><h3>Module 7 · Formal</h3><p>Assertions that hold every cycle, bounded search for a counterexample, and why a proof needs induction.</p><ol><li class="on" aria-current="page"><a href="/learn/assertions/">19 Assertions</a></li><li><a href="/learn/bounded-model-checking/">20 Bounded model checking</a></li><li><a href="/learn/induction/">21 Induction</a></li></ol></li><li><h3>Module 8 · Mutation</h3><p>Break the design on purpose and count what the tests catch.</p><ol><li><a href="/learn/kill-the-mutant/">22 Kill the mutant</a></li><li><a href="/learn/surviving-mutants/">23 Surviving mutants</a></li><li><a href="/learn/test-strength/">24 Test strength</a></li></ol></li><li><h3>Module 9 · Sign-off</h3><p>One command, every receipt, a clean verdict you can show.</p><ol><li><a href="/learn/build-verify/">25 Build and verify</a></li><li><a href="/learn/board-selftest/">26 The board self-tests</a></li><li><a href="/learn/verify-capstone/">27 Capstone: verify a module whole</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/back-to-three-values/index.html b/apps/website/public/learn/back-to-three-values/index.html
index 03b5c728df..099ec02a82 100644
--- a/apps/website/public/learn/back-to-three-values/index.html
+++ b/apps/website/public/learn/back-to-three-values/index.html
@@ -103,5 +103,6 @@ <h1>Back to three values</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Lab: our own research</h3><p>A number format of our own, an honest scoreboard, and a model's tables multiplied on the board.</p><ol><li><a href="/learn/t27-number-format/">01 T27: a number format of our own</a></li><li><a href="/learn/an-honest-scoreboard/">02 An honest scoreboard</a></li><li><a href="/learn/a-model-on-the-board/">03 A model's tables on the board</a></li></ol></li><li><h3>Module 2 · AI numbers: the MX block</h3><p>How AI chips keep weights in a few bits: one shared scale per block, the scale byte itself, and what one outlier does to its neighbours.</p><ol><li><a href="/learn/one-scale-per-block/">04 One scale for a block of weights</a></li><li><a href="/learn/scale-byte-native/">05 The scale byte on a real machine</a></li><li><a href="/learn/one-outlier/">06 One outlier, many zeros</a></li></ol></li><li><h3>Module 3 · Ternary weights</h3><p>Weights that are only minus, zero or plus a scale, the five rules a ternary alphabet must pass, and a test pass that checked nothing.</p><ol><li><a href="/learn/phi-scaled-ternary/">07 Weights of minus phi, zero and plus phi</a></li><li><a href="/learn/a-pass-that-checked-nothing/">08 A pass that checked nothing</a></li><li><a href="/learn/five-rules-for-weights/">09 Five rules for a weight alphabet</a></li></ol></li><li><h3>Module 4 · The Ternary Network Float</h3><p>A rule the compiler enforces before any test runs, and a 17-bit float whose exponent is four balanced trits.</p><ol><li><a href="/learn/blocked-is-not-failed/">10 Blocked is not failed</a></li><li><a href="/learn/four-trit-exponent/">11 A float whose exponent is four trits</a></li><li><a href="/learn/negate-twice/">12 Negate twice, get the number back</a></li></ol></li><li><h3>Module 5 · Arithmetic on signed numbers</h3><p>Multiply two signed numbers, add them when their signs differ, and do both at once in a multiply-accumulate.</p><ol><li><a href="/learn/a-sign-is-one-bit/">13 A sign is one bit</a></li><li><a href="/learn/opposite-signs-subtract/">14 Opposite signs subtract</a></li><li><a href="/learn/multiply-then-add/">15 Multiply, then add</a></li></ol></li><li><h3>Module 6 · Parts of a neuron</h3><p>A ReLU that bends at zero, a power of two for softmax, and an argmax that names the answer.</p><ol><li><a href="/learn/a-bend-at-zero/">16 A bend at zero</a></li><li><a href="/learn/two-to-the-x/">17 Two to the x</a></li><li><a href="/learn/pick-the-largest/">18 Pick the largest</a></li></ol></li><li><h3>Module 7 · Learning from a mistake</h3><p>A loss that prices a wrong guess in bits, one step that moves a weight against its gradient, and the hidden layer that XOR needs.</p><ol><li><a href="/learn/a-loss-in-bits/">19 A loss in bits</a></li><li><a href="/learn/one-step-downhill/">20 One step downhill</a></li><li><a href="/learn/xor-needs-a-bend/">21 XOR needs a bend</a></li></ol></li><li><h3>Module 8 · BitNet: ternary networks</h3><p>A threshold that squeezes a sum back to three values, one neuron that becomes a different function when its weights change, and a neuron that reads its inputs 27 trits at a time.</p><ol><li class="on" aria-current="page"><a href="/learn/back-to-three-values/">22 Back to three values</a></li><li><a href="/learn/same-neuron-new-weights/">23 Same neuron, new weights</a></li><li><a href="/learn/a-neuron-in-chunks/">24 A neuron in chunks</a></li></ol></li><li><h3>Module 9 · The ternary MAC as a chip</h3><p>The 27-trit dot product as wires with no register, the same sum added into a register on every clock, and a small whole network to close the course.</p><ol><li><a href="/learn/a-dot-product-in-wires/">25 A dot product made of wires</a></li><li><a href="/learn/add-it-up-every-clock/">26 Add it up on every clock</a></li><li><a href="/learn/a-whole-network/">27 A whole network</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/balanced-ternary/index.html b/apps/website/public/learn/balanced-ternary/index.html
index d77a645a17..0540eb3887 100644
--- a/apps/website/public/learn/balanced-ternary/index.html
+++ b/apps/website/public/learn/balanced-ternary/index.html
@@ -769,5 +769,6 @@ <h1>Balanced ternary</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Three values</h3><p>Why three, how balanced ternary writes every number without a sign, and what flipping and cutting trits do.</p><ol><li><a href="/learn/why-three/">01 Why three</a></li><li class="on" aria-current="page"><a href="/learn/balanced-ternary/">02 Balanced ternary</a></li><li><a href="/learn/flip-and-cut/">03 Flip to negate, cut to round</a></li></ol></li><li><h3>Module 2 · Logic with unknown</h3><p>Kleene's three-valued gates, two gates binary has no twin for, and addition as a pair of tables.</p><ol><li><a href="/learn/kleene-logic/">04 Three-valued logic</a></li><li><a href="/learn/consensus-and-any/">05 Consensus and any</a></li><li><a href="/learn/sum-and-carry/">06 Sum and carry</a></li></ol></li><li><h3>Module 3 · Adding trits</h3><p>The half adder, the full adder and a carry that ripples left, one place at a time.</p><ol><li><a href="/learn/half-adder/">07 The half adder</a></li><li><a href="/learn/full-adder/">08 The full adder</a></li><li><a href="/learn/ripple-carry/">09 Ripple carry</a></li></ol></li><li><h3>Module 4 · Multiplying without a multiplier</h3><p>Copy, drop or flip: a product by one trit, long multiplication, and a MAC that only adds.</p><ol><li><a href="/learn/times-a-trit/">10 Times one trit</a></li><li><a href="/learn/long-multiplication/">11 Long multiplication</a></li><li><a href="/learn/ternary-mac/">12 A ternary MAC</a></li></ol></li><li><h3>Module 5 · Trits in binary memory</h3><p>Two bits per trit, five trits per byte, and the bits a 27-trit word needs.</p><ol><li><a href="/learn/two-bits-per-trit/">13 Two bits per trit</a></li><li><a href="/learn/five-trits-a-byte/">14 Five trits in a byte</a></li><li><a href="/learn/a-word-of-27/">15 A word of 27 trits</a></li></ol></li><li><h3>Module 6 · The TRI-27 instruction word</h3><p>The 32-bit word the Trinity emulator decodes: its fields, its 47 opcodes and its 15-bit immediate.</p><ol><li><a href="/learn/the-instruction-word/">16 The instruction word</a></li><li><a href="/learn/the-opcode-map/">17 The opcode map</a></li><li><a href="/learn/the-immediate/">18 The immediate</a></li></ol></li><li><h3>Module 7 · A ternary machine</h3><p>An ALU built from this course's adders, a three-way jump, and a program you can step.</p><ol><li><a href="/learn/a-ternary-alu/">19 A ternary ALU</a></li><li><a href="/learn/three-way-jump/">20 A three-way jump</a></li><li><a href="/learn/run-a-program/">21 Run a program</a></li></ol></li><li><h3>Module 8 · Three answers</h3><p>Compare at the highest differing trit, find a number in thirds, sort with three-way compares.</p><ol><li><a href="/learn/compare-at-the-top/">22 Compare at the top</a></li><li><a href="/learn/questions-in-thirds/">23 Questions in thirds</a></li><li><a href="/learn/sort-by-three/">24 Sort with three-way compares</a></li></ol></li><li><h3>Module 9 · Ternary neurons</h3><p>Weights of -1, 0, +1: one neuron, a detector and a small layer, with no multiplier anywhere.</p><ol><li><a href="/learn/one-ternary-neuron/">25 One ternary neuron</a></li><li><a href="/learn/a-bar-detector/">26 A bar detector</a></li><li><a href="/learn/a-tiny-layer/">27 A tiny layer</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/baud-and-the-divisor/index.html b/apps/website/public/learn/baud-and-the-divisor/index.html
index 572a77bf1b..74036a5541 100644
--- a/apps/website/public/learn/baud-and-the-divisor/index.html
+++ b/apps/website/public/learn/baud-and-the-divisor/index.html
@@ -299,5 +299,6 @@ <h1>Baud and the divisor</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a bus is</h3><p>Why a bus exists at all: a conversation on wires, framed and addressed, and who is allowed to talk.</p><ol><li><a href="/learn/wires-that-talk/">01 Wires that talk</a></li><li><a href="/learn/framing-a-conversation/">02 Framing a conversation</a></li><li><a href="/learn/master-and-servant/">03 Master and servant</a></li></ol></li><li><h3>Module 2 · UART</h3><p>The two-wire bus with no clock: the frame, the divisor that sets the rate, and the status a driver polls.</p><ol><li><a href="/learn/the-frame/">04 The frame: start, data, stop</a></li><li class="on" aria-current="page"><a href="/learn/baud-and-the-divisor/">05 Baud and the divisor</a></li><li><a href="/learn/status-and-fifos/">06 Status and FIFOs</a></li></ol></li><li><h3>Module 3 · SPI</h3><p>The clocked conversation: four modes, a prescaler ladder, and a chip select per servant.</p><ol><li><a href="/learn/modes-cpol-cpha/">07 Modes: CPOL and CPHA</a></li><li><a href="/learn/the-prescaler-ladder/">08 The prescaler ladder</a></li><li><a href="/learn/chip-select-and-widths/">09 Chip select and widths</a></li></ol></li><li><h3>Module 4 · APB</h3><p>The register bus: PSEL and PENABLE, strobes and wait states, and how many address bits a peripheral count costs.</p><ol><li><a href="/learn/psel-and-penable/">10 PSEL and PENABLE</a></li><li><a href="/learn/strobes-and-waits/">11 Strobes and waits</a></li><li><a href="/learn/address-decode/">12 Address decode</a></li></ol></li><li><h3>Module 5 · AXI4</h3><p>The five channels: address, data and response in both directions, lite or full, bursts and IDs.</p><ol><li><a href="/learn/five-channels/">13 Five channels</a></li><li><a href="/learn/lite-or-full/">14 Lite or full</a></li><li><a href="/learn/bursts-and-ids/">15 Bursts and IDs</a></li></ol></li><li><h3>Module 6 · Memory</h3><p>What sits on the far side of every bus: memory maps, port kinds, and latency that a wait state must cover.</p><ol><li><a href="/learn/memory-maps/">16 Memory maps</a></li><li><a href="/learn/ports-and-kinds/">17 Ports and kinds</a></li><li><a href="/learn/latency/">18 Latency</a></li></ol></li><li><h3>Module 7 · Bridges</h3><p>Why designs grow more than one bus, and the packet bridge that moves work between them.</p><ol><li><a href="/learn/why-bridges/">19 Why bridges</a></li><li><a href="/learn/the-packet-bridge/">20 The packet bridge</a></li><li><a href="/learn/everything-in-one-bench/">21 Everything in one bench</a></li></ol></li><li><h3>Module 8 · Ethernet</h3><p>Frames, the frame check sequence, RGMII timing, and the pre-registered steps of a real bring-up.</p><ol><li><a href="/learn/frames-and-the-fcs/">22 Frames and the FCS</a></li><li><a href="/learn/rgmii-timing/">23 RGMII timing</a></li><li><a href="/learn/bring-up-on-the-bench/">24 Bring-up on the bench</a></li></ol></li><li><h3>Module 9 · The bench</h3><p>The discipline that guards real hardware: who holds the IO, taking and giving the lock, and what runs next.</p><ol><li><a href="/learn/who-holds-the-io/">25 Who holds the IO</a></li><li><a href="/learn/claim-and-release/">26 Claim and release</a></li><li><a href="/learn/capstone-what-runs-next/">27 Capstone: what runs next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/bins-of-a-spectrum/index.html b/apps/website/public/learn/bins-of-a-spectrum/index.html
index e0b7c2a8bf..80171b2bc5 100644
--- a/apps/website/public/learn/bins-of-a-spectrum/index.html
+++ b/apps/website/public/learn/bins-of-a-spectrum/index.html
@@ -1122,5 +1122,6 @@ <h1>Bins of a spectrum</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Adding numbers</h3><p>Why addition is slow: the carry that walks up the word, the dedicated chain in every slice, prefix networks and carry-save trees.</p><ol><li><a href="/learn/the-carry-chain/">01 The carry chain</a></li><li><a href="/learn/prefix-adders/">02 Looking ahead: prefix adders</a></li><li><a href="/learn/carry-save/">03 Three numbers at once</a></li></ol></li><li><h3>Module 2 · Multiplying</h3><p>A product is a sum of shifted copies; Booth recoding halves them, and the DSP48E1 slice does the rest in one block.</p><ol><li><a href="/learn/shift-and-add/">04 Shift and add</a></li><li><a href="/learn/booths-trick/">05 Booth's trick</a></li><li><a href="/learn/the-dsp48e1-slice/">06 The DSP48E1 slice</a></li></ol></li><li><h3>Module 3 · Fixed point</h3><p>Where the binary point sits, what rounding does to a value and to its average, and what each bit of a quantizer buys.</p><ol><li><a href="/learn/q-formats/">07 Q formats</a></li><li><a href="/learn/rounding-and-saturation/">08 Rounding and saturation</a></li><li><a href="/learn/quantization-noise/">09 Quantization noise</a></li></ol></li><li><h3>Module 4 · Functions in hardware</h3><p>Sine, cosine, angle and length from shifts and adds, and when a table is the better answer.</p><ol><li><a href="/learn/rotating-by-shifts/">10 Rotating by shifts</a></li><li><a href="/learn/angle-and-magnitude/">11 Angle and magnitude</a></li><li><a href="/learn/tables-and-interpolation/">12 Tables and interpolation</a></li></ol></li><li><h3>Module 5 · Signals and sampling</h3><p>What sampling does to a frequency, how an oscillator is built from an adder, and what a DFT bin measures.</p><ol><li><a href="/learn/sampling-and-aliasing/">13 Sampling and aliasing</a></li><li><a href="/learn/the-nco/">14 A numerically controlled oscillator</a></li><li class="on" aria-current="page"><a href="/learn/bins-of-a-spectrum/">15 Bins of a spectrum</a></li></ol></li><li><h3>Module 6 · FIR filters</h3><p>The moving average, a windowed-sinc design with integer taps, and folding the taps onto DSP slices.</p><ol><li><a href="/learn/the-moving-average/">16 The moving average</a></li><li><a href="/learn/designing-taps/">17 Designing taps</a></li><li><a href="/learn/folding-taps-onto-slices/">18 Folding taps onto slices</a></li></ol></li><li><h3>Module 7 · Multirate</h3><p>Lowering the sample rate without folding noise in: decimation, the CIC filter and the polyphase form.</p><ol><li><a href="/learn/lowering-the-rate/">19 Lowering the rate</a></li><li><a href="/learn/the-cic-filter/">20 The CIC filter</a></li><li><a href="/learn/polyphase/">21 Polyphase</a></li></ol></li><li><h3>Module 8 · The FFT</h3><p>N log N instead of N^2: butterflies, the bit-reversed input order and the bits each stage adds.</p><ol><li><a href="/learn/the-butterfly/">22 The butterfly</a></li><li><a href="/learn/twiddles-and-bit-reversal/">23 Twiddles and bit reversal</a></li><li><a href="/learn/growth-through-the-stages/">24 Growth through the stages</a></li></ol></li><li><h3>Module 9 · On the bench</h3><p>From the arithmetic to the board: a correlator from the modem spec, a budget of slices and timing, and the capstone filter.</p><ol><li><a href="/learn/a-correlator-finds-the-frame/">25 A correlator finds the frame</a></li><li><a href="/learn/does-it-fit-and-close-timing/">26 Does it fit and close timing?</a></li><li><a href="/learn/capstone-a-filter-on-the-board/">27 Capstone: a filter on the board</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/bits-and-trits/index.html b/apps/website/public/learn/bits-and-trits/index.html
index d036210bd5..d747323fa6 100644
--- a/apps/website/public/learn/bits-and-trits/index.html
+++ b/apps/website/public/learn/bits-and-trits/index.html
@@ -140,5 +140,6 @@ <h1>Bits, trits and GF16</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The chip</h3><p>What an FPGA is, which chip we use, and how its pins meet the board.</p><ol><li><a href="/learn/fpga-is-a-grid/">01 An FPGA is a grid you configure</a></li><li><a href="/learn/which-chip/">02 Which chip is this?</a></li><li><a href="/learn/pins-and-package/">03 Pins, balls and the package</a></li></ol></li><li><h3>Module 2 · Numbers in hardware</h3><p>Bits, trits and number formats, and what arithmetic costs in logic.</p><ol><li class="on" aria-current="page"><a href="/learn/bits-and-trits/">04 Bits, trits and GF16</a></li><li><a href="/learn/what-arithmetic-costs/">05 What arithmetic costs</a></li><li><a href="/learn/how-much-fits/">06 How much fits on the chip</a></li></ol></li><li><h3>Module 3 · Your t27 program</h3><p>Write a spec, test it, and see why compiling is not the same as being right.</p><ol><li><a href="/learn/hello-t27/">07 Hello, t27</a></li><li><a href="/learn/tests-are-the-spec/">08 Tests are part of the spec</a></li><li><a href="/learn/compiles-is-not-correct/">09 Compiles is not the same as correct</a></li></ol></li><li><h3>Module 4 · Inside t27c</h3><p>How the compiler reads a spec and what it writes, including the native t27b.</p><ol><li><a href="/learn/compiler-stages/">10 The stages of t27c</a></li><li><a href="/learn/t27c-on-a-real-machine/">11 t27c on a real machine</a></li><li><a href="/learn/seven-backends-and-t27b/">12 Seven backends, and t27b</a></li></ol></li><li><h3>Module 5 · From spec to hardware</h3><p>The Verilog t27c writes, the cells it becomes, and what one LUT does.</p><ol><li><a href="/learn/spec-to-verilog/">13 From a spec to Verilog</a></li><li><a href="/learn/count-the-cells/">14 Count the cells</a></li><li><a href="/learn/one-lut-up-close/">15 One LUT, up close</a></li></ol></li><li><h3>Module 6 · Reading synthesis</h3><p>What yosys reports about your design, and which warnings matter.</p><ol><li><a href="/learn/synthesis-warnings/">16 Synthesis warnings that matter</a></li><li><a href="/learn/state-machines/">17 State machines</a></li><li><a href="/learn/where-the-luts-go/">18 Where the LUTs go</a></li></ol></li><li><h3>Module 7 · Place, route, timing</h3><p>Where the cells land on the die, and whether the clock is met.</p><ol><li><a href="/learn/placement/">19 Placement: where the cells land</a></li><li><a href="/learn/logic-depth/">20 Logic depth</a></li><li><a href="/learn/timing-slack/">21 Timing and slack</a></li></ol></li><li><h3>Module 8 · The bitstream</h3><p>How a routed design becomes the bits the chip loads, with no Vivado.</p><ol><li><a href="/learn/fasm-to-frames/">22 From FASM to frames</a></li><li><a href="/learn/a-build-receipt/">23 A build, itemised</a></li><li><a href="/learn/read-a-bitstream/">24 Read a bitstream</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>Ask the chip who it is, load the bits, and check them.</p><ol><li><a href="/learn/jtag-asks-the-chip/">25 JTAG asks the chip</a></li><li><a href="/learn/same-bits-on-the-board/">26 The same bits, on the board</a></li><li><a href="/learn/capstone/">27 Capstone: out-guess yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/blocked-is-not-failed/index.html b/apps/website/public/learn/blocked-is-not-failed/index.html
index ab3c4b04fe..614039425c 100644
--- a/apps/website/public/learn/blocked-is-not-failed/index.html
+++ b/apps/website/public/learn/blocked-is-not-failed/index.html
@@ -616,5 +616,6 @@ <h1>Blocked is not failed</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Lab: our own research</h3><p>A number format of our own, an honest scoreboard, and a model's tables multiplied on the board.</p><ol><li><a href="/learn/t27-number-format/">01 T27: a number format of our own</a></li><li><a href="/learn/an-honest-scoreboard/">02 An honest scoreboard</a></li><li><a href="/learn/a-model-on-the-board/">03 A model's tables on the board</a></li></ol></li><li><h3>Module 2 · AI numbers: the MX block</h3><p>How AI chips keep weights in a few bits: one shared scale per block, the scale byte itself, and what one outlier does to its neighbours.</p><ol><li><a href="/learn/one-scale-per-block/">04 One scale for a block of weights</a></li><li><a href="/learn/scale-byte-native/">05 The scale byte on a real machine</a></li><li><a href="/learn/one-outlier/">06 One outlier, many zeros</a></li></ol></li><li><h3>Module 3 · Ternary weights</h3><p>Weights that are only minus, zero or plus a scale, the five rules a ternary alphabet must pass, and a test pass that checked nothing.</p><ol><li><a href="/learn/phi-scaled-ternary/">07 Weights of minus phi, zero and plus phi</a></li><li><a href="/learn/a-pass-that-checked-nothing/">08 A pass that checked nothing</a></li><li><a href="/learn/five-rules-for-weights/">09 Five rules for a weight alphabet</a></li></ol></li><li><h3>Module 4 · The Ternary Network Float</h3><p>A rule the compiler enforces before any test runs, and a 17-bit float whose exponent is four balanced trits.</p><ol><li class="on" aria-current="page"><a href="/learn/blocked-is-not-failed/">10 Blocked is not failed</a></li><li><a href="/learn/four-trit-exponent/">11 A float whose exponent is four trits</a></li><li><a href="/learn/negate-twice/">12 Negate twice, get the number back</a></li></ol></li><li><h3>Module 5 · Arithmetic on signed numbers</h3><p>Multiply two signed numbers, add them when their signs differ, and do both at once in a multiply-accumulate.</p><ol><li><a href="/learn/a-sign-is-one-bit/">13 A sign is one bit</a></li><li><a href="/learn/opposite-signs-subtract/">14 Opposite signs subtract</a></li><li><a href="/learn/multiply-then-add/">15 Multiply, then add</a></li></ol></li><li><h3>Module 6 · Parts of a neuron</h3><p>A ReLU that bends at zero, a power of two for softmax, and an argmax that names the answer.</p><ol><li><a href="/learn/a-bend-at-zero/">16 A bend at zero</a></li><li><a href="/learn/two-to-the-x/">17 Two to the x</a></li><li><a href="/learn/pick-the-largest/">18 Pick the largest</a></li></ol></li><li><h3>Module 7 · Learning from a mistake</h3><p>A loss that prices a wrong guess in bits, one step that moves a weight against its gradient, and the hidden layer that XOR needs.</p><ol><li><a href="/learn/a-loss-in-bits/">19 A loss in bits</a></li><li><a href="/learn/one-step-downhill/">20 One step downhill</a></li><li><a href="/learn/xor-needs-a-bend/">21 XOR needs a bend</a></li></ol></li><li><h3>Module 8 · BitNet: ternary networks</h3><p>A threshold that squeezes a sum back to three values, one neuron that becomes a different function when its weights change, and a neuron that reads its inputs 27 trits at a time.</p><ol><li><a href="/learn/back-to-three-values/">22 Back to three values</a></li><li><a href="/learn/same-neuron-new-weights/">23 Same neuron, new weights</a></li><li><a href="/learn/a-neuron-in-chunks/">24 A neuron in chunks</a></li></ol></li><li><h3>Module 9 · The ternary MAC as a chip</h3><p>The 27-trit dot product as wires with no register, the same sum added into a register on every clock, and a small whole network to close the course.</p><ol><li><a href="/learn/a-dot-product-in-wires/">25 A dot product made of wires</a></li><li><a href="/learn/add-it-up-every-clock/">26 Add it up on every clock</a></li><li><a href="/learn/a-whole-network/">27 A whole network</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/board-selftest/index.html b/apps/website/public/learn/board-selftest/index.html
index a5ac504fd1..aee40e6ce0 100644
--- a/apps/website/public/learn/board-selftest/index.html
+++ b/apps/website/public/learn/board-selftest/index.html
@@ -110,5 +110,6 @@ <h1>The board self-tests</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Why verify</h3><p>Designs that compile and are wrong, the model that decides, and the plan written before the code.</p><ol><li><a href="/learn/bugs-that-compile/">01 Bugs that compile</a></li><li><a href="/learn/golden-model/">02 The golden model</a></li><li><a href="/learn/plan-before-code/">03 Plan before code</a></li></ol></li><li><h3>Module 2 · Testbenches</h3><p>Stimulus, checks and a verdict, written as one spec beside the design it judges.</p><ol><li><a href="/learn/anatomy-of-a-tb/">04 Anatomy of a testbench</a></li><li><a href="/learn/self-checking/">05 Self-checking, not staring</a></li><li><a href="/learn/directed-vs-random/">06 Directed versus random</a></li></ol></li><li><h3>Module 3 · Waveforms</h3><p>A trace of every signal, read the way a hardware engineer reads it, and two runs compared.</p><ol><li><a href="/learn/vcd-format/">07 The VCD format</a></li><li><a href="/learn/debug-with-waves/">08 Debug with waves</a></li><li><a href="/learn/compare-traces/">09 Compare two traces</a></li></ol></li><li><h3>Module 4 · Conformance vectors</h3><p>Cases with the answer written beside them, kept where the compiler can reach them.</p><ol><li><a href="/learn/what-a-vector-is/">10 What a vector is</a></li><li><a href="/learn/golden-ruler-vectors/">11 Golden ruler vectors</a></li><li><a href="/learn/end-to-end-scenarios/">12 End-to-end scenarios</a></li></ol></li><li><h3>Module 5 · Cosimulation</h3><p>Spec, simulator and board agreeing on the bench Artix-7 XC7A200T, and what to do when they do not.</p><ol><li><a href="/learn/three-way-agreement/">13 Three-way agreement</a></li><li><a href="/learn/simulator-internals/">14 Inside the simulator</a></li><li><a href="/learn/when-they-disagree/">15 When they disagree</a></li></ol></li><li><h3>Module 6 · Coverage</h3><p>What the tests touched: lines, toggles, states, and what that number hides.</p><ol><li><a href="/learn/line-and-toggle/">16 Line and toggle</a></li><li><a href="/learn/fsm-coverage/">17 FSM coverage</a></li><li><a href="/learn/coverage-lies/">18 Coverage lies</a></li></ol></li><li><h3>Module 7 · Formal</h3><p>Assertions that hold every cycle, bounded search for a counterexample, and why a proof needs induction.</p><ol><li><a href="/learn/assertions/">19 Assertions</a></li><li><a href="/learn/bounded-model-checking/">20 Bounded model checking</a></li><li><a href="/learn/induction/">21 Induction</a></li></ol></li><li><h3>Module 8 · Mutation</h3><p>Break the design on purpose and count what the tests catch.</p><ol><li><a href="/learn/kill-the-mutant/">22 Kill the mutant</a></li><li><a href="/learn/surviving-mutants/">23 Surviving mutants</a></li><li><a href="/learn/test-strength/">24 Test strength</a></li></ol></li><li><h3>Module 9 · Sign-off</h3><p>One command, every receipt, a clean verdict you can show.</p><ol><li><a href="/learn/build-verify/">25 Build and verify</a></li><li class="on" aria-current="page"><a href="/learn/board-selftest/">26 The board self-tests</a></li><li><a href="/learn/verify-capstone/">27 Capstone: verify a module whole</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/booths-trick/index.html b/apps/website/public/learn/booths-trick/index.html
index 8cf844c736..bc13a87a07 100644
--- a/apps/website/public/learn/booths-trick/index.html
+++ b/apps/website/public/learn/booths-trick/index.html
@@ -372,5 +372,6 @@ <h1>Booth's trick</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Adding numbers</h3><p>Why addition is slow: the carry that walks up the word, the dedicated chain in every slice, prefix networks and carry-save trees.</p><ol><li><a href="/learn/the-carry-chain/">01 The carry chain</a></li><li><a href="/learn/prefix-adders/">02 Looking ahead: prefix adders</a></li><li><a href="/learn/carry-save/">03 Three numbers at once</a></li></ol></li><li><h3>Module 2 · Multiplying</h3><p>A product is a sum of shifted copies; Booth recoding halves them, and the DSP48E1 slice does the rest in one block.</p><ol><li><a href="/learn/shift-and-add/">04 Shift and add</a></li><li class="on" aria-current="page"><a href="/learn/booths-trick/">05 Booth's trick</a></li><li><a href="/learn/the-dsp48e1-slice/">06 The DSP48E1 slice</a></li></ol></li><li><h3>Module 3 · Fixed point</h3><p>Where the binary point sits, what rounding does to a value and to its average, and what each bit of a quantizer buys.</p><ol><li><a href="/learn/q-formats/">07 Q formats</a></li><li><a href="/learn/rounding-and-saturation/">08 Rounding and saturation</a></li><li><a href="/learn/quantization-noise/">09 Quantization noise</a></li></ol></li><li><h3>Module 4 · Functions in hardware</h3><p>Sine, cosine, angle and length from shifts and adds, and when a table is the better answer.</p><ol><li><a href="/learn/rotating-by-shifts/">10 Rotating by shifts</a></li><li><a href="/learn/angle-and-magnitude/">11 Angle and magnitude</a></li><li><a href="/learn/tables-and-interpolation/">12 Tables and interpolation</a></li></ol></li><li><h3>Module 5 · Signals and sampling</h3><p>What sampling does to a frequency, how an oscillator is built from an adder, and what a DFT bin measures.</p><ol><li><a href="/learn/sampling-and-aliasing/">13 Sampling and aliasing</a></li><li><a href="/learn/the-nco/">14 A numerically controlled oscillator</a></li><li><a href="/learn/bins-of-a-spectrum/">15 Bins of a spectrum</a></li></ol></li><li><h3>Module 6 · FIR filters</h3><p>The moving average, a windowed-sinc design with integer taps, and folding the taps onto DSP slices.</p><ol><li><a href="/learn/the-moving-average/">16 The moving average</a></li><li><a href="/learn/designing-taps/">17 Designing taps</a></li><li><a href="/learn/folding-taps-onto-slices/">18 Folding taps onto slices</a></li></ol></li><li><h3>Module 7 · Multirate</h3><p>Lowering the sample rate without folding noise in: decimation, the CIC filter and the polyphase form.</p><ol><li><a href="/learn/lowering-the-rate/">19 Lowering the rate</a></li><li><a href="/learn/the-cic-filter/">20 The CIC filter</a></li><li><a href="/learn/polyphase/">21 Polyphase</a></li></ol></li><li><h3>Module 8 · The FFT</h3><p>N log N instead of N^2: butterflies, the bit-reversed input order and the bits each stage adds.</p><ol><li><a href="/learn/the-butterfly/">22 The butterfly</a></li><li><a href="/learn/twiddles-and-bit-reversal/">23 Twiddles and bit reversal</a></li><li><a href="/learn/growth-through-the-stages/">24 Growth through the stages</a></li></ol></li><li><h3>Module 9 · On the bench</h3><p>From the arithmetic to the board: a correlator from the modem spec, a budget of slices and timing, and the capstone filter.</p><ol><li><a href="/learn/a-correlator-finds-the-frame/">25 A correlator finds the frame</a></li><li><a href="/learn/does-it-fit-and-close-timing/">26 Does it fit and close timing?</a></li><li><a href="/learn/capstone-a-filter-on-the-board/">27 Capstone: a filter on the board</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/bounded-model-checking/index.html b/apps/website/public/learn/bounded-model-checking/index.html
index 2fb6557465..f40c45875a 100644
--- a/apps/website/public/learn/bounded-model-checking/index.html
+++ b/apps/website/public/learn/bounded-model-checking/index.html
@@ -424,5 +424,6 @@ <h1>Bounded model checking</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Why verify</h3><p>Designs that compile and are wrong, the model that decides, and the plan written before the code.</p><ol><li><a href="/learn/bugs-that-compile/">01 Bugs that compile</a></li><li><a href="/learn/golden-model/">02 The golden model</a></li><li><a href="/learn/plan-before-code/">03 Plan before code</a></li></ol></li><li><h3>Module 2 · Testbenches</h3><p>Stimulus, checks and a verdict, written as one spec beside the design it judges.</p><ol><li><a href="/learn/anatomy-of-a-tb/">04 Anatomy of a testbench</a></li><li><a href="/learn/self-checking/">05 Self-checking, not staring</a></li><li><a href="/learn/directed-vs-random/">06 Directed versus random</a></li></ol></li><li><h3>Module 3 · Waveforms</h3><p>A trace of every signal, read the way a hardware engineer reads it, and two runs compared.</p><ol><li><a href="/learn/vcd-format/">07 The VCD format</a></li><li><a href="/learn/debug-with-waves/">08 Debug with waves</a></li><li><a href="/learn/compare-traces/">09 Compare two traces</a></li></ol></li><li><h3>Module 4 · Conformance vectors</h3><p>Cases with the answer written beside them, kept where the compiler can reach them.</p><ol><li><a href="/learn/what-a-vector-is/">10 What a vector is</a></li><li><a href="/learn/golden-ruler-vectors/">11 Golden ruler vectors</a></li><li><a href="/learn/end-to-end-scenarios/">12 End-to-end scenarios</a></li></ol></li><li><h3>Module 5 · Cosimulation</h3><p>Spec, simulator and board agreeing on the bench Artix-7 XC7A200T, and what to do when they do not.</p><ol><li><a href="/learn/three-way-agreement/">13 Three-way agreement</a></li><li><a href="/learn/simulator-internals/">14 Inside the simulator</a></li><li><a href="/learn/when-they-disagree/">15 When they disagree</a></li></ol></li><li><h3>Module 6 · Coverage</h3><p>What the tests touched: lines, toggles, states, and what that number hides.</p><ol><li><a href="/learn/line-and-toggle/">16 Line and toggle</a></li><li><a href="/learn/fsm-coverage/">17 FSM coverage</a></li><li><a href="/learn/coverage-lies/">18 Coverage lies</a></li></ol></li><li><h3>Module 7 · Formal</h3><p>Assertions that hold every cycle, bounded search for a counterexample, and why a proof needs induction.</p><ol><li><a href="/learn/assertions/">19 Assertions</a></li><li class="on" aria-current="page"><a href="/learn/bounded-model-checking/">20 Bounded model checking</a></li><li><a href="/learn/induction/">21 Induction</a></li></ol></li><li><h3>Module 8 · Mutation</h3><p>Break the design on purpose and count what the tests catch.</p><ol><li><a href="/learn/kill-the-mutant/">22 Kill the mutant</a></li><li><a href="/learn/surviving-mutants/">23 Surviving mutants</a></li><li><a href="/learn/test-strength/">24 Test strength</a></li></ol></li><li><h3>Module 9 · Sign-off</h3><p>One command, every receipt, a clean verdict you can show.</p><ol><li><a href="/learn/build-verify/">25 Build and verify</a></li><li><a href="/learn/board-selftest/">26 The board self-tests</a></li><li><a href="/learn/verify-capstone/">27 Capstone: verify a module whole</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/bring-up-on-the-bench/index.html b/apps/website/public/learn/bring-up-on-the-bench/index.html
index cbf12929f1..69d90bccff 100644
--- a/apps/website/public/learn/bring-up-on-the-bench/index.html
+++ b/apps/website/public/learn/bring-up-on-the-bench/index.html
@@ -109,5 +109,6 @@ <h1>Bring-up on the bench</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a bus is</h3><p>Why a bus exists at all: a conversation on wires, framed and addressed, and who is allowed to talk.</p><ol><li><a href="/learn/wires-that-talk/">01 Wires that talk</a></li><li><a href="/learn/framing-a-conversation/">02 Framing a conversation</a></li><li><a href="/learn/master-and-servant/">03 Master and servant</a></li></ol></li><li><h3>Module 2 · UART</h3><p>The two-wire bus with no clock: the frame, the divisor that sets the rate, and the status a driver polls.</p><ol><li><a href="/learn/the-frame/">04 The frame: start, data, stop</a></li><li><a href="/learn/baud-and-the-divisor/">05 Baud and the divisor</a></li><li><a href="/learn/status-and-fifos/">06 Status and FIFOs</a></li></ol></li><li><h3>Module 3 · SPI</h3><p>The clocked conversation: four modes, a prescaler ladder, and a chip select per servant.</p><ol><li><a href="/learn/modes-cpol-cpha/">07 Modes: CPOL and CPHA</a></li><li><a href="/learn/the-prescaler-ladder/">08 The prescaler ladder</a></li><li><a href="/learn/chip-select-and-widths/">09 Chip select and widths</a></li></ol></li><li><h3>Module 4 · APB</h3><p>The register bus: PSEL and PENABLE, strobes and wait states, and how many address bits a peripheral count costs.</p><ol><li><a href="/learn/psel-and-penable/">10 PSEL and PENABLE</a></li><li><a href="/learn/strobes-and-waits/">11 Strobes and waits</a></li><li><a href="/learn/address-decode/">12 Address decode</a></li></ol></li><li><h3>Module 5 · AXI4</h3><p>The five channels: address, data and response in both directions, lite or full, bursts and IDs.</p><ol><li><a href="/learn/five-channels/">13 Five channels</a></li><li><a href="/learn/lite-or-full/">14 Lite or full</a></li><li><a href="/learn/bursts-and-ids/">15 Bursts and IDs</a></li></ol></li><li><h3>Module 6 · Memory</h3><p>What sits on the far side of every bus: memory maps, port kinds, and latency that a wait state must cover.</p><ol><li><a href="/learn/memory-maps/">16 Memory maps</a></li><li><a href="/learn/ports-and-kinds/">17 Ports and kinds</a></li><li><a href="/learn/latency/">18 Latency</a></li></ol></li><li><h3>Module 7 · Bridges</h3><p>Why designs grow more than one bus, and the packet bridge that moves work between them.</p><ol><li><a href="/learn/why-bridges/">19 Why bridges</a></li><li><a href="/learn/the-packet-bridge/">20 The packet bridge</a></li><li><a href="/learn/everything-in-one-bench/">21 Everything in one bench</a></li></ol></li><li><h3>Module 8 · Ethernet</h3><p>Frames, the frame check sequence, RGMII timing, and the pre-registered steps of a real bring-up.</p><ol><li><a href="/learn/frames-and-the-fcs/">22 Frames and the FCS</a></li><li><a href="/learn/rgmii-timing/">23 RGMII timing</a></li><li class="on" aria-current="page"><a href="/learn/bring-up-on-the-bench/">24 Bring-up on the bench</a></li></ol></li><li><h3>Module 9 · The bench</h3><p>The discipline that guards real hardware: who holds the IO, taking and giving the lock, and what runs next.</p><ol><li><a href="/learn/who-holds-the-io/">25 Who holds the IO</a></li><li><a href="/learn/claim-and-release/">26 Claim and release</a></li><li><a href="/learn/capstone-what-runs-next/">27 Capstone: what runs next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/bugs-that-compile/index.html b/apps/website/public/learn/bugs-that-compile/index.html
index 5c8d5ec1da..e2489eaf62 100644
--- a/apps/website/public/learn/bugs-that-compile/index.html
+++ b/apps/website/public/learn/bugs-that-compile/index.html
@@ -466,5 +466,6 @@ <h1>Bugs that compile</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Why verify</h3><p>Designs that compile and are wrong, the model that decides, and the plan written before the code.</p><ol><li class="on" aria-current="page"><a href="/learn/bugs-that-compile/">01 Bugs that compile</a></li><li><a href="/learn/golden-model/">02 The golden model</a></li><li><a href="/learn/plan-before-code/">03 Plan before code</a></li></ol></li><li><h3>Module 2 · Testbenches</h3><p>Stimulus, checks and a verdict, written as one spec beside the design it judges.</p><ol><li><a href="/learn/anatomy-of-a-tb/">04 Anatomy of a testbench</a></li><li><a href="/learn/self-checking/">05 Self-checking, not staring</a></li><li><a href="/learn/directed-vs-random/">06 Directed versus random</a></li></ol></li><li><h3>Module 3 · Waveforms</h3><p>A trace of every signal, read the way a hardware engineer reads it, and two runs compared.</p><ol><li><a href="/learn/vcd-format/">07 The VCD format</a></li><li><a href="/learn/debug-with-waves/">08 Debug with waves</a></li><li><a href="/learn/compare-traces/">09 Compare two traces</a></li></ol></li><li><h3>Module 4 · Conformance vectors</h3><p>Cases with the answer written beside them, kept where the compiler can reach them.</p><ol><li><a href="/learn/what-a-vector-is/">10 What a vector is</a></li><li><a href="/learn/golden-ruler-vectors/">11 Golden ruler vectors</a></li><li><a href="/learn/end-to-end-scenarios/">12 End-to-end scenarios</a></li></ol></li><li><h3>Module 5 · Cosimulation</h3><p>Spec, simulator and board agreeing on the bench Artix-7 XC7A200T, and what to do when they do not.</p><ol><li><a href="/learn/three-way-agreement/">13 Three-way agreement</a></li><li><a href="/learn/simulator-internals/">14 Inside the simulator</a></li><li><a href="/learn/when-they-disagree/">15 When they disagree</a></li></ol></li><li><h3>Module 6 · Coverage</h3><p>What the tests touched: lines, toggles, states, and what that number hides.</p><ol><li><a href="/learn/line-and-toggle/">16 Line and toggle</a></li><li><a href="/learn/fsm-coverage/">17 FSM coverage</a></li><li><a href="/learn/coverage-lies/">18 Coverage lies</a></li></ol></li><li><h3>Module 7 · Formal</h3><p>Assertions that hold every cycle, bounded search for a counterexample, and why a proof needs induction.</p><ol><li><a href="/learn/assertions/">19 Assertions</a></li><li><a href="/learn/bounded-model-checking/">20 Bounded model checking</a></li><li><a href="/learn/induction/">21 Induction</a></li></ol></li><li><h3>Module 8 · Mutation</h3><p>Break the design on purpose and count what the tests catch.</p><ol><li><a href="/learn/kill-the-mutant/">22 Kill the mutant</a></li><li><a href="/learn/surviving-mutants/">23 Surviving mutants</a></li><li><a href="/learn/test-strength/">24 Test strength</a></li></ol></li><li><h3>Module 9 · Sign-off</h3><p>One command, every receipt, a clean verdict you can show.</p><ol><li><a href="/learn/build-verify/">25 Build and verify</a></li><li><a href="/learn/board-selftest/">26 The board self-tests</a></li><li><a href="/learn/verify-capstone/">27 Capstone: verify a module whole</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/build-verify/index.html b/apps/website/public/learn/build-verify/index.html
index b7c661af55..4d2337e333 100644
--- a/apps/website/public/learn/build-verify/index.html
+++ b/apps/website/public/learn/build-verify/index.html
@@ -193,5 +193,6 @@ <h1>Build and verify</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Why verify</h3><p>Designs that compile and are wrong, the model that decides, and the plan written before the code.</p><ol><li><a href="/learn/bugs-that-compile/">01 Bugs that compile</a></li><li><a href="/learn/golden-model/">02 The golden model</a></li><li><a href="/learn/plan-before-code/">03 Plan before code</a></li></ol></li><li><h3>Module 2 · Testbenches</h3><p>Stimulus, checks and a verdict, written as one spec beside the design it judges.</p><ol><li><a href="/learn/anatomy-of-a-tb/">04 Anatomy of a testbench</a></li><li><a href="/learn/self-checking/">05 Self-checking, not staring</a></li><li><a href="/learn/directed-vs-random/">06 Directed versus random</a></li></ol></li><li><h3>Module 3 · Waveforms</h3><p>A trace of every signal, read the way a hardware engineer reads it, and two runs compared.</p><ol><li><a href="/learn/vcd-format/">07 The VCD format</a></li><li><a href="/learn/debug-with-waves/">08 Debug with waves</a></li><li><a href="/learn/compare-traces/">09 Compare two traces</a></li></ol></li><li><h3>Module 4 · Conformance vectors</h3><p>Cases with the answer written beside them, kept where the compiler can reach them.</p><ol><li><a href="/learn/what-a-vector-is/">10 What a vector is</a></li><li><a href="/learn/golden-ruler-vectors/">11 Golden ruler vectors</a></li><li><a href="/learn/end-to-end-scenarios/">12 End-to-end scenarios</a></li></ol></li><li><h3>Module 5 · Cosimulation</h3><p>Spec, simulator and board agreeing on the bench Artix-7 XC7A200T, and what to do when they do not.</p><ol><li><a href="/learn/three-way-agreement/">13 Three-way agreement</a></li><li><a href="/learn/simulator-internals/">14 Inside the simulator</a></li><li><a href="/learn/when-they-disagree/">15 When they disagree</a></li></ol></li><li><h3>Module 6 · Coverage</h3><p>What the tests touched: lines, toggles, states, and what that number hides.</p><ol><li><a href="/learn/line-and-toggle/">16 Line and toggle</a></li><li><a href="/learn/fsm-coverage/">17 FSM coverage</a></li><li><a href="/learn/coverage-lies/">18 Coverage lies</a></li></ol></li><li><h3>Module 7 · Formal</h3><p>Assertions that hold every cycle, bounded search for a counterexample, and why a proof needs induction.</p><ol><li><a href="/learn/assertions/">19 Assertions</a></li><li><a href="/learn/bounded-model-checking/">20 Bounded model checking</a></li><li><a href="/learn/induction/">21 Induction</a></li></ol></li><li><h3>Module 8 · Mutation</h3><p>Break the design on purpose and count what the tests catch.</p><ol><li><a href="/learn/kill-the-mutant/">22 Kill the mutant</a></li><li><a href="/learn/surviving-mutants/">23 Surviving mutants</a></li><li><a href="/learn/test-strength/">24 Test strength</a></li></ol></li><li><h3>Module 9 · Sign-off</h3><p>One command, every receipt, a clean verdict you can show.</p><ol><li class="on" aria-current="page"><a href="/learn/build-verify/">25 Build and verify</a></li><li><a href="/learn/board-selftest/">26 The board self-tests</a></li><li><a href="/learn/verify-capstone/">27 Capstone: verify a module whole</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/bursts-and-ids/index.html b/apps/website/public/learn/bursts-and-ids/index.html
index 8798f18921..db3fc8f5c7 100644
--- a/apps/website/public/learn/bursts-and-ids/index.html
+++ b/apps/website/public/learn/bursts-and-ids/index.html
@@ -280,5 +280,6 @@ <h1>Bursts and IDs</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a bus is</h3><p>Why a bus exists at all: a conversation on wires, framed and addressed, and who is allowed to talk.</p><ol><li><a href="/learn/wires-that-talk/">01 Wires that talk</a></li><li><a href="/learn/framing-a-conversation/">02 Framing a conversation</a></li><li><a href="/learn/master-and-servant/">03 Master and servant</a></li></ol></li><li><h3>Module 2 · UART</h3><p>The two-wire bus with no clock: the frame, the divisor that sets the rate, and the status a driver polls.</p><ol><li><a href="/learn/the-frame/">04 The frame: start, data, stop</a></li><li><a href="/learn/baud-and-the-divisor/">05 Baud and the divisor</a></li><li><a href="/learn/status-and-fifos/">06 Status and FIFOs</a></li></ol></li><li><h3>Module 3 · SPI</h3><p>The clocked conversation: four modes, a prescaler ladder, and a chip select per servant.</p><ol><li><a href="/learn/modes-cpol-cpha/">07 Modes: CPOL and CPHA</a></li><li><a href="/learn/the-prescaler-ladder/">08 The prescaler ladder</a></li><li><a href="/learn/chip-select-and-widths/">09 Chip select and widths</a></li></ol></li><li><h3>Module 4 · APB</h3><p>The register bus: PSEL and PENABLE, strobes and wait states, and how many address bits a peripheral count costs.</p><ol><li><a href="/learn/psel-and-penable/">10 PSEL and PENABLE</a></li><li><a href="/learn/strobes-and-waits/">11 Strobes and waits</a></li><li><a href="/learn/address-decode/">12 Address decode</a></li></ol></li><li><h3>Module 5 · AXI4</h3><p>The five channels: address, data and response in both directions, lite or full, bursts and IDs.</p><ol><li><a href="/learn/five-channels/">13 Five channels</a></li><li><a href="/learn/lite-or-full/">14 Lite or full</a></li><li class="on" aria-current="page"><a href="/learn/bursts-and-ids/">15 Bursts and IDs</a></li></ol></li><li><h3>Module 6 · Memory</h3><p>What sits on the far side of every bus: memory maps, port kinds, and latency that a wait state must cover.</p><ol><li><a href="/learn/memory-maps/">16 Memory maps</a></li><li><a href="/learn/ports-and-kinds/">17 Ports and kinds</a></li><li><a href="/learn/latency/">18 Latency</a></li></ol></li><li><h3>Module 7 · Bridges</h3><p>Why designs grow more than one bus, and the packet bridge that moves work between them.</p><ol><li><a href="/learn/why-bridges/">19 Why bridges</a></li><li><a href="/learn/the-packet-bridge/">20 The packet bridge</a></li><li><a href="/learn/everything-in-one-bench/">21 Everything in one bench</a></li></ol></li><li><h3>Module 8 · Ethernet</h3><p>Frames, the frame check sequence, RGMII timing, and the pre-registered steps of a real bring-up.</p><ol><li><a href="/learn/frames-and-the-fcs/">22 Frames and the FCS</a></li><li><a href="/learn/rgmii-timing/">23 RGMII timing</a></li><li><a href="/learn/bring-up-on-the-bench/">24 Bring-up on the bench</a></li></ol></li><li><h3>Module 9 · The bench</h3><p>The discipline that guards real hardware: who holds the IO, taking and giving the lock, and what runs next.</p><ol><li><a href="/learn/who-holds-the-io/">25 Who holds the IO</a></li><li><a href="/learn/claim-and-release/">26 Claim and release</a></li><li><a href="/learn/capstone-what-runs-next/">27 Capstone: what runs next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/buses-and-peripherals/index.html b/apps/website/public/learn/buses-and-peripherals/index.html
index 00e6227227..bd54c5ba5f 100644
--- a/apps/website/public/learn/buses-and-peripherals/index.html
+++ b/apps/website/public/learn/buses-and-peripherals/index.html
@@ -67,5 +67,6 @@ <h1>Buses and peripherals with t27: UART, SPI, APB, AXI4, Ethernet</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a bus is</h3><p>Why a bus exists at all: a conversation on wires, framed and addressed, and who is allowed to talk.</p><ol><li><a href="/learn/wires-that-talk/">01 Wires that talk</a></li><li><a href="/learn/framing-a-conversation/">02 Framing a conversation</a></li><li><a href="/learn/master-and-servant/">03 Master and servant</a></li></ol></li><li><h3>Module 2 · UART</h3><p>The two-wire bus with no clock: the frame, the divisor that sets the rate, and the status a driver polls.</p><ol><li><a href="/learn/the-frame/">04 The frame: start, data, stop</a></li><li><a href="/learn/baud-and-the-divisor/">05 Baud and the divisor</a></li><li><a href="/learn/status-and-fifos/">06 Status and FIFOs</a></li></ol></li><li><h3>Module 3 · SPI</h3><p>The clocked conversation: four modes, a prescaler ladder, and a chip select per servant.</p><ol><li><a href="/learn/modes-cpol-cpha/">07 Modes: CPOL and CPHA</a></li><li><a href="/learn/the-prescaler-ladder/">08 The prescaler ladder</a></li><li><a href="/learn/chip-select-and-widths/">09 Chip select and widths</a></li></ol></li><li><h3>Module 4 · APB</h3><p>The register bus: PSEL and PENABLE, strobes and wait states, and how many address bits a peripheral count costs.</p><ol><li><a href="/learn/psel-and-penable/">10 PSEL and PENABLE</a></li><li><a href="/learn/strobes-and-waits/">11 Strobes and waits</a></li><li><a href="/learn/address-decode/">12 Address decode</a></li></ol></li><li><h3>Module 5 · AXI4</h3><p>The five channels: address, data and response in both directions, lite or full, bursts and IDs.</p><ol><li><a href="/learn/five-channels/">13 Five channels</a></li><li><a href="/learn/lite-or-full/">14 Lite or full</a></li><li><a href="/learn/bursts-and-ids/">15 Bursts and IDs</a></li></ol></li><li><h3>Module 6 · Memory</h3><p>What sits on the far side of every bus: memory maps, port kinds, and latency that a wait state must cover.</p><ol><li><a href="/learn/memory-maps/">16 Memory maps</a></li><li><a href="/learn/ports-and-kinds/">17 Ports and kinds</a></li><li><a href="/learn/latency/">18 Latency</a></li></ol></li><li><h3>Module 7 · Bridges</h3><p>Why designs grow more than one bus, and the packet bridge that moves work between them.</p><ol><li><a href="/learn/why-bridges/">19 Why bridges</a></li><li><a href="/learn/the-packet-bridge/">20 The packet bridge</a></li><li><a href="/learn/everything-in-one-bench/">21 Everything in one bench</a></li></ol></li><li><h3>Module 8 · Ethernet</h3><p>Frames, the frame check sequence, RGMII timing, and the pre-registered steps of a real bring-up.</p><ol><li><a href="/learn/frames-and-the-fcs/">22 Frames and the FCS</a></li><li><a href="/learn/rgmii-timing/">23 RGMII timing</a></li><li><a href="/learn/bring-up-on-the-bench/">24 Bring-up on the bench</a></li></ol></li><li><h3>Module 9 · The bench</h3><p>The discipline that guards real hardware: who holds the IO, taking and giving the lock, and what runs next.</p><ol><li><a href="/learn/who-holds-the-io/">25 Who holds the IO</a></li><li><a href="/learn/claim-and-release/">26 Claim and release</a></li><li><a href="/learn/capstone-what-runs-next/">27 Capstone: what runs next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/capstone-a-filter-on-the-board/index.html b/apps/website/public/learn/capstone-a-filter-on-the-board/index.html
index bf09b06e2e..f12368fd15 100644
--- a/apps/website/public/learn/capstone-a-filter-on-the-board/index.html
+++ b/apps/website/public/learn/capstone-a-filter-on-the-board/index.html
@@ -1122,5 +1122,6 @@ <h1>Capstone: a filter on the board</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Adding numbers</h3><p>Why addition is slow: the carry that walks up the word, the dedicated chain in every slice, prefix networks and carry-save trees.</p><ol><li><a href="/learn/the-carry-chain/">01 The carry chain</a></li><li><a href="/learn/prefix-adders/">02 Looking ahead: prefix adders</a></li><li><a href="/learn/carry-save/">03 Three numbers at once</a></li></ol></li><li><h3>Module 2 · Multiplying</h3><p>A product is a sum of shifted copies; Booth recoding halves them, and the DSP48E1 slice does the rest in one block.</p><ol><li><a href="/learn/shift-and-add/">04 Shift and add</a></li><li><a href="/learn/booths-trick/">05 Booth's trick</a></li><li><a href="/learn/the-dsp48e1-slice/">06 The DSP48E1 slice</a></li></ol></li><li><h3>Module 3 · Fixed point</h3><p>Where the binary point sits, what rounding does to a value and to its average, and what each bit of a quantizer buys.</p><ol><li><a href="/learn/q-formats/">07 Q formats</a></li><li><a href="/learn/rounding-and-saturation/">08 Rounding and saturation</a></li><li><a href="/learn/quantization-noise/">09 Quantization noise</a></li></ol></li><li><h3>Module 4 · Functions in hardware</h3><p>Sine, cosine, angle and length from shifts and adds, and when a table is the better answer.</p><ol><li><a href="/learn/rotating-by-shifts/">10 Rotating by shifts</a></li><li><a href="/learn/angle-and-magnitude/">11 Angle and magnitude</a></li><li><a href="/learn/tables-and-interpolation/">12 Tables and interpolation</a></li></ol></li><li><h3>Module 5 · Signals and sampling</h3><p>What sampling does to a frequency, how an oscillator is built from an adder, and what a DFT bin measures.</p><ol><li><a href="/learn/sampling-and-aliasing/">13 Sampling and aliasing</a></li><li><a href="/learn/the-nco/">14 A numerically controlled oscillator</a></li><li><a href="/learn/bins-of-a-spectrum/">15 Bins of a spectrum</a></li></ol></li><li><h3>Module 6 · FIR filters</h3><p>The moving average, a windowed-sinc design with integer taps, and folding the taps onto DSP slices.</p><ol><li><a href="/learn/the-moving-average/">16 The moving average</a></li><li><a href="/learn/designing-taps/">17 Designing taps</a></li><li><a href="/learn/folding-taps-onto-slices/">18 Folding taps onto slices</a></li></ol></li><li><h3>Module 7 · Multirate</h3><p>Lowering the sample rate without folding noise in: decimation, the CIC filter and the polyphase form.</p><ol><li><a href="/learn/lowering-the-rate/">19 Lowering the rate</a></li><li><a href="/learn/the-cic-filter/">20 The CIC filter</a></li><li><a href="/learn/polyphase/">21 Polyphase</a></li></ol></li><li><h3>Module 8 · The FFT</h3><p>N log N instead of N^2: butterflies, the bit-reversed input order and the bits each stage adds.</p><ol><li><a href="/learn/the-butterfly/">22 The butterfly</a></li><li><a href="/learn/twiddles-and-bit-reversal/">23 Twiddles and bit reversal</a></li><li><a href="/learn/growth-through-the-stages/">24 Growth through the stages</a></li></ol></li><li><h3>Module 9 · On the bench</h3><p>From the arithmetic to the board: a correlator from the modem spec, a budget of slices and timing, and the capstone filter.</p><ol><li><a href="/learn/a-correlator-finds-the-frame/">25 A correlator finds the frame</a></li><li><a href="/learn/does-it-fit-and-close-timing/">26 Does it fit and close timing?</a></li><li class="on" aria-current="page"><a href="/learn/capstone-a-filter-on-the-board/">27 Capstone: a filter on the board</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/capstone-diff/index.html b/apps/website/public/learn/capstone-diff/index.html
index f90bcec437..ac96a7049d 100644
--- a/apps/website/public/learn/capstone-diff/index.html
+++ b/apps/website/public/learn/capstone-diff/index.html
@@ -434,5 +434,6 @@ <h1>Capstone: change it and diff it</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a clock is</h3><p>One edge, one world: what shares a clock edge shares a world, the period and the jitter of a real edge, and where the clock enters a board.</p><ol><li><a href="/learn/one-edge-one-world/">01 One edge, one world</a></li><li><a href="/learn/period-and-jitter/">02 A period, an edge, and jitter</a></li><li><a href="/learn/where-the-clock-enters/">03 Where the clock enters</a></li></ol></li><li><h3>Module 2 · Clock trees</h3><p>Skew and insertion delay, the global buffer network, and the trap of gating a clock with logic.</p><ol><li><a href="/learn/skew-and-insertion/">04 Skew and insertion delay</a></li><li><a href="/learn/global-buffers/">05 The global buffer</a></li><li><a href="/learn/gated-clock-trap/">06 The gated-clock trap</a></li></ol></li><li><h3>Module 3 · PLL and MMCM</h3><p>Multiply and divide one clock into another, move its phase in steps of the VCO, and which clocks the analyzer treats as related.</p><ol><li><a href="/learn/multiply-and-divide/">07 Multiply and divide</a></li><li><a href="/learn/phase-and-lock/">08 Phase and lock</a></li><li><a href="/learn/related-or-not/">09 Related or not</a></li></ol></li><li><h3>Module 4 · Resets</h3><p>Assert asynchronously, release synchronously: the three reset kinds, the release pipe, and the tree a reset grows.</p><ol><li><a href="/learn/sync-or-async/">10 Sync or async</a></li><li><a href="/learn/reset-release/">11 The release pipe</a></li><li><a href="/learn/reset-fanout/">12 Reset fanout</a></li></ol></li><li><h3>Module 5 · Metastability</h3><p>The setup-hold window, the mean time between failures in integer arithmetic, and the two flops that fix it.</p><ol><li><a href="/learn/setup-hold-window/">13 The setup-hold window</a></li><li><a href="/learn/mtbf/">14 Mean time between failures</a></li><li><a href="/learn/two-flop-sync/">15 The two-flop synchronizer</a></li></ol></li><li><h3>Module 6 · Crossing many bits</h3><p>Why a binary bus tears, why Gray code does not, and the handshake that moves a pulse between worlds.</p><ol><li><a href="/learn/why-binary-fails/">16 Why binary fails</a></li><li><a href="/learn/gray-code/">17 Gray code</a></li><li><a href="/learn/handshakes-and-pulses/">18 Handshakes and pulses</a></li></ol></li><li><h3>Module 7 · The asynchronous FIFO</h3><p>Pointers, flags and depth: the buffer that moves a stream between two clocks.</p><ol><li><a href="/learn/pointers-and-flags/">19 Pointers and flags</a></li><li><a href="/learn/depth-sizing/">20 Sizing the depth</a></li><li><a href="/learn/fifo-testbench/">21 The FIFO testbench</a></li></ol></li><li><h3>Module 8 · Constraints</h3><p>The lines that tell the analyzer what a clock is, which paths not to check, and what the pins must meet.</p><ol><li><a href="/learn/create-clock/">22 create_clock</a></li><li><a href="/learn/false-path-and-max-delay/">23 False paths and max delay</a></li><li><a href="/learn/io-timing/">24 IO timing</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>A CDC report, one crossing captured at the flip-flops, and the bitstream diff that closes the course.</p><ol><li><a href="/learn/cdc-report/">25 The CDC report</a></li><li><a href="/learn/capture-a-crossing/">26 Capture a crossing</a></li><li class="on" aria-current="page"><a href="/learn/capstone-diff/">27 Capstone: change it and diff it</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/capstone-what-runs-next/index.html b/apps/website/public/learn/capstone-what-runs-next/index.html
index 171c28847c..116ab93c9f 100644
--- a/apps/website/public/learn/capstone-what-runs-next/index.html
+++ b/apps/website/public/learn/capstone-what-runs-next/index.html
@@ -300,5 +300,6 @@ <h1>Capstone: what runs next</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a bus is</h3><p>Why a bus exists at all: a conversation on wires, framed and addressed, and who is allowed to talk.</p><ol><li><a href="/learn/wires-that-talk/">01 Wires that talk</a></li><li><a href="/learn/framing-a-conversation/">02 Framing a conversation</a></li><li><a href="/learn/master-and-servant/">03 Master and servant</a></li></ol></li><li><h3>Module 2 · UART</h3><p>The two-wire bus with no clock: the frame, the divisor that sets the rate, and the status a driver polls.</p><ol><li><a href="/learn/the-frame/">04 The frame: start, data, stop</a></li><li><a href="/learn/baud-and-the-divisor/">05 Baud and the divisor</a></li><li><a href="/learn/status-and-fifos/">06 Status and FIFOs</a></li></ol></li><li><h3>Module 3 · SPI</h3><p>The clocked conversation: four modes, a prescaler ladder, and a chip select per servant.</p><ol><li><a href="/learn/modes-cpol-cpha/">07 Modes: CPOL and CPHA</a></li><li><a href="/learn/the-prescaler-ladder/">08 The prescaler ladder</a></li><li><a href="/learn/chip-select-and-widths/">09 Chip select and widths</a></li></ol></li><li><h3>Module 4 · APB</h3><p>The register bus: PSEL and PENABLE, strobes and wait states, and how many address bits a peripheral count costs.</p><ol><li><a href="/learn/psel-and-penable/">10 PSEL and PENABLE</a></li><li><a href="/learn/strobes-and-waits/">11 Strobes and waits</a></li><li><a href="/learn/address-decode/">12 Address decode</a></li></ol></li><li><h3>Module 5 · AXI4</h3><p>The five channels: address, data and response in both directions, lite or full, bursts and IDs.</p><ol><li><a href="/learn/five-channels/">13 Five channels</a></li><li><a href="/learn/lite-or-full/">14 Lite or full</a></li><li><a href="/learn/bursts-and-ids/">15 Bursts and IDs</a></li></ol></li><li><h3>Module 6 · Memory</h3><p>What sits on the far side of every bus: memory maps, port kinds, and latency that a wait state must cover.</p><ol><li><a href="/learn/memory-maps/">16 Memory maps</a></li><li><a href="/learn/ports-and-kinds/">17 Ports and kinds</a></li><li><a href="/learn/latency/">18 Latency</a></li></ol></li><li><h3>Module 7 · Bridges</h3><p>Why designs grow more than one bus, and the packet bridge that moves work between them.</p><ol><li><a href="/learn/why-bridges/">19 Why bridges</a></li><li><a href="/learn/the-packet-bridge/">20 The packet bridge</a></li><li><a href="/learn/everything-in-one-bench/">21 Everything in one bench</a></li></ol></li><li><h3>Module 8 · Ethernet</h3><p>Frames, the frame check sequence, RGMII timing, and the pre-registered steps of a real bring-up.</p><ol><li><a href="/learn/frames-and-the-fcs/">22 Frames and the FCS</a></li><li><a href="/learn/rgmii-timing/">23 RGMII timing</a></li><li><a href="/learn/bring-up-on-the-bench/">24 Bring-up on the bench</a></li></ol></li><li><h3>Module 9 · The bench</h3><p>The discipline that guards real hardware: who holds the IO, taking and giving the lock, and what runs next.</p><ol><li><a href="/learn/who-holds-the-io/">25 Who holds the IO</a></li><li><a href="/learn/claim-and-release/">26 Claim and release</a></li><li class="on" aria-current="page"><a href="/learn/capstone-what-runs-next/">27 Capstone: what runs next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/capstone/index.html b/apps/website/public/learn/capstone/index.html
index e6921553f6..17ee528cd2 100644
--- a/apps/website/public/learn/capstone/index.html
+++ b/apps/website/public/learn/capstone/index.html
@@ -322,5 +322,6 @@ <h1>Capstone: out-guess yosys</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The chip</h3><p>What an FPGA is, which chip we use, and how its pins meet the board.</p><ol><li><a href="/learn/fpga-is-a-grid/">01 An FPGA is a grid you configure</a></li><li><a href="/learn/which-chip/">02 Which chip is this?</a></li><li><a href="/learn/pins-and-package/">03 Pins, balls and the package</a></li></ol></li><li><h3>Module 2 · Numbers in hardware</h3><p>Bits, trits and number formats, and what arithmetic costs in logic.</p><ol><li><a href="/learn/bits-and-trits/">04 Bits, trits and GF16</a></li><li><a href="/learn/what-arithmetic-costs/">05 What arithmetic costs</a></li><li><a href="/learn/how-much-fits/">06 How much fits on the chip</a></li></ol></li><li><h3>Module 3 · Your t27 program</h3><p>Write a spec, test it, and see why compiling is not the same as being right.</p><ol><li><a href="/learn/hello-t27/">07 Hello, t27</a></li><li><a href="/learn/tests-are-the-spec/">08 Tests are part of the spec</a></li><li><a href="/learn/compiles-is-not-correct/">09 Compiles is not the same as correct</a></li></ol></li><li><h3>Module 4 · Inside t27c</h3><p>How the compiler reads a spec and what it writes, including the native t27b.</p><ol><li><a href="/learn/compiler-stages/">10 The stages of t27c</a></li><li><a href="/learn/t27c-on-a-real-machine/">11 t27c on a real machine</a></li><li><a href="/learn/seven-backends-and-t27b/">12 Seven backends, and t27b</a></li></ol></li><li><h3>Module 5 · From spec to hardware</h3><p>The Verilog t27c writes, the cells it becomes, and what one LUT does.</p><ol><li><a href="/learn/spec-to-verilog/">13 From a spec to Verilog</a></li><li><a href="/learn/count-the-cells/">14 Count the cells</a></li><li><a href="/learn/one-lut-up-close/">15 One LUT, up close</a></li></ol></li><li><h3>Module 6 · Reading synthesis</h3><p>What yosys reports about your design, and which warnings matter.</p><ol><li><a href="/learn/synthesis-warnings/">16 Synthesis warnings that matter</a></li><li><a href="/learn/state-machines/">17 State machines</a></li><li><a href="/learn/where-the-luts-go/">18 Where the LUTs go</a></li></ol></li><li><h3>Module 7 · Place, route, timing</h3><p>Where the cells land on the die, and whether the clock is met.</p><ol><li><a href="/learn/placement/">19 Placement: where the cells land</a></li><li><a href="/learn/logic-depth/">20 Logic depth</a></li><li><a href="/learn/timing-slack/">21 Timing and slack</a></li></ol></li><li><h3>Module 8 · The bitstream</h3><p>How a routed design becomes the bits the chip loads, with no Vivado.</p><ol><li><a href="/learn/fasm-to-frames/">22 From FASM to frames</a></li><li><a href="/learn/a-build-receipt/">23 A build, itemised</a></li><li><a href="/learn/read-a-bitstream/">24 Read a bitstream</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>Ask the chip who it is, load the bits, and check them.</p><ol><li><a href="/learn/jtag-asks-the-chip/">25 JTAG asks the chip</a></li><li><a href="/learn/same-bits-on-the-board/">26 The same bits, on the board</a></li><li class="on" aria-current="page"><a href="/learn/capstone/">27 Capstone: out-guess yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/capture-a-crossing/index.html b/apps/website/public/learn/capture-a-crossing/index.html
index 7cc16865cf..61d9bb8b4d 100644
--- a/apps/website/public/learn/capture-a-crossing/index.html
+++ b/apps/website/public/learn/capture-a-crossing/index.html
@@ -110,5 +110,6 @@ <h1>Capture a crossing</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a clock is</h3><p>One edge, one world: what shares a clock edge shares a world, the period and the jitter of a real edge, and where the clock enters a board.</p><ol><li><a href="/learn/one-edge-one-world/">01 One edge, one world</a></li><li><a href="/learn/period-and-jitter/">02 A period, an edge, and jitter</a></li><li><a href="/learn/where-the-clock-enters/">03 Where the clock enters</a></li></ol></li><li><h3>Module 2 · Clock trees</h3><p>Skew and insertion delay, the global buffer network, and the trap of gating a clock with logic.</p><ol><li><a href="/learn/skew-and-insertion/">04 Skew and insertion delay</a></li><li><a href="/learn/global-buffers/">05 The global buffer</a></li><li><a href="/learn/gated-clock-trap/">06 The gated-clock trap</a></li></ol></li><li><h3>Module 3 · PLL and MMCM</h3><p>Multiply and divide one clock into another, move its phase in steps of the VCO, and which clocks the analyzer treats as related.</p><ol><li><a href="/learn/multiply-and-divide/">07 Multiply and divide</a></li><li><a href="/learn/phase-and-lock/">08 Phase and lock</a></li><li><a href="/learn/related-or-not/">09 Related or not</a></li></ol></li><li><h3>Module 4 · Resets</h3><p>Assert asynchronously, release synchronously: the three reset kinds, the release pipe, and the tree a reset grows.</p><ol><li><a href="/learn/sync-or-async/">10 Sync or async</a></li><li><a href="/learn/reset-release/">11 The release pipe</a></li><li><a href="/learn/reset-fanout/">12 Reset fanout</a></li></ol></li><li><h3>Module 5 · Metastability</h3><p>The setup-hold window, the mean time between failures in integer arithmetic, and the two flops that fix it.</p><ol><li><a href="/learn/setup-hold-window/">13 The setup-hold window</a></li><li><a href="/learn/mtbf/">14 Mean time between failures</a></li><li><a href="/learn/two-flop-sync/">15 The two-flop synchronizer</a></li></ol></li><li><h3>Module 6 · Crossing many bits</h3><p>Why a binary bus tears, why Gray code does not, and the handshake that moves a pulse between worlds.</p><ol><li><a href="/learn/why-binary-fails/">16 Why binary fails</a></li><li><a href="/learn/gray-code/">17 Gray code</a></li><li><a href="/learn/handshakes-and-pulses/">18 Handshakes and pulses</a></li></ol></li><li><h3>Module 7 · The asynchronous FIFO</h3><p>Pointers, flags and depth: the buffer that moves a stream between two clocks.</p><ol><li><a href="/learn/pointers-and-flags/">19 Pointers and flags</a></li><li><a href="/learn/depth-sizing/">20 Sizing the depth</a></li><li><a href="/learn/fifo-testbench/">21 The FIFO testbench</a></li></ol></li><li><h3>Module 8 · Constraints</h3><p>The lines that tell the analyzer what a clock is, which paths not to check, and what the pins must meet.</p><ol><li><a href="/learn/create-clock/">22 create_clock</a></li><li><a href="/learn/false-path-and-max-delay/">23 False paths and max delay</a></li><li><a href="/learn/io-timing/">24 IO timing</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>A CDC report, one crossing captured at the flip-flops, and the bitstream diff that closes the course.</p><ol><li><a href="/learn/cdc-report/">25 The CDC report</a></li><li class="on" aria-current="page"><a href="/learn/capture-a-crossing/">26 Capture a crossing</a></li><li><a href="/learn/capstone-diff/">27 Capstone: change it and diff it</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/carry-save/index.html b/apps/website/public/learn/carry-save/index.html
index 0ecea716dd..54cef45607 100644
--- a/apps/website/public/learn/carry-save/index.html
+++ b/apps/website/public/learn/carry-save/index.html
@@ -554,5 +554,6 @@ <h1>Three numbers at once</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Adding numbers</h3><p>Why addition is slow: the carry that walks up the word, the dedicated chain in every slice, prefix networks and carry-save trees.</p><ol><li><a href="/learn/the-carry-chain/">01 The carry chain</a></li><li><a href="/learn/prefix-adders/">02 Looking ahead: prefix adders</a></li><li class="on" aria-current="page"><a href="/learn/carry-save/">03 Three numbers at once</a></li></ol></li><li><h3>Module 2 · Multiplying</h3><p>A product is a sum of shifted copies; Booth recoding halves them, and the DSP48E1 slice does the rest in one block.</p><ol><li><a href="/learn/shift-and-add/">04 Shift and add</a></li><li><a href="/learn/booths-trick/">05 Booth's trick</a></li><li><a href="/learn/the-dsp48e1-slice/">06 The DSP48E1 slice</a></li></ol></li><li><h3>Module 3 · Fixed point</h3><p>Where the binary point sits, what rounding does to a value and to its average, and what each bit of a quantizer buys.</p><ol><li><a href="/learn/q-formats/">07 Q formats</a></li><li><a href="/learn/rounding-and-saturation/">08 Rounding and saturation</a></li><li><a href="/learn/quantization-noise/">09 Quantization noise</a></li></ol></li><li><h3>Module 4 · Functions in hardware</h3><p>Sine, cosine, angle and length from shifts and adds, and when a table is the better answer.</p><ol><li><a href="/learn/rotating-by-shifts/">10 Rotating by shifts</a></li><li><a href="/learn/angle-and-magnitude/">11 Angle and magnitude</a></li><li><a href="/learn/tables-and-interpolation/">12 Tables and interpolation</a></li></ol></li><li><h3>Module 5 · Signals and sampling</h3><p>What sampling does to a frequency, how an oscillator is built from an adder, and what a DFT bin measures.</p><ol><li><a href="/learn/sampling-and-aliasing/">13 Sampling and aliasing</a></li><li><a href="/learn/the-nco/">14 A numerically controlled oscillator</a></li><li><a href="/learn/bins-of-a-spectrum/">15 Bins of a spectrum</a></li></ol></li><li><h3>Module 6 · FIR filters</h3><p>The moving average, a windowed-sinc design with integer taps, and folding the taps onto DSP slices.</p><ol><li><a href="/learn/the-moving-average/">16 The moving average</a></li><li><a href="/learn/designing-taps/">17 Designing taps</a></li><li><a href="/learn/folding-taps-onto-slices/">18 Folding taps onto slices</a></li></ol></li><li><h3>Module 7 · Multirate</h3><p>Lowering the sample rate without folding noise in: decimation, the CIC filter and the polyphase form.</p><ol><li><a href="/learn/lowering-the-rate/">19 Lowering the rate</a></li><li><a href="/learn/the-cic-filter/">20 The CIC filter</a></li><li><a href="/learn/polyphase/">21 Polyphase</a></li></ol></li><li><h3>Module 8 · The FFT</h3><p>N log N instead of N^2: butterflies, the bit-reversed input order and the bits each stage adds.</p><ol><li><a href="/learn/the-butterfly/">22 The butterfly</a></li><li><a href="/learn/twiddles-and-bit-reversal/">23 Twiddles and bit reversal</a></li><li><a href="/learn/growth-through-the-stages/">24 Growth through the stages</a></li></ol></li><li><h3>Module 9 · On the bench</h3><p>From the arithmetic to the board: a correlator from the modem spec, a budget of slices and timing, and the capstone filter.</p><ol><li><a href="/learn/a-correlator-finds-the-frame/">25 A correlator finds the frame</a></li><li><a href="/learn/does-it-fit-and-close-timing/">26 Does it fit and close timing?</a></li><li><a href="/learn/capstone-a-filter-on-the-board/">27 Capstone: a filter on the board</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/cdc-report/index.html b/apps/website/public/learn/cdc-report/index.html
index 57bea0d079..231f68fd07 100644
--- a/apps/website/public/learn/cdc-report/index.html
+++ b/apps/website/public/learn/cdc-report/index.html
@@ -448,5 +448,6 @@ <h1>The CDC report</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a clock is</h3><p>One edge, one world: what shares a clock edge shares a world, the period and the jitter of a real edge, and where the clock enters a board.</p><ol><li><a href="/learn/one-edge-one-world/">01 One edge, one world</a></li><li><a href="/learn/period-and-jitter/">02 A period, an edge, and jitter</a></li><li><a href="/learn/where-the-clock-enters/">03 Where the clock enters</a></li></ol></li><li><h3>Module 2 · Clock trees</h3><p>Skew and insertion delay, the global buffer network, and the trap of gating a clock with logic.</p><ol><li><a href="/learn/skew-and-insertion/">04 Skew and insertion delay</a></li><li><a href="/learn/global-buffers/">05 The global buffer</a></li><li><a href="/learn/gated-clock-trap/">06 The gated-clock trap</a></li></ol></li><li><h3>Module 3 · PLL and MMCM</h3><p>Multiply and divide one clock into another, move its phase in steps of the VCO, and which clocks the analyzer treats as related.</p><ol><li><a href="/learn/multiply-and-divide/">07 Multiply and divide</a></li><li><a href="/learn/phase-and-lock/">08 Phase and lock</a></li><li><a href="/learn/related-or-not/">09 Related or not</a></li></ol></li><li><h3>Module 4 · Resets</h3><p>Assert asynchronously, release synchronously: the three reset kinds, the release pipe, and the tree a reset grows.</p><ol><li><a href="/learn/sync-or-async/">10 Sync or async</a></li><li><a href="/learn/reset-release/">11 The release pipe</a></li><li><a href="/learn/reset-fanout/">12 Reset fanout</a></li></ol></li><li><h3>Module 5 · Metastability</h3><p>The setup-hold window, the mean time between failures in integer arithmetic, and the two flops that fix it.</p><ol><li><a href="/learn/setup-hold-window/">13 The setup-hold window</a></li><li><a href="/learn/mtbf/">14 Mean time between failures</a></li><li><a href="/learn/two-flop-sync/">15 The two-flop synchronizer</a></li></ol></li><li><h3>Module 6 · Crossing many bits</h3><p>Why a binary bus tears, why Gray code does not, and the handshake that moves a pulse between worlds.</p><ol><li><a href="/learn/why-binary-fails/">16 Why binary fails</a></li><li><a href="/learn/gray-code/">17 Gray code</a></li><li><a href="/learn/handshakes-and-pulses/">18 Handshakes and pulses</a></li></ol></li><li><h3>Module 7 · The asynchronous FIFO</h3><p>Pointers, flags and depth: the buffer that moves a stream between two clocks.</p><ol><li><a href="/learn/pointers-and-flags/">19 Pointers and flags</a></li><li><a href="/learn/depth-sizing/">20 Sizing the depth</a></li><li><a href="/learn/fifo-testbench/">21 The FIFO testbench</a></li></ol></li><li><h3>Module 8 · Constraints</h3><p>The lines that tell the analyzer what a clock is, which paths not to check, and what the pins must meet.</p><ol><li><a href="/learn/create-clock/">22 create_clock</a></li><li><a href="/learn/false-path-and-max-delay/">23 False paths and max delay</a></li><li><a href="/learn/io-timing/">24 IO timing</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>A CDC report, one crossing captured at the flip-flops, and the bitstream diff that closes the course.</p><ol><li class="on" aria-current="page"><a href="/learn/cdc-report/">25 The CDC report</a></li><li><a href="/learn/capture-a-crossing/">26 Capture a crossing</a></li><li><a href="/learn/capstone-diff/">27 Capstone: change it and diff it</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/chip-select-and-widths/index.html b/apps/website/public/learn/chip-select-and-widths/index.html
index 3e71cf85ae..628f6640a0 100644
--- a/apps/website/public/learn/chip-select-and-widths/index.html
+++ b/apps/website/public/learn/chip-select-and-widths/index.html
@@ -209,5 +209,6 @@ <h1>Chip select and widths</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a bus is</h3><p>Why a bus exists at all: a conversation on wires, framed and addressed, and who is allowed to talk.</p><ol><li><a href="/learn/wires-that-talk/">01 Wires that talk</a></li><li><a href="/learn/framing-a-conversation/">02 Framing a conversation</a></li><li><a href="/learn/master-and-servant/">03 Master and servant</a></li></ol></li><li><h3>Module 2 · UART</h3><p>The two-wire bus with no clock: the frame, the divisor that sets the rate, and the status a driver polls.</p><ol><li><a href="/learn/the-frame/">04 The frame: start, data, stop</a></li><li><a href="/learn/baud-and-the-divisor/">05 Baud and the divisor</a></li><li><a href="/learn/status-and-fifos/">06 Status and FIFOs</a></li></ol></li><li><h3>Module 3 · SPI</h3><p>The clocked conversation: four modes, a prescaler ladder, and a chip select per servant.</p><ol><li><a href="/learn/modes-cpol-cpha/">07 Modes: CPOL and CPHA</a></li><li><a href="/learn/the-prescaler-ladder/">08 The prescaler ladder</a></li><li class="on" aria-current="page"><a href="/learn/chip-select-and-widths/">09 Chip select and widths</a></li></ol></li><li><h3>Module 4 · APB</h3><p>The register bus: PSEL and PENABLE, strobes and wait states, and how many address bits a peripheral count costs.</p><ol><li><a href="/learn/psel-and-penable/">10 PSEL and PENABLE</a></li><li><a href="/learn/strobes-and-waits/">11 Strobes and waits</a></li><li><a href="/learn/address-decode/">12 Address decode</a></li></ol></li><li><h3>Module 5 · AXI4</h3><p>The five channels: address, data and response in both directions, lite or full, bursts and IDs.</p><ol><li><a href="/learn/five-channels/">13 Five channels</a></li><li><a href="/learn/lite-or-full/">14 Lite or full</a></li><li><a href="/learn/bursts-and-ids/">15 Bursts and IDs</a></li></ol></li><li><h3>Module 6 · Memory</h3><p>What sits on the far side of every bus: memory maps, port kinds, and latency that a wait state must cover.</p><ol><li><a href="/learn/memory-maps/">16 Memory maps</a></li><li><a href="/learn/ports-and-kinds/">17 Ports and kinds</a></li><li><a href="/learn/latency/">18 Latency</a></li></ol></li><li><h3>Module 7 · Bridges</h3><p>Why designs grow more than one bus, and the packet bridge that moves work between them.</p><ol><li><a href="/learn/why-bridges/">19 Why bridges</a></li><li><a href="/learn/the-packet-bridge/">20 The packet bridge</a></li><li><a href="/learn/everything-in-one-bench/">21 Everything in one bench</a></li></ol></li><li><h3>Module 8 · Ethernet</h3><p>Frames, the frame check sequence, RGMII timing, and the pre-registered steps of a real bring-up.</p><ol><li><a href="/learn/frames-and-the-fcs/">22 Frames and the FCS</a></li><li><a href="/learn/rgmii-timing/">23 RGMII timing</a></li><li><a href="/learn/bring-up-on-the-bench/">24 Bring-up on the bench</a></li></ol></li><li><h3>Module 9 · The bench</h3><p>The discipline that guards real hardware: who holds the IO, taking and giving the lock, and what runs next.</p><ol><li><a href="/learn/who-holds-the-io/">25 Who holds the IO</a></li><li><a href="/learn/claim-and-release/">26 Claim and release</a></li><li><a href="/learn/capstone-what-runs-next/">27 Capstone: what runs next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/claim-and-release/index.html b/apps/website/public/learn/claim-and-release/index.html
index 583b8c064b..7112b2f48c 100644
--- a/apps/website/public/learn/claim-and-release/index.html
+++ b/apps/website/public/learn/claim-and-release/index.html
@@ -109,5 +109,6 @@ <h1>Claim and release</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a bus is</h3><p>Why a bus exists at all: a conversation on wires, framed and addressed, and who is allowed to talk.</p><ol><li><a href="/learn/wires-that-talk/">01 Wires that talk</a></li><li><a href="/learn/framing-a-conversation/">02 Framing a conversation</a></li><li><a href="/learn/master-and-servant/">03 Master and servant</a></li></ol></li><li><h3>Module 2 · UART</h3><p>The two-wire bus with no clock: the frame, the divisor that sets the rate, and the status a driver polls.</p><ol><li><a href="/learn/the-frame/">04 The frame: start, data, stop</a></li><li><a href="/learn/baud-and-the-divisor/">05 Baud and the divisor</a></li><li><a href="/learn/status-and-fifos/">06 Status and FIFOs</a></li></ol></li><li><h3>Module 3 · SPI</h3><p>The clocked conversation: four modes, a prescaler ladder, and a chip select per servant.</p><ol><li><a href="/learn/modes-cpol-cpha/">07 Modes: CPOL and CPHA</a></li><li><a href="/learn/the-prescaler-ladder/">08 The prescaler ladder</a></li><li><a href="/learn/chip-select-and-widths/">09 Chip select and widths</a></li></ol></li><li><h3>Module 4 · APB</h3><p>The register bus: PSEL and PENABLE, strobes and wait states, and how many address bits a peripheral count costs.</p><ol><li><a href="/learn/psel-and-penable/">10 PSEL and PENABLE</a></li><li><a href="/learn/strobes-and-waits/">11 Strobes and waits</a></li><li><a href="/learn/address-decode/">12 Address decode</a></li></ol></li><li><h3>Module 5 · AXI4</h3><p>The five channels: address, data and response in both directions, lite or full, bursts and IDs.</p><ol><li><a href="/learn/five-channels/">13 Five channels</a></li><li><a href="/learn/lite-or-full/">14 Lite or full</a></li><li><a href="/learn/bursts-and-ids/">15 Bursts and IDs</a></li></ol></li><li><h3>Module 6 · Memory</h3><p>What sits on the far side of every bus: memory maps, port kinds, and latency that a wait state must cover.</p><ol><li><a href="/learn/memory-maps/">16 Memory maps</a></li><li><a href="/learn/ports-and-kinds/">17 Ports and kinds</a></li><li><a href="/learn/latency/">18 Latency</a></li></ol></li><li><h3>Module 7 · Bridges</h3><p>Why designs grow more than one bus, and the packet bridge that moves work between them.</p><ol><li><a href="/learn/why-bridges/">19 Why bridges</a></li><li><a href="/learn/the-packet-bridge/">20 The packet bridge</a></li><li><a href="/learn/everything-in-one-bench/">21 Everything in one bench</a></li></ol></li><li><h3>Module 8 · Ethernet</h3><p>Frames, the frame check sequence, RGMII timing, and the pre-registered steps of a real bring-up.</p><ol><li><a href="/learn/frames-and-the-fcs/">22 Frames and the FCS</a></li><li><a href="/learn/rgmii-timing/">23 RGMII timing</a></li><li><a href="/learn/bring-up-on-the-bench/">24 Bring-up on the bench</a></li></ol></li><li><h3>Module 9 · The bench</h3><p>The discipline that guards real hardware: who holds the IO, taking and giving the lock, and what runs next.</p><ol><li><a href="/learn/who-holds-the-io/">25 Who holds the IO</a></li><li class="on" aria-current="page"><a href="/learn/claim-and-release/">26 Claim and release</a></li><li><a href="/learn/capstone-what-runs-next/">27 Capstone: what runs next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/clocks-and-cdc/index.html b/apps/website/public/learn/clocks-and-cdc/index.html
index 43486ed565..b8de797039 100644
--- a/apps/website/public/learn/clocks-and-cdc/index.html
+++ b/apps/website/public/learn/clocks-and-cdc/index.html
@@ -67,5 +67,6 @@ <h1>Clocking, resets and clock-domain crossings with t27</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a clock is</h3><p>One edge, one world: what shares a clock edge shares a world, the period and the jitter of a real edge, and where the clock enters a board.</p><ol><li><a href="/learn/one-edge-one-world/">01 One edge, one world</a></li><li><a href="/learn/period-and-jitter/">02 A period, an edge, and jitter</a></li><li><a href="/learn/where-the-clock-enters/">03 Where the clock enters</a></li></ol></li><li><h3>Module 2 · Clock trees</h3><p>Skew and insertion delay, the global buffer network, and the trap of gating a clock with logic.</p><ol><li><a href="/learn/skew-and-insertion/">04 Skew and insertion delay</a></li><li><a href="/learn/global-buffers/">05 The global buffer</a></li><li><a href="/learn/gated-clock-trap/">06 The gated-clock trap</a></li></ol></li><li><h3>Module 3 · PLL and MMCM</h3><p>Multiply and divide one clock into another, move its phase in steps of the VCO, and which clocks the analyzer treats as related.</p><ol><li><a href="/learn/multiply-and-divide/">07 Multiply and divide</a></li><li><a href="/learn/phase-and-lock/">08 Phase and lock</a></li><li><a href="/learn/related-or-not/">09 Related or not</a></li></ol></li><li><h3>Module 4 · Resets</h3><p>Assert asynchronously, release synchronously: the three reset kinds, the release pipe, and the tree a reset grows.</p><ol><li><a href="/learn/sync-or-async/">10 Sync or async</a></li><li><a href="/learn/reset-release/">11 The release pipe</a></li><li><a href="/learn/reset-fanout/">12 Reset fanout</a></li></ol></li><li><h3>Module 5 · Metastability</h3><p>The setup-hold window, the mean time between failures in integer arithmetic, and the two flops that fix it.</p><ol><li><a href="/learn/setup-hold-window/">13 The setup-hold window</a></li><li><a href="/learn/mtbf/">14 Mean time between failures</a></li><li><a href="/learn/two-flop-sync/">15 The two-flop synchronizer</a></li></ol></li><li><h3>Module 6 · Crossing many bits</h3><p>Why a binary bus tears, why Gray code does not, and the handshake that moves a pulse between worlds.</p><ol><li><a href="/learn/why-binary-fails/">16 Why binary fails</a></li><li><a href="/learn/gray-code/">17 Gray code</a></li><li><a href="/learn/handshakes-and-pulses/">18 Handshakes and pulses</a></li></ol></li><li><h3>Module 7 · The asynchronous FIFO</h3><p>Pointers, flags and depth: the buffer that moves a stream between two clocks.</p><ol><li><a href="/learn/pointers-and-flags/">19 Pointers and flags</a></li><li><a href="/learn/depth-sizing/">20 Sizing the depth</a></li><li><a href="/learn/fifo-testbench/">21 The FIFO testbench</a></li></ol></li><li><h3>Module 8 · Constraints</h3><p>The lines that tell the analyzer what a clock is, which paths not to check, and what the pins must meet.</p><ol><li><a href="/learn/create-clock/">22 create_clock</a></li><li><a href="/learn/false-path-and-max-delay/">23 False paths and max delay</a></li><li><a href="/learn/io-timing/">24 IO timing</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>A CDC report, one crossing captured at the flip-flops, and the bitstream diff that closes the course.</p><ol><li><a href="/learn/cdc-report/">25 The CDC report</a></li><li><a href="/learn/capture-a-crossing/">26 Capture a crossing</a></li><li><a href="/learn/capstone-diff/">27 Capstone: change it and diff it</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/comments-and-prose/index.html b/apps/website/public/learn/comments-and-prose/index.html
index 55db3dfb25..f4c2a609a6 100644
--- a/apps/website/public/learn/comments-and-prose/index.html
+++ b/apps/website/public/learn/comments-and-prose/index.html
@@ -85,5 +85,6 @@ <h1>Comments and prose lines</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · A spec file</h3><p>What a .t27 file holds: a module line, prose lines and comments.</p><ol><li><a href="/learn/what-a-spec-is/">01 What a spec is</a></li><li><a href="/learn/the-module-line/">02 The module line</a></li><li class="on" aria-current="page"><a href="/learn/comments-and-prose/">03 Comments and prose lines</a></li></ol></li><li><h3>Module 2 · Values and types</h3><p>Constants, the width of a whole number, and true, false and text.</p><ol><li><a href="/learn/constants/">04 Constants</a></li><li><a href="/learn/whole-numbers/">05 Whole numbers and their widths</a></li><li><a href="/learn/true-false-and-text/">06 True, false and text</a></li></ol></li><li><h3>Module 3 · Arrays, trits and expressions</h3><p>Lists of values, the three values of a trit, and what operators compute.</p><ol><li><a href="/learn/arrays/">07 Arrays</a></li><li><a href="/learn/trits/">08 Trits: minus one, zero, one</a></li><li><a href="/learn/expressions/">09 Expressions</a></li></ol></li><li><h3>Module 4 · Tests and functions</h3><p>A test block, several small tests in one spec, and a function with typed inputs.</p><ol><li><a href="/learn/a-test-block/">10 A test block</a></li><li><a href="/learn/many-tests/">11 Many tests in one spec</a></li><li><a href="/learn/functions/">12 Functions</a></li></ol></li><li><h3>Module 5 · Inside a function</h3><p>Names with let and var, choices with if and switch, and loops.</p><ol><li><a href="/learn/local-names/">13 Local names: let and var</a></li><li><a href="/learn/if-and-else/">14 Choices: if, else and switch</a></li><li><a href="/learn/loops/">15 Loops: while and for</a></li></ol></li><li><h3>Module 6 · Visibility and your own types</h3><p>What pub shares with other modules, and the structs and enums you declare.</p><ol><li><a href="/learn/pub-or-private/">16 pub or private</a></li><li><a href="/learn/structs/">17 Structs</a></li><li><a href="/learn/enums/">18 Enums</a></li></ol></li><li><h3>Module 7 · Modules, rules and gen-ts</h3><p>Reading another module, stating a rule that always holds, and a spec turned into TypeScript.</p><ol><li><a href="/learn/use-other-modules/">19 Using other modules</a></li><li><a href="/learn/invariants/">20 Invariants and benches</a></li><li><a href="/learn/gen-ts/">21 gen-ts: a spec becomes TypeScript</a></li></ol></li><li><h3>Module 8 · The compiler and tri</h3><p>The backends t27c writes, the stages it reads a spec in, and the tri command.</p><ol><li><a href="/learn/seven-backends/">22 Seven backends</a></li><li><a href="/learn/stages-of-t27c/">23 The stages of t27c</a></li><li><a href="/learn/tri-commands/">24 tri: the command line</a></li></ol></li><li><h3>Module 9 · Errors, a program, what next</h3><p>Reading a compiler error, one small complete program, and the course to take next.</p><ol><li><a href="/learn/reading-errors/">25 Reading compiler errors</a></li><li><a href="/learn/a-small-program/">26 A small complete program</a></li><li><a href="/learn/where-next/">27 Where to go next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/compare-at-the-top/index.html b/apps/website/public/learn/compare-at-the-top/index.html
index 845eb5d160..5881bc9128 100644
--- a/apps/website/public/learn/compare-at-the-top/index.html
+++ b/apps/website/public/learn/compare-at-the-top/index.html
@@ -769,5 +769,6 @@ <h1>Compare at the top</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Three values</h3><p>Why three, how balanced ternary writes every number without a sign, and what flipping and cutting trits do.</p><ol><li><a href="/learn/why-three/">01 Why three</a></li><li><a href="/learn/balanced-ternary/">02 Balanced ternary</a></li><li><a href="/learn/flip-and-cut/">03 Flip to negate, cut to round</a></li></ol></li><li><h3>Module 2 · Logic with unknown</h3><p>Kleene's three-valued gates, two gates binary has no twin for, and addition as a pair of tables.</p><ol><li><a href="/learn/kleene-logic/">04 Three-valued logic</a></li><li><a href="/learn/consensus-and-any/">05 Consensus and any</a></li><li><a href="/learn/sum-and-carry/">06 Sum and carry</a></li></ol></li><li><h3>Module 3 · Adding trits</h3><p>The half adder, the full adder and a carry that ripples left, one place at a time.</p><ol><li><a href="/learn/half-adder/">07 The half adder</a></li><li><a href="/learn/full-adder/">08 The full adder</a></li><li><a href="/learn/ripple-carry/">09 Ripple carry</a></li></ol></li><li><h3>Module 4 · Multiplying without a multiplier</h3><p>Copy, drop or flip: a product by one trit, long multiplication, and a MAC that only adds.</p><ol><li><a href="/learn/times-a-trit/">10 Times one trit</a></li><li><a href="/learn/long-multiplication/">11 Long multiplication</a></li><li><a href="/learn/ternary-mac/">12 A ternary MAC</a></li></ol></li><li><h3>Module 5 · Trits in binary memory</h3><p>Two bits per trit, five trits per byte, and the bits a 27-trit word needs.</p><ol><li><a href="/learn/two-bits-per-trit/">13 Two bits per trit</a></li><li><a href="/learn/five-trits-a-byte/">14 Five trits in a byte</a></li><li><a href="/learn/a-word-of-27/">15 A word of 27 trits</a></li></ol></li><li><h3>Module 6 · The TRI-27 instruction word</h3><p>The 32-bit word the Trinity emulator decodes: its fields, its 47 opcodes and its 15-bit immediate.</p><ol><li><a href="/learn/the-instruction-word/">16 The instruction word</a></li><li><a href="/learn/the-opcode-map/">17 The opcode map</a></li><li><a href="/learn/the-immediate/">18 The immediate</a></li></ol></li><li><h3>Module 7 · A ternary machine</h3><p>An ALU built from this course's adders, a three-way jump, and a program you can step.</p><ol><li><a href="/learn/a-ternary-alu/">19 A ternary ALU</a></li><li><a href="/learn/three-way-jump/">20 A three-way jump</a></li><li><a href="/learn/run-a-program/">21 Run a program</a></li></ol></li><li><h3>Module 8 · Three answers</h3><p>Compare at the highest differing trit, find a number in thirds, sort with three-way compares.</p><ol><li class="on" aria-current="page"><a href="/learn/compare-at-the-top/">22 Compare at the top</a></li><li><a href="/learn/questions-in-thirds/">23 Questions in thirds</a></li><li><a href="/learn/sort-by-three/">24 Sort with three-way compares</a></li></ol></li><li><h3>Module 9 · Ternary neurons</h3><p>Weights of -1, 0, +1: one neuron, a detector and a small layer, with no multiplier anywhere.</p><ol><li><a href="/learn/one-ternary-neuron/">25 One ternary neuron</a></li><li><a href="/learn/a-bar-detector/">26 A bar detector</a></li><li><a href="/learn/a-tiny-layer/">27 A tiny layer</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/compare-traces/index.html b/apps/website/public/learn/compare-traces/index.html
index 6933ed793a..d7722ef9e4 100644
--- a/apps/website/public/learn/compare-traces/index.html
+++ b/apps/website/public/learn/compare-traces/index.html
@@ -181,5 +181,6 @@ <h1>Compare two traces</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Why verify</h3><p>Designs that compile and are wrong, the model that decides, and the plan written before the code.</p><ol><li><a href="/learn/bugs-that-compile/">01 Bugs that compile</a></li><li><a href="/learn/golden-model/">02 The golden model</a></li><li><a href="/learn/plan-before-code/">03 Plan before code</a></li></ol></li><li><h3>Module 2 · Testbenches</h3><p>Stimulus, checks and a verdict, written as one spec beside the design it judges.</p><ol><li><a href="/learn/anatomy-of-a-tb/">04 Anatomy of a testbench</a></li><li><a href="/learn/self-checking/">05 Self-checking, not staring</a></li><li><a href="/learn/directed-vs-random/">06 Directed versus random</a></li></ol></li><li><h3>Module 3 · Waveforms</h3><p>A trace of every signal, read the way a hardware engineer reads it, and two runs compared.</p><ol><li><a href="/learn/vcd-format/">07 The VCD format</a></li><li><a href="/learn/debug-with-waves/">08 Debug with waves</a></li><li class="on" aria-current="page"><a href="/learn/compare-traces/">09 Compare two traces</a></li></ol></li><li><h3>Module 4 · Conformance vectors</h3><p>Cases with the answer written beside them, kept where the compiler can reach them.</p><ol><li><a href="/learn/what-a-vector-is/">10 What a vector is</a></li><li><a href="/learn/golden-ruler-vectors/">11 Golden ruler vectors</a></li><li><a href="/learn/end-to-end-scenarios/">12 End-to-end scenarios</a></li></ol></li><li><h3>Module 5 · Cosimulation</h3><p>Spec, simulator and board agreeing on the bench Artix-7 XC7A200T, and what to do when they do not.</p><ol><li><a href="/learn/three-way-agreement/">13 Three-way agreement</a></li><li><a href="/learn/simulator-internals/">14 Inside the simulator</a></li><li><a href="/learn/when-they-disagree/">15 When they disagree</a></li></ol></li><li><h3>Module 6 · Coverage</h3><p>What the tests touched: lines, toggles, states, and what that number hides.</p><ol><li><a href="/learn/line-and-toggle/">16 Line and toggle</a></li><li><a href="/learn/fsm-coverage/">17 FSM coverage</a></li><li><a href="/learn/coverage-lies/">18 Coverage lies</a></li></ol></li><li><h3>Module 7 · Formal</h3><p>Assertions that hold every cycle, bounded search for a counterexample, and why a proof needs induction.</p><ol><li><a href="/learn/assertions/">19 Assertions</a></li><li><a href="/learn/bounded-model-checking/">20 Bounded model checking</a></li><li><a href="/learn/induction/">21 Induction</a></li></ol></li><li><h3>Module 8 · Mutation</h3><p>Break the design on purpose and count what the tests catch.</p><ol><li><a href="/learn/kill-the-mutant/">22 Kill the mutant</a></li><li><a href="/learn/surviving-mutants/">23 Surviving mutants</a></li><li><a href="/learn/test-strength/">24 Test strength</a></li></ol></li><li><h3>Module 9 · Sign-off</h3><p>One command, every receipt, a clean verdict you can show.</p><ol><li><a href="/learn/build-verify/">25 Build and verify</a></li><li><a href="/learn/board-selftest/">26 The board self-tests</a></li><li><a href="/learn/verify-capstone/">27 Capstone: verify a module whole</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/compiler-stages/index.html b/apps/website/public/learn/compiler-stages/index.html
index 45827f2e44..62b9177043 100644
--- a/apps/website/public/learn/compiler-stages/index.html
+++ b/apps/website/public/learn/compiler-stages/index.html
@@ -210,5 +210,6 @@ <h1>The stages of t27c</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The chip</h3><p>What an FPGA is, which chip we use, and how its pins meet the board.</p><ol><li><a href="/learn/fpga-is-a-grid/">01 An FPGA is a grid you configure</a></li><li><a href="/learn/which-chip/">02 Which chip is this?</a></li><li><a href="/learn/pins-and-package/">03 Pins, balls and the package</a></li></ol></li><li><h3>Module 2 · Numbers in hardware</h3><p>Bits, trits and number formats, and what arithmetic costs in logic.</p><ol><li><a href="/learn/bits-and-trits/">04 Bits, trits and GF16</a></li><li><a href="/learn/what-arithmetic-costs/">05 What arithmetic costs</a></li><li><a href="/learn/how-much-fits/">06 How much fits on the chip</a></li></ol></li><li><h3>Module 3 · Your t27 program</h3><p>Write a spec, test it, and see why compiling is not the same as being right.</p><ol><li><a href="/learn/hello-t27/">07 Hello, t27</a></li><li><a href="/learn/tests-are-the-spec/">08 Tests are part of the spec</a></li><li><a href="/learn/compiles-is-not-correct/">09 Compiles is not the same as correct</a></li></ol></li><li><h3>Module 4 · Inside t27c</h3><p>How the compiler reads a spec and what it writes, including the native t27b.</p><ol><li class="on" aria-current="page"><a href="/learn/compiler-stages/">10 The stages of t27c</a></li><li><a href="/learn/t27c-on-a-real-machine/">11 t27c on a real machine</a></li><li><a href="/learn/seven-backends-and-t27b/">12 Seven backends, and t27b</a></li></ol></li><li><h3>Module 5 · From spec to hardware</h3><p>The Verilog t27c writes, the cells it becomes, and what one LUT does.</p><ol><li><a href="/learn/spec-to-verilog/">13 From a spec to Verilog</a></li><li><a href="/learn/count-the-cells/">14 Count the cells</a></li><li><a href="/learn/one-lut-up-close/">15 One LUT, up close</a></li></ol></li><li><h3>Module 6 · Reading synthesis</h3><p>What yosys reports about your design, and which warnings matter.</p><ol><li><a href="/learn/synthesis-warnings/">16 Synthesis warnings that matter</a></li><li><a href="/learn/state-machines/">17 State machines</a></li><li><a href="/learn/where-the-luts-go/">18 Where the LUTs go</a></li></ol></li><li><h3>Module 7 · Place, route, timing</h3><p>Where the cells land on the die, and whether the clock is met.</p><ol><li><a href="/learn/placement/">19 Placement: where the cells land</a></li><li><a href="/learn/logic-depth/">20 Logic depth</a></li><li><a href="/learn/timing-slack/">21 Timing and slack</a></li></ol></li><li><h3>Module 8 · The bitstream</h3><p>How a routed design becomes the bits the chip loads, with no Vivado.</p><ol><li><a href="/learn/fasm-to-frames/">22 From FASM to frames</a></li><li><a href="/learn/a-build-receipt/">23 A build, itemised</a></li><li><a href="/learn/read-a-bitstream/">24 Read a bitstream</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>Ask the chip who it is, load the bits, and check them.</p><ol><li><a href="/learn/jtag-asks-the-chip/">25 JTAG asks the chip</a></li><li><a href="/learn/same-bits-on-the-board/">26 The same bits, on the board</a></li><li><a href="/learn/capstone/">27 Capstone: out-guess yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/compiles-is-not-correct/index.html b/apps/website/public/learn/compiles-is-not-correct/index.html
index 59cd0aeb76..e416539d9d 100644
--- a/apps/website/public/learn/compiles-is-not-correct/index.html
+++ b/apps/website/public/learn/compiles-is-not-correct/index.html
@@ -179,5 +179,6 @@ <h1>Compiles is not the same as correct</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The chip</h3><p>What an FPGA is, which chip we use, and how its pins meet the board.</p><ol><li><a href="/learn/fpga-is-a-grid/">01 An FPGA is a grid you configure</a></li><li><a href="/learn/which-chip/">02 Which chip is this?</a></li><li><a href="/learn/pins-and-package/">03 Pins, balls and the package</a></li></ol></li><li><h3>Module 2 · Numbers in hardware</h3><p>Bits, trits and number formats, and what arithmetic costs in logic.</p><ol><li><a href="/learn/bits-and-trits/">04 Bits, trits and GF16</a></li><li><a href="/learn/what-arithmetic-costs/">05 What arithmetic costs</a></li><li><a href="/learn/how-much-fits/">06 How much fits on the chip</a></li></ol></li><li><h3>Module 3 · Your t27 program</h3><p>Write a spec, test it, and see why compiling is not the same as being right.</p><ol><li><a href="/learn/hello-t27/">07 Hello, t27</a></li><li><a href="/learn/tests-are-the-spec/">08 Tests are part of the spec</a></li><li class="on" aria-current="page"><a href="/learn/compiles-is-not-correct/">09 Compiles is not the same as correct</a></li></ol></li><li><h3>Module 4 · Inside t27c</h3><p>How the compiler reads a spec and what it writes, including the native t27b.</p><ol><li><a href="/learn/compiler-stages/">10 The stages of t27c</a></li><li><a href="/learn/t27c-on-a-real-machine/">11 t27c on a real machine</a></li><li><a href="/learn/seven-backends-and-t27b/">12 Seven backends, and t27b</a></li></ol></li><li><h3>Module 5 · From spec to hardware</h3><p>The Verilog t27c writes, the cells it becomes, and what one LUT does.</p><ol><li><a href="/learn/spec-to-verilog/">13 From a spec to Verilog</a></li><li><a href="/learn/count-the-cells/">14 Count the cells</a></li><li><a href="/learn/one-lut-up-close/">15 One LUT, up close</a></li></ol></li><li><h3>Module 6 · Reading synthesis</h3><p>What yosys reports about your design, and which warnings matter.</p><ol><li><a href="/learn/synthesis-warnings/">16 Synthesis warnings that matter</a></li><li><a href="/learn/state-machines/">17 State machines</a></li><li><a href="/learn/where-the-luts-go/">18 Where the LUTs go</a></li></ol></li><li><h3>Module 7 · Place, route, timing</h3><p>Where the cells land on the die, and whether the clock is met.</p><ol><li><a href="/learn/placement/">19 Placement: where the cells land</a></li><li><a href="/learn/logic-depth/">20 Logic depth</a></li><li><a href="/learn/timing-slack/">21 Timing and slack</a></li></ol></li><li><h3>Module 8 · The bitstream</h3><p>How a routed design becomes the bits the chip loads, with no Vivado.</p><ol><li><a href="/learn/fasm-to-frames/">22 From FASM to frames</a></li><li><a href="/learn/a-build-receipt/">23 A build, itemised</a></li><li><a href="/learn/read-a-bitstream/">24 Read a bitstream</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>Ask the chip who it is, load the bits, and check them.</p><ol><li><a href="/learn/jtag-asks-the-chip/">25 JTAG asks the chip</a></li><li><a href="/learn/same-bits-on-the-board/">26 The same bits, on the board</a></li><li><a href="/learn/capstone/">27 Capstone: out-guess yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/consensus-and-any/index.html b/apps/website/public/learn/consensus-and-any/index.html
index da418119e7..91a71e0ca6 100644
--- a/apps/website/public/learn/consensus-and-any/index.html
+++ b/apps/website/public/learn/consensus-and-any/index.html
@@ -769,5 +769,6 @@ <h1>Consensus and any</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Three values</h3><p>Why three, how balanced ternary writes every number without a sign, and what flipping and cutting trits do.</p><ol><li><a href="/learn/why-three/">01 Why three</a></li><li><a href="/learn/balanced-ternary/">02 Balanced ternary</a></li><li><a href="/learn/flip-and-cut/">03 Flip to negate, cut to round</a></li></ol></li><li><h3>Module 2 · Logic with unknown</h3><p>Kleene's three-valued gates, two gates binary has no twin for, and addition as a pair of tables.</p><ol><li><a href="/learn/kleene-logic/">04 Three-valued logic</a></li><li class="on" aria-current="page"><a href="/learn/consensus-and-any/">05 Consensus and any</a></li><li><a href="/learn/sum-and-carry/">06 Sum and carry</a></li></ol></li><li><h3>Module 3 · Adding trits</h3><p>The half adder, the full adder and a carry that ripples left, one place at a time.</p><ol><li><a href="/learn/half-adder/">07 The half adder</a></li><li><a href="/learn/full-adder/">08 The full adder</a></li><li><a href="/learn/ripple-carry/">09 Ripple carry</a></li></ol></li><li><h3>Module 4 · Multiplying without a multiplier</h3><p>Copy, drop or flip: a product by one trit, long multiplication, and a MAC that only adds.</p><ol><li><a href="/learn/times-a-trit/">10 Times one trit</a></li><li><a href="/learn/long-multiplication/">11 Long multiplication</a></li><li><a href="/learn/ternary-mac/">12 A ternary MAC</a></li></ol></li><li><h3>Module 5 · Trits in binary memory</h3><p>Two bits per trit, five trits per byte, and the bits a 27-trit word needs.</p><ol><li><a href="/learn/two-bits-per-trit/">13 Two bits per trit</a></li><li><a href="/learn/five-trits-a-byte/">14 Five trits in a byte</a></li><li><a href="/learn/a-word-of-27/">15 A word of 27 trits</a></li></ol></li><li><h3>Module 6 · The TRI-27 instruction word</h3><p>The 32-bit word the Trinity emulator decodes: its fields, its 47 opcodes and its 15-bit immediate.</p><ol><li><a href="/learn/the-instruction-word/">16 The instruction word</a></li><li><a href="/learn/the-opcode-map/">17 The opcode map</a></li><li><a href="/learn/the-immediate/">18 The immediate</a></li></ol></li><li><h3>Module 7 · A ternary machine</h3><p>An ALU built from this course's adders, a three-way jump, and a program you can step.</p><ol><li><a href="/learn/a-ternary-alu/">19 A ternary ALU</a></li><li><a href="/learn/three-way-jump/">20 A three-way jump</a></li><li><a href="/learn/run-a-program/">21 Run a program</a></li></ol></li><li><h3>Module 8 · Three answers</h3><p>Compare at the highest differing trit, find a number in thirds, sort with three-way compares.</p><ol><li><a href="/learn/compare-at-the-top/">22 Compare at the top</a></li><li><a href="/learn/questions-in-thirds/">23 Questions in thirds</a></li><li><a href="/learn/sort-by-three/">24 Sort with three-way compares</a></li></ol></li><li><h3>Module 9 · Ternary neurons</h3><p>Weights of -1, 0, +1: one neuron, a detector and a small layer, with no multiplier anywhere.</p><ol><li><a href="/learn/one-ternary-neuron/">25 One ternary neuron</a></li><li><a href="/learn/a-bar-detector/">26 A bar detector</a></li><li><a href="/learn/a-tiny-layer/">27 A tiny layer</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/constants/index.html b/apps/website/public/learn/constants/index.html
index b200cc7292..9e017d8528 100644
--- a/apps/website/public/learn/constants/index.html
+++ b/apps/website/public/learn/constants/index.html
@@ -83,5 +83,6 @@ <h1>Constants</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · A spec file</h3><p>What a .t27 file holds: a module line, prose lines and comments.</p><ol><li><a href="/learn/what-a-spec-is/">01 What a spec is</a></li><li><a href="/learn/the-module-line/">02 The module line</a></li><li><a href="/learn/comments-and-prose/">03 Comments and prose lines</a></li></ol></li><li><h3>Module 2 · Values and types</h3><p>Constants, the width of a whole number, and true, false and text.</p><ol><li class="on" aria-current="page"><a href="/learn/constants/">04 Constants</a></li><li><a href="/learn/whole-numbers/">05 Whole numbers and their widths</a></li><li><a href="/learn/true-false-and-text/">06 True, false and text</a></li></ol></li><li><h3>Module 3 · Arrays, trits and expressions</h3><p>Lists of values, the three values of a trit, and what operators compute.</p><ol><li><a href="/learn/arrays/">07 Arrays</a></li><li><a href="/learn/trits/">08 Trits: minus one, zero, one</a></li><li><a href="/learn/expressions/">09 Expressions</a></li></ol></li><li><h3>Module 4 · Tests and functions</h3><p>A test block, several small tests in one spec, and a function with typed inputs.</p><ol><li><a href="/learn/a-test-block/">10 A test block</a></li><li><a href="/learn/many-tests/">11 Many tests in one spec</a></li><li><a href="/learn/functions/">12 Functions</a></li></ol></li><li><h3>Module 5 · Inside a function</h3><p>Names with let and var, choices with if and switch, and loops.</p><ol><li><a href="/learn/local-names/">13 Local names: let and var</a></li><li><a href="/learn/if-and-else/">14 Choices: if, else and switch</a></li><li><a href="/learn/loops/">15 Loops: while and for</a></li></ol></li><li><h3>Module 6 · Visibility and your own types</h3><p>What pub shares with other modules, and the structs and enums you declare.</p><ol><li><a href="/learn/pub-or-private/">16 pub or private</a></li><li><a href="/learn/structs/">17 Structs</a></li><li><a href="/learn/enums/">18 Enums</a></li></ol></li><li><h3>Module 7 · Modules, rules and gen-ts</h3><p>Reading another module, stating a rule that always holds, and a spec turned into TypeScript.</p><ol><li><a href="/learn/use-other-modules/">19 Using other modules</a></li><li><a href="/learn/invariants/">20 Invariants and benches</a></li><li><a href="/learn/gen-ts/">21 gen-ts: a spec becomes TypeScript</a></li></ol></li><li><h3>Module 8 · The compiler and tri</h3><p>The backends t27c writes, the stages it reads a spec in, and the tri command.</p><ol><li><a href="/learn/seven-backends/">22 Seven backends</a></li><li><a href="/learn/stages-of-t27c/">23 The stages of t27c</a></li><li><a href="/learn/tri-commands/">24 tri: the command line</a></li></ol></li><li><h3>Module 9 · Errors, a program, what next</h3><p>Reading a compiler error, one small complete program, and the course to take next.</p><ol><li><a href="/learn/reading-errors/">25 Reading compiler errors</a></li><li><a href="/learn/a-small-program/">26 A small complete program</a></li><li><a href="/learn/where-next/">27 Where to go next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/count-the-cells/index.html b/apps/website/public/learn/count-the-cells/index.html
index eda627c0b9..e57923e98d 100644
--- a/apps/website/public/learn/count-the-cells/index.html
+++ b/apps/website/public/learn/count-the-cells/index.html
@@ -222,5 +222,6 @@ <h1>Count the cells</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The chip</h3><p>What an FPGA is, which chip we use, and how its pins meet the board.</p><ol><li><a href="/learn/fpga-is-a-grid/">01 An FPGA is a grid you configure</a></li><li><a href="/learn/which-chip/">02 Which chip is this?</a></li><li><a href="/learn/pins-and-package/">03 Pins, balls and the package</a></li></ol></li><li><h3>Module 2 · Numbers in hardware</h3><p>Bits, trits and number formats, and what arithmetic costs in logic.</p><ol><li><a href="/learn/bits-and-trits/">04 Bits, trits and GF16</a></li><li><a href="/learn/what-arithmetic-costs/">05 What arithmetic costs</a></li><li><a href="/learn/how-much-fits/">06 How much fits on the chip</a></li></ol></li><li><h3>Module 3 · Your t27 program</h3><p>Write a spec, test it, and see why compiling is not the same as being right.</p><ol><li><a href="/learn/hello-t27/">07 Hello, t27</a></li><li><a href="/learn/tests-are-the-spec/">08 Tests are part of the spec</a></li><li><a href="/learn/compiles-is-not-correct/">09 Compiles is not the same as correct</a></li></ol></li><li><h3>Module 4 · Inside t27c</h3><p>How the compiler reads a spec and what it writes, including the native t27b.</p><ol><li><a href="/learn/compiler-stages/">10 The stages of t27c</a></li><li><a href="/learn/t27c-on-a-real-machine/">11 t27c on a real machine</a></li><li><a href="/learn/seven-backends-and-t27b/">12 Seven backends, and t27b</a></li></ol></li><li><h3>Module 5 · From spec to hardware</h3><p>The Verilog t27c writes, the cells it becomes, and what one LUT does.</p><ol><li><a href="/learn/spec-to-verilog/">13 From a spec to Verilog</a></li><li class="on" aria-current="page"><a href="/learn/count-the-cells/">14 Count the cells</a></li><li><a href="/learn/one-lut-up-close/">15 One LUT, up close</a></li></ol></li><li><h3>Module 6 · Reading synthesis</h3><p>What yosys reports about your design, and which warnings matter.</p><ol><li><a href="/learn/synthesis-warnings/">16 Synthesis warnings that matter</a></li><li><a href="/learn/state-machines/">17 State machines</a></li><li><a href="/learn/where-the-luts-go/">18 Where the LUTs go</a></li></ol></li><li><h3>Module 7 · Place, route, timing</h3><p>Where the cells land on the die, and whether the clock is met.</p><ol><li><a href="/learn/placement/">19 Placement: where the cells land</a></li><li><a href="/learn/logic-depth/">20 Logic depth</a></li><li><a href="/learn/timing-slack/">21 Timing and slack</a></li></ol></li><li><h3>Module 8 · The bitstream</h3><p>How a routed design becomes the bits the chip loads, with no Vivado.</p><ol><li><a href="/learn/fasm-to-frames/">22 From FASM to frames</a></li><li><a href="/learn/a-build-receipt/">23 A build, itemised</a></li><li><a href="/learn/read-a-bitstream/">24 Read a bitstream</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>Ask the chip who it is, load the bits, and check them.</p><ol><li><a href="/learn/jtag-asks-the-chip/">25 JTAG asks the chip</a></li><li><a href="/learn/same-bits-on-the-board/">26 The same bits, on the board</a></li><li><a href="/learn/capstone/">27 Capstone: out-guess yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/course/index.html b/apps/website/public/learn/course/index.html
index 127e6e2e3d..08d57667a3 100644
--- a/apps/website/public/learn/course/index.html
+++ b/apps/website/public/learn/course/index.html
@@ -67,5 +67,6 @@ <h1>From zero to a chip: FPGA programming with t27</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The chip</h3><p>What an FPGA is, which chip we use, and how its pins meet the board.</p><ol><li><a href="/learn/fpga-is-a-grid/">01 An FPGA is a grid you configure</a></li><li><a href="/learn/which-chip/">02 Which chip is this?</a></li><li><a href="/learn/pins-and-package/">03 Pins, balls and the package</a></li></ol></li><li><h3>Module 2 · Numbers in hardware</h3><p>Bits, trits and number formats, and what arithmetic costs in logic.</p><ol><li><a href="/learn/bits-and-trits/">04 Bits, trits and GF16</a></li><li><a href="/learn/what-arithmetic-costs/">05 What arithmetic costs</a></li><li><a href="/learn/how-much-fits/">06 How much fits on the chip</a></li></ol></li><li><h3>Module 3 · Your t27 program</h3><p>Write a spec, test it, and see why compiling is not the same as being right.</p><ol><li><a href="/learn/hello-t27/">07 Hello, t27</a></li><li><a href="/learn/tests-are-the-spec/">08 Tests are part of the spec</a></li><li><a href="/learn/compiles-is-not-correct/">09 Compiles is not the same as correct</a></li></ol></li><li><h3>Module 4 · Inside t27c</h3><p>How the compiler reads a spec and what it writes, including the native t27b.</p><ol><li><a href="/learn/compiler-stages/">10 The stages of t27c</a></li><li><a href="/learn/t27c-on-a-real-machine/">11 t27c on a real machine</a></li><li><a href="/learn/seven-backends-and-t27b/">12 Seven backends, and t27b</a></li></ol></li><li><h3>Module 5 · From spec to hardware</h3><p>The Verilog t27c writes, the cells it becomes, and what one LUT does.</p><ol><li><a href="/learn/spec-to-verilog/">13 From a spec to Verilog</a></li><li><a href="/learn/count-the-cells/">14 Count the cells</a></li><li><a href="/learn/one-lut-up-close/">15 One LUT, up close</a></li></ol></li><li><h3>Module 6 · Reading synthesis</h3><p>What yosys reports about your design, and which warnings matter.</p><ol><li><a href="/learn/synthesis-warnings/">16 Synthesis warnings that matter</a></li><li><a href="/learn/state-machines/">17 State machines</a></li><li><a href="/learn/where-the-luts-go/">18 Where the LUTs go</a></li></ol></li><li><h3>Module 7 · Place, route, timing</h3><p>Where the cells land on the die, and whether the clock is met.</p><ol><li><a href="/learn/placement/">19 Placement: where the cells land</a></li><li><a href="/learn/logic-depth/">20 Logic depth</a></li><li><a href="/learn/timing-slack/">21 Timing and slack</a></li></ol></li><li><h3>Module 8 · The bitstream</h3><p>How a routed design becomes the bits the chip loads, with no Vivado.</p><ol><li><a href="/learn/fasm-to-frames/">22 From FASM to frames</a></li><li><a href="/learn/a-build-receipt/">23 A build, itemised</a></li><li><a href="/learn/read-a-bitstream/">24 Read a bitstream</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>Ask the chip who it is, load the bits, and check them.</p><ol><li><a href="/learn/jtag-asks-the-chip/">25 JTAG asks the chip</a></li><li><a href="/learn/same-bits-on-the-board/">26 The same bits, on the board</a></li><li><a href="/learn/capstone/">27 Capstone: out-guess yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/coverage-lies/index.html b/apps/website/public/learn/coverage-lies/index.html
index 3e3dff5107..ed22a2e567 100644
--- a/apps/website/public/learn/coverage-lies/index.html
+++ b/apps/website/public/learn/coverage-lies/index.html
@@ -304,5 +304,6 @@ <h1>Coverage lies</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Why verify</h3><p>Designs that compile and are wrong, the model that decides, and the plan written before the code.</p><ol><li><a href="/learn/bugs-that-compile/">01 Bugs that compile</a></li><li><a href="/learn/golden-model/">02 The golden model</a></li><li><a href="/learn/plan-before-code/">03 Plan before code</a></li></ol></li><li><h3>Module 2 · Testbenches</h3><p>Stimulus, checks and a verdict, written as one spec beside the design it judges.</p><ol><li><a href="/learn/anatomy-of-a-tb/">04 Anatomy of a testbench</a></li><li><a href="/learn/self-checking/">05 Self-checking, not staring</a></li><li><a href="/learn/directed-vs-random/">06 Directed versus random</a></li></ol></li><li><h3>Module 3 · Waveforms</h3><p>A trace of every signal, read the way a hardware engineer reads it, and two runs compared.</p><ol><li><a href="/learn/vcd-format/">07 The VCD format</a></li><li><a href="/learn/debug-with-waves/">08 Debug with waves</a></li><li><a href="/learn/compare-traces/">09 Compare two traces</a></li></ol></li><li><h3>Module 4 · Conformance vectors</h3><p>Cases with the answer written beside them, kept where the compiler can reach them.</p><ol><li><a href="/learn/what-a-vector-is/">10 What a vector is</a></li><li><a href="/learn/golden-ruler-vectors/">11 Golden ruler vectors</a></li><li><a href="/learn/end-to-end-scenarios/">12 End-to-end scenarios</a></li></ol></li><li><h3>Module 5 · Cosimulation</h3><p>Spec, simulator and board agreeing on the bench Artix-7 XC7A200T, and what to do when they do not.</p><ol><li><a href="/learn/three-way-agreement/">13 Three-way agreement</a></li><li><a href="/learn/simulator-internals/">14 Inside the simulator</a></li><li><a href="/learn/when-they-disagree/">15 When they disagree</a></li></ol></li><li><h3>Module 6 · Coverage</h3><p>What the tests touched: lines, toggles, states, and what that number hides.</p><ol><li><a href="/learn/line-and-toggle/">16 Line and toggle</a></li><li><a href="/learn/fsm-coverage/">17 FSM coverage</a></li><li class="on" aria-current="page"><a href="/learn/coverage-lies/">18 Coverage lies</a></li></ol></li><li><h3>Module 7 · Formal</h3><p>Assertions that hold every cycle, bounded search for a counterexample, and why a proof needs induction.</p><ol><li><a href="/learn/assertions/">19 Assertions</a></li><li><a href="/learn/bounded-model-checking/">20 Bounded model checking</a></li><li><a href="/learn/induction/">21 Induction</a></li></ol></li><li><h3>Module 8 · Mutation</h3><p>Break the design on purpose and count what the tests catch.</p><ol><li><a href="/learn/kill-the-mutant/">22 Kill the mutant</a></li><li><a href="/learn/surviving-mutants/">23 Surviving mutants</a></li><li><a href="/learn/test-strength/">24 Test strength</a></li></ol></li><li><h3>Module 9 · Sign-off</h3><p>One command, every receipt, a clean verdict you can show.</p><ol><li><a href="/learn/build-verify/">25 Build and verify</a></li><li><a href="/learn/board-selftest/">26 The board self-tests</a></li><li><a href="/learn/verify-capstone/">27 Capstone: verify a module whole</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/create-clock/index.html b/apps/website/public/learn/create-clock/index.html
index 1fbaa3a5ad..132da40fdb 100644
--- a/apps/website/public/learn/create-clock/index.html
+++ b/apps/website/public/learn/create-clock/index.html
@@ -364,5 +364,6 @@ <h1>create_clock</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a clock is</h3><p>One edge, one world: what shares a clock edge shares a world, the period and the jitter of a real edge, and where the clock enters a board.</p><ol><li><a href="/learn/one-edge-one-world/">01 One edge, one world</a></li><li><a href="/learn/period-and-jitter/">02 A period, an edge, and jitter</a></li><li><a href="/learn/where-the-clock-enters/">03 Where the clock enters</a></li></ol></li><li><h3>Module 2 · Clock trees</h3><p>Skew and insertion delay, the global buffer network, and the trap of gating a clock with logic.</p><ol><li><a href="/learn/skew-and-insertion/">04 Skew and insertion delay</a></li><li><a href="/learn/global-buffers/">05 The global buffer</a></li><li><a href="/learn/gated-clock-trap/">06 The gated-clock trap</a></li></ol></li><li><h3>Module 3 · PLL and MMCM</h3><p>Multiply and divide one clock into another, move its phase in steps of the VCO, and which clocks the analyzer treats as related.</p><ol><li><a href="/learn/multiply-and-divide/">07 Multiply and divide</a></li><li><a href="/learn/phase-and-lock/">08 Phase and lock</a></li><li><a href="/learn/related-or-not/">09 Related or not</a></li></ol></li><li><h3>Module 4 · Resets</h3><p>Assert asynchronously, release synchronously: the three reset kinds, the release pipe, and the tree a reset grows.</p><ol><li><a href="/learn/sync-or-async/">10 Sync or async</a></li><li><a href="/learn/reset-release/">11 The release pipe</a></li><li><a href="/learn/reset-fanout/">12 Reset fanout</a></li></ol></li><li><h3>Module 5 · Metastability</h3><p>The setup-hold window, the mean time between failures in integer arithmetic, and the two flops that fix it.</p><ol><li><a href="/learn/setup-hold-window/">13 The setup-hold window</a></li><li><a href="/learn/mtbf/">14 Mean time between failures</a></li><li><a href="/learn/two-flop-sync/">15 The two-flop synchronizer</a></li></ol></li><li><h3>Module 6 · Crossing many bits</h3><p>Why a binary bus tears, why Gray code does not, and the handshake that moves a pulse between worlds.</p><ol><li><a href="/learn/why-binary-fails/">16 Why binary fails</a></li><li><a href="/learn/gray-code/">17 Gray code</a></li><li><a href="/learn/handshakes-and-pulses/">18 Handshakes and pulses</a></li></ol></li><li><h3>Module 7 · The asynchronous FIFO</h3><p>Pointers, flags and depth: the buffer that moves a stream between two clocks.</p><ol><li><a href="/learn/pointers-and-flags/">19 Pointers and flags</a></li><li><a href="/learn/depth-sizing/">20 Sizing the depth</a></li><li><a href="/learn/fifo-testbench/">21 The FIFO testbench</a></li></ol></li><li><h3>Module 8 · Constraints</h3><p>The lines that tell the analyzer what a clock is, which paths not to check, and what the pins must meet.</p><ol><li class="on" aria-current="page"><a href="/learn/create-clock/">22 create_clock</a></li><li><a href="/learn/false-path-and-max-delay/">23 False paths and max delay</a></li><li><a href="/learn/io-timing/">24 IO timing</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>A CDC report, one crossing captured at the flip-flops, and the bitstream diff that closes the course.</p><ol><li><a href="/learn/cdc-report/">25 The CDC report</a></li><li><a href="/learn/capture-a-crossing/">26 Capture a crossing</a></li><li><a href="/learn/capstone-diff/">27 Capstone: change it and diff it</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/debug-with-waves/index.html b/apps/website/public/learn/debug-with-waves/index.html
index 04f5e681d0..0534c92a06 100644
--- a/apps/website/public/learn/debug-with-waves/index.html
+++ b/apps/website/public/learn/debug-with-waves/index.html
@@ -161,5 +161,6 @@ <h1>Debug with waves</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Why verify</h3><p>Designs that compile and are wrong, the model that decides, and the plan written before the code.</p><ol><li><a href="/learn/bugs-that-compile/">01 Bugs that compile</a></li><li><a href="/learn/golden-model/">02 The golden model</a></li><li><a href="/learn/plan-before-code/">03 Plan before code</a></li></ol></li><li><h3>Module 2 · Testbenches</h3><p>Stimulus, checks and a verdict, written as one spec beside the design it judges.</p><ol><li><a href="/learn/anatomy-of-a-tb/">04 Anatomy of a testbench</a></li><li><a href="/learn/self-checking/">05 Self-checking, not staring</a></li><li><a href="/learn/directed-vs-random/">06 Directed versus random</a></li></ol></li><li><h3>Module 3 · Waveforms</h3><p>A trace of every signal, read the way a hardware engineer reads it, and two runs compared.</p><ol><li><a href="/learn/vcd-format/">07 The VCD format</a></li><li class="on" aria-current="page"><a href="/learn/debug-with-waves/">08 Debug with waves</a></li><li><a href="/learn/compare-traces/">09 Compare two traces</a></li></ol></li><li><h3>Module 4 · Conformance vectors</h3><p>Cases with the answer written beside them, kept where the compiler can reach them.</p><ol><li><a href="/learn/what-a-vector-is/">10 What a vector is</a></li><li><a href="/learn/golden-ruler-vectors/">11 Golden ruler vectors</a></li><li><a href="/learn/end-to-end-scenarios/">12 End-to-end scenarios</a></li></ol></li><li><h3>Module 5 · Cosimulation</h3><p>Spec, simulator and board agreeing on the bench Artix-7 XC7A200T, and what to do when they do not.</p><ol><li><a href="/learn/three-way-agreement/">13 Three-way agreement</a></li><li><a href="/learn/simulator-internals/">14 Inside the simulator</a></li><li><a href="/learn/when-they-disagree/">15 When they disagree</a></li></ol></li><li><h3>Module 6 · Coverage</h3><p>What the tests touched: lines, toggles, states, and what that number hides.</p><ol><li><a href="/learn/line-and-toggle/">16 Line and toggle</a></li><li><a href="/learn/fsm-coverage/">17 FSM coverage</a></li><li><a href="/learn/coverage-lies/">18 Coverage lies</a></li></ol></li><li><h3>Module 7 · Formal</h3><p>Assertions that hold every cycle, bounded search for a counterexample, and why a proof needs induction.</p><ol><li><a href="/learn/assertions/">19 Assertions</a></li><li><a href="/learn/bounded-model-checking/">20 Bounded model checking</a></li><li><a href="/learn/induction/">21 Induction</a></li></ol></li><li><h3>Module 8 · Mutation</h3><p>Break the design on purpose and count what the tests catch.</p><ol><li><a href="/learn/kill-the-mutant/">22 Kill the mutant</a></li><li><a href="/learn/surviving-mutants/">23 Surviving mutants</a></li><li><a href="/learn/test-strength/">24 Test strength</a></li></ol></li><li><h3>Module 9 · Sign-off</h3><p>One command, every receipt, a clean verdict you can show.</p><ol><li><a href="/learn/build-verify/">25 Build and verify</a></li><li><a href="/learn/board-selftest/">26 The board self-tests</a></li><li><a href="/learn/verify-capstone/">27 Capstone: verify a module whole</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/decode-in-one-step/index.html b/apps/website/public/learn/decode-in-one-step/index.html
index ea3508d4c1..066fe75a30 100644
--- a/apps/website/public/learn/decode-in-one-step/index.html
+++ b/apps/website/public/learn/decode-in-one-step/index.html
@@ -191,5 +191,6 @@ <h1>Decode in one step</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The rule and its numbers</h3><p>One rule splits every width, the ratio it aims at, and the Lucas numbers behind the 3.</p><ol><li><a href="/learn/a-float-cut-by-phi/">01 A float cut by phi</a></li><li><a href="/learn/phi-as-a-ratio/">02 Phi as a ratio</a></li><li><a href="/learn/lucas-numbers-stay-whole/">03 Lucas numbers stay whole</a></li></ol></li><li><h3>Module 2 · Why phi, why three</h3><p>Why the split is phi, why base three, and how a spec checks GF16 keeps phi.</p><ol><li><a href="/learn/why-the-split-is-phi/">04 Why the split is phi</a></li><li><a href="/learn/why-base-three/">05 Why base three</a></li><li><a href="/learn/phi-in-sixteen-bits/">06 Phi in sixteen bits</a></li></ol></li><li><h3>Module 3 · The small rungs: GF4 to GF8</h3><p>GF4, GF6 and GF8, the fewest bits, where rounding to whole bits costs the most.</p><ol><li><a href="/learn/gf4-four-bits/">07 GF4: four bits</a></li><li><a href="/learn/gf6-six-bits/">08 GF6: six bits</a></li><li><a href="/learn/gf8-one-byte/">09 GF8: one byte</a></li></ol></li><li><h3>Module 4 · Ten to fourteen bits</h3><p>GF10, GF12 and GF14, and how the distance from 1 / phi moves as the word grows.</p><ol><li><a href="/learn/gf10-ten-bits/">10 GF10: ten bits</a></li><li><a href="/learn/gf12-twelve-bits/">11 GF12: twelve bits</a></li><li><a href="/learn/gf14-fourteen-bits/">12 GF14: fourteen bits</a></li></ol></li><li><h3>Module 5 · GF16 at work</h3><p>The primary 16-bit format, a two-term dot product in GF-T16, then GF20 and GF24.</p><ol><li><a href="/learn/gf16-the-primary/">13 GF16: the primary format</a></li><li><a href="/learn/two-products-one-sum/">14 Two products, one sum</a></li><li><a href="/learn/gf24-twenty-four-bits/">15 GF20 and GF24</a></li></ol></li><li><h3>Module 6 · GF32 to GF64</h3><p>GF32 beside IEEE single, GF48 with no IEEE twin, GF64 beside IEEE double.</p><ol><li><a href="/learn/gf32-a-single/">16 GF32: a single</a></li><li><a href="/learn/gf48-forty-eight-bits/">17 GF48: forty-eight bits</a></li><li><a href="/learn/gf64-a-double/">18 GF64: a double</a></li></ol></li><li><h3>Module 7 · GF96 to GF256</h3><p>GF96, GF128 and GF256, where the specs hold the layout with invariants.</p><ol><li><a href="/learn/gf96-ninety-six-bits/">19 GF96: ninety-six bits</a></li><li><a href="/learn/gf128-a-quad/">20 GF128: a quad</a></li><li><a href="/learn/gf256-two-five-six/">21 GF256: two hundred fifty-six bits</a></li></ol></li><li><h3>Module 8 · The widest rungs, then trits</h3><p>GF512 and GF1024, the two widest rungs, then GF-T8, where the exponent moves to trits.</p><ol><li><a href="/learn/gf512-five-twelve/">22 GF512: five hundred twelve bits</a></li><li><a href="/learn/gf1024-one-kilobit/">23 GF1024: one kilobit</a></li><li><a href="/learn/gft8-exponent-in-trits/">24 GF-T8: an exponent in trits</a></li></ol></li><li><h3>Module 9 · More trits, then the decode</h3><p>GF-T16 and GF-T32, then why fixed fields decode in parallel and a posit does not.</p><ol><li><a href="/learn/gft16-six-trits/">25 GF-T16: six trits</a></li><li><a href="/learn/gft32-twelve-trits/">26 GF-T32: twelve trits</a></li><li class="on" aria-current="page"><a href="/learn/decode-in-one-step/">27 Decode in one step</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/depth-sizing/index.html b/apps/website/public/learn/depth-sizing/index.html
index 969af576b6..799e5e73cc 100644
--- a/apps/website/public/learn/depth-sizing/index.html
+++ b/apps/website/public/learn/depth-sizing/index.html
@@ -434,5 +434,6 @@ <h1>Sizing the depth</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a clock is</h3><p>One edge, one world: what shares a clock edge shares a world, the period and the jitter of a real edge, and where the clock enters a board.</p><ol><li><a href="/learn/one-edge-one-world/">01 One edge, one world</a></li><li><a href="/learn/period-and-jitter/">02 A period, an edge, and jitter</a></li><li><a href="/learn/where-the-clock-enters/">03 Where the clock enters</a></li></ol></li><li><h3>Module 2 · Clock trees</h3><p>Skew and insertion delay, the global buffer network, and the trap of gating a clock with logic.</p><ol><li><a href="/learn/skew-and-insertion/">04 Skew and insertion delay</a></li><li><a href="/learn/global-buffers/">05 The global buffer</a></li><li><a href="/learn/gated-clock-trap/">06 The gated-clock trap</a></li></ol></li><li><h3>Module 3 · PLL and MMCM</h3><p>Multiply and divide one clock into another, move its phase in steps of the VCO, and which clocks the analyzer treats as related.</p><ol><li><a href="/learn/multiply-and-divide/">07 Multiply and divide</a></li><li><a href="/learn/phase-and-lock/">08 Phase and lock</a></li><li><a href="/learn/related-or-not/">09 Related or not</a></li></ol></li><li><h3>Module 4 · Resets</h3><p>Assert asynchronously, release synchronously: the three reset kinds, the release pipe, and the tree a reset grows.</p><ol><li><a href="/learn/sync-or-async/">10 Sync or async</a></li><li><a href="/learn/reset-release/">11 The release pipe</a></li><li><a href="/learn/reset-fanout/">12 Reset fanout</a></li></ol></li><li><h3>Module 5 · Metastability</h3><p>The setup-hold window, the mean time between failures in integer arithmetic, and the two flops that fix it.</p><ol><li><a href="/learn/setup-hold-window/">13 The setup-hold window</a></li><li><a href="/learn/mtbf/">14 Mean time between failures</a></li><li><a href="/learn/two-flop-sync/">15 The two-flop synchronizer</a></li></ol></li><li><h3>Module 6 · Crossing many bits</h3><p>Why a binary bus tears, why Gray code does not, and the handshake that moves a pulse between worlds.</p><ol><li><a href="/learn/why-binary-fails/">16 Why binary fails</a></li><li><a href="/learn/gray-code/">17 Gray code</a></li><li><a href="/learn/handshakes-and-pulses/">18 Handshakes and pulses</a></li></ol></li><li><h3>Module 7 · The asynchronous FIFO</h3><p>Pointers, flags and depth: the buffer that moves a stream between two clocks.</p><ol><li><a href="/learn/pointers-and-flags/">19 Pointers and flags</a></li><li class="on" aria-current="page"><a href="/learn/depth-sizing/">20 Sizing the depth</a></li><li><a href="/learn/fifo-testbench/">21 The FIFO testbench</a></li></ol></li><li><h3>Module 8 · Constraints</h3><p>The lines that tell the analyzer what a clock is, which paths not to check, and what the pins must meet.</p><ol><li><a href="/learn/create-clock/">22 create_clock</a></li><li><a href="/learn/false-path-and-max-delay/">23 False paths and max delay</a></li><li><a href="/learn/io-timing/">24 IO timing</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>A CDC report, one crossing captured at the flip-flops, and the bitstream diff that closes the course.</p><ol><li><a href="/learn/cdc-report/">25 The CDC report</a></li><li><a href="/learn/capture-a-crossing/">26 Capture a crossing</a></li><li><a href="/learn/capstone-diff/">27 Capstone: change it and diff it</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/designing-taps/index.html b/apps/website/public/learn/designing-taps/index.html
index 3c570252d4..1df78a53e2 100644
--- a/apps/website/public/learn/designing-taps/index.html
+++ b/apps/website/public/learn/designing-taps/index.html
@@ -1122,5 +1122,6 @@ <h1>Designing taps</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Adding numbers</h3><p>Why addition is slow: the carry that walks up the word, the dedicated chain in every slice, prefix networks and carry-save trees.</p><ol><li><a href="/learn/the-carry-chain/">01 The carry chain</a></li><li><a href="/learn/prefix-adders/">02 Looking ahead: prefix adders</a></li><li><a href="/learn/carry-save/">03 Three numbers at once</a></li></ol></li><li><h3>Module 2 · Multiplying</h3><p>A product is a sum of shifted copies; Booth recoding halves them, and the DSP48E1 slice does the rest in one block.</p><ol><li><a href="/learn/shift-and-add/">04 Shift and add</a></li><li><a href="/learn/booths-trick/">05 Booth's trick</a></li><li><a href="/learn/the-dsp48e1-slice/">06 The DSP48E1 slice</a></li></ol></li><li><h3>Module 3 · Fixed point</h3><p>Where the binary point sits, what rounding does to a value and to its average, and what each bit of a quantizer buys.</p><ol><li><a href="/learn/q-formats/">07 Q formats</a></li><li><a href="/learn/rounding-and-saturation/">08 Rounding and saturation</a></li><li><a href="/learn/quantization-noise/">09 Quantization noise</a></li></ol></li><li><h3>Module 4 · Functions in hardware</h3><p>Sine, cosine, angle and length from shifts and adds, and when a table is the better answer.</p><ol><li><a href="/learn/rotating-by-shifts/">10 Rotating by shifts</a></li><li><a href="/learn/angle-and-magnitude/">11 Angle and magnitude</a></li><li><a href="/learn/tables-and-interpolation/">12 Tables and interpolation</a></li></ol></li><li><h3>Module 5 · Signals and sampling</h3><p>What sampling does to a frequency, how an oscillator is built from an adder, and what a DFT bin measures.</p><ol><li><a href="/learn/sampling-and-aliasing/">13 Sampling and aliasing</a></li><li><a href="/learn/the-nco/">14 A numerically controlled oscillator</a></li><li><a href="/learn/bins-of-a-spectrum/">15 Bins of a spectrum</a></li></ol></li><li><h3>Module 6 · FIR filters</h3><p>The moving average, a windowed-sinc design with integer taps, and folding the taps onto DSP slices.</p><ol><li><a href="/learn/the-moving-average/">16 The moving average</a></li><li class="on" aria-current="page"><a href="/learn/designing-taps/">17 Designing taps</a></li><li><a href="/learn/folding-taps-onto-slices/">18 Folding taps onto slices</a></li></ol></li><li><h3>Module 7 · Multirate</h3><p>Lowering the sample rate without folding noise in: decimation, the CIC filter and the polyphase form.</p><ol><li><a href="/learn/lowering-the-rate/">19 Lowering the rate</a></li><li><a href="/learn/the-cic-filter/">20 The CIC filter</a></li><li><a href="/learn/polyphase/">21 Polyphase</a></li></ol></li><li><h3>Module 8 · The FFT</h3><p>N log N instead of N^2: butterflies, the bit-reversed input order and the bits each stage adds.</p><ol><li><a href="/learn/the-butterfly/">22 The butterfly</a></li><li><a href="/learn/twiddles-and-bit-reversal/">23 Twiddles and bit reversal</a></li><li><a href="/learn/growth-through-the-stages/">24 Growth through the stages</a></li></ol></li><li><h3>Module 9 · On the bench</h3><p>From the arithmetic to the board: a correlator from the modem spec, a budget of slices and timing, and the capstone filter.</p><ol><li><a href="/learn/a-correlator-finds-the-frame/">25 A correlator finds the frame</a></li><li><a href="/learn/does-it-fit-and-close-timing/">26 Does it fit and close timing?</a></li><li><a href="/learn/capstone-a-filter-on-the-board/">27 Capstone: a filter on the board</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/directed-vs-random/index.html b/apps/website/public/learn/directed-vs-random/index.html
index 6906f4c76c..5fdb4bae0b 100644
--- a/apps/website/public/learn/directed-vs-random/index.html
+++ b/apps/website/public/learn/directed-vs-random/index.html
@@ -466,5 +466,6 @@ <h1>Directed versus random</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Why verify</h3><p>Designs that compile and are wrong, the model that decides, and the plan written before the code.</p><ol><li><a href="/learn/bugs-that-compile/">01 Bugs that compile</a></li><li><a href="/learn/golden-model/">02 The golden model</a></li><li><a href="/learn/plan-before-code/">03 Plan before code</a></li></ol></li><li><h3>Module 2 · Testbenches</h3><p>Stimulus, checks and a verdict, written as one spec beside the design it judges.</p><ol><li><a href="/learn/anatomy-of-a-tb/">04 Anatomy of a testbench</a></li><li><a href="/learn/self-checking/">05 Self-checking, not staring</a></li><li class="on" aria-current="page"><a href="/learn/directed-vs-random/">06 Directed versus random</a></li></ol></li><li><h3>Module 3 · Waveforms</h3><p>A trace of every signal, read the way a hardware engineer reads it, and two runs compared.</p><ol><li><a href="/learn/vcd-format/">07 The VCD format</a></li><li><a href="/learn/debug-with-waves/">08 Debug with waves</a></li><li><a href="/learn/compare-traces/">09 Compare two traces</a></li></ol></li><li><h3>Module 4 · Conformance vectors</h3><p>Cases with the answer written beside them, kept where the compiler can reach them.</p><ol><li><a href="/learn/what-a-vector-is/">10 What a vector is</a></li><li><a href="/learn/golden-ruler-vectors/">11 Golden ruler vectors</a></li><li><a href="/learn/end-to-end-scenarios/">12 End-to-end scenarios</a></li></ol></li><li><h3>Module 5 · Cosimulation</h3><p>Spec, simulator and board agreeing on the bench Artix-7 XC7A200T, and what to do when they do not.</p><ol><li><a href="/learn/three-way-agreement/">13 Three-way agreement</a></li><li><a href="/learn/simulator-internals/">14 Inside the simulator</a></li><li><a href="/learn/when-they-disagree/">15 When they disagree</a></li></ol></li><li><h3>Module 6 · Coverage</h3><p>What the tests touched: lines, toggles, states, and what that number hides.</p><ol><li><a href="/learn/line-and-toggle/">16 Line and toggle</a></li><li><a href="/learn/fsm-coverage/">17 FSM coverage</a></li><li><a href="/learn/coverage-lies/">18 Coverage lies</a></li></ol></li><li><h3>Module 7 · Formal</h3><p>Assertions that hold every cycle, bounded search for a counterexample, and why a proof needs induction.</p><ol><li><a href="/learn/assertions/">19 Assertions</a></li><li><a href="/learn/bounded-model-checking/">20 Bounded model checking</a></li><li><a href="/learn/induction/">21 Induction</a></li></ol></li><li><h3>Module 8 · Mutation</h3><p>Break the design on purpose and count what the tests catch.</p><ol><li><a href="/learn/kill-the-mutant/">22 Kill the mutant</a></li><li><a href="/learn/surviving-mutants/">23 Surviving mutants</a></li><li><a href="/learn/test-strength/">24 Test strength</a></li></ol></li><li><h3>Module 9 · Sign-off</h3><p>One command, every receipt, a clean verdict you can show.</p><ol><li><a href="/learn/build-verify/">25 Build and verify</a></li><li><a href="/learn/board-selftest/">26 The board self-tests</a></li><li><a href="/learn/verify-capstone/">27 Capstone: verify a module whole</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/does-it-fit-and-close-timing/index.html b/apps/website/public/learn/does-it-fit-and-close-timing/index.html
index 621e1200e8..eabdc6705a 100644
--- a/apps/website/public/learn/does-it-fit-and-close-timing/index.html
+++ b/apps/website/public/learn/does-it-fit-and-close-timing/index.html
@@ -1122,5 +1122,6 @@ <h1>Does it fit and close timing?</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Adding numbers</h3><p>Why addition is slow: the carry that walks up the word, the dedicated chain in every slice, prefix networks and carry-save trees.</p><ol><li><a href="/learn/the-carry-chain/">01 The carry chain</a></li><li><a href="/learn/prefix-adders/">02 Looking ahead: prefix adders</a></li><li><a href="/learn/carry-save/">03 Three numbers at once</a></li></ol></li><li><h3>Module 2 · Multiplying</h3><p>A product is a sum of shifted copies; Booth recoding halves them, and the DSP48E1 slice does the rest in one block.</p><ol><li><a href="/learn/shift-and-add/">04 Shift and add</a></li><li><a href="/learn/booths-trick/">05 Booth's trick</a></li><li><a href="/learn/the-dsp48e1-slice/">06 The DSP48E1 slice</a></li></ol></li><li><h3>Module 3 · Fixed point</h3><p>Where the binary point sits, what rounding does to a value and to its average, and what each bit of a quantizer buys.</p><ol><li><a href="/learn/q-formats/">07 Q formats</a></li><li><a href="/learn/rounding-and-saturation/">08 Rounding and saturation</a></li><li><a href="/learn/quantization-noise/">09 Quantization noise</a></li></ol></li><li><h3>Module 4 · Functions in hardware</h3><p>Sine, cosine, angle and length from shifts and adds, and when a table is the better answer.</p><ol><li><a href="/learn/rotating-by-shifts/">10 Rotating by shifts</a></li><li><a href="/learn/angle-and-magnitude/">11 Angle and magnitude</a></li><li><a href="/learn/tables-and-interpolation/">12 Tables and interpolation</a></li></ol></li><li><h3>Module 5 · Signals and sampling</h3><p>What sampling does to a frequency, how an oscillator is built from an adder, and what a DFT bin measures.</p><ol><li><a href="/learn/sampling-and-aliasing/">13 Sampling and aliasing</a></li><li><a href="/learn/the-nco/">14 A numerically controlled oscillator</a></li><li><a href="/learn/bins-of-a-spectrum/">15 Bins of a spectrum</a></li></ol></li><li><h3>Module 6 · FIR filters</h3><p>The moving average, a windowed-sinc design with integer taps, and folding the taps onto DSP slices.</p><ol><li><a href="/learn/the-moving-average/">16 The moving average</a></li><li><a href="/learn/designing-taps/">17 Designing taps</a></li><li><a href="/learn/folding-taps-onto-slices/">18 Folding taps onto slices</a></li></ol></li><li><h3>Module 7 · Multirate</h3><p>Lowering the sample rate without folding noise in: decimation, the CIC filter and the polyphase form.</p><ol><li><a href="/learn/lowering-the-rate/">19 Lowering the rate</a></li><li><a href="/learn/the-cic-filter/">20 The CIC filter</a></li><li><a href="/learn/polyphase/">21 Polyphase</a></li></ol></li><li><h3>Module 8 · The FFT</h3><p>N log N instead of N^2: butterflies, the bit-reversed input order and the bits each stage adds.</p><ol><li><a href="/learn/the-butterfly/">22 The butterfly</a></li><li><a href="/learn/twiddles-and-bit-reversal/">23 Twiddles and bit reversal</a></li><li><a href="/learn/growth-through-the-stages/">24 Growth through the stages</a></li></ol></li><li><h3>Module 9 · On the bench</h3><p>From the arithmetic to the board: a correlator from the modem spec, a budget of slices and timing, and the capstone filter.</p><ol><li><a href="/learn/a-correlator-finds-the-frame/">25 A correlator finds the frame</a></li><li class="on" aria-current="page"><a href="/learn/does-it-fit-and-close-timing/">26 Does it fit and close timing?</a></li><li><a href="/learn/capstone-a-filter-on-the-board/">27 Capstone: a filter on the board</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/end-to-end-scenarios/index.html b/apps/website/public/learn/end-to-end-scenarios/index.html
index 7869b4fbdf..527163a184 100644
--- a/apps/website/public/learn/end-to-end-scenarios/index.html
+++ b/apps/website/public/learn/end-to-end-scenarios/index.html
@@ -322,5 +322,6 @@ <h1>End-to-end scenarios</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Why verify</h3><p>Designs that compile and are wrong, the model that decides, and the plan written before the code.</p><ol><li><a href="/learn/bugs-that-compile/">01 Bugs that compile</a></li><li><a href="/learn/golden-model/">02 The golden model</a></li><li><a href="/learn/plan-before-code/">03 Plan before code</a></li></ol></li><li><h3>Module 2 · Testbenches</h3><p>Stimulus, checks and a verdict, written as one spec beside the design it judges.</p><ol><li><a href="/learn/anatomy-of-a-tb/">04 Anatomy of a testbench</a></li><li><a href="/learn/self-checking/">05 Self-checking, not staring</a></li><li><a href="/learn/directed-vs-random/">06 Directed versus random</a></li></ol></li><li><h3>Module 3 · Waveforms</h3><p>A trace of every signal, read the way a hardware engineer reads it, and two runs compared.</p><ol><li><a href="/learn/vcd-format/">07 The VCD format</a></li><li><a href="/learn/debug-with-waves/">08 Debug with waves</a></li><li><a href="/learn/compare-traces/">09 Compare two traces</a></li></ol></li><li><h3>Module 4 · Conformance vectors</h3><p>Cases with the answer written beside them, kept where the compiler can reach them.</p><ol><li><a href="/learn/what-a-vector-is/">10 What a vector is</a></li><li><a href="/learn/golden-ruler-vectors/">11 Golden ruler vectors</a></li><li class="on" aria-current="page"><a href="/learn/end-to-end-scenarios/">12 End-to-end scenarios</a></li></ol></li><li><h3>Module 5 · Cosimulation</h3><p>Spec, simulator and board agreeing on the bench Artix-7 XC7A200T, and what to do when they do not.</p><ol><li><a href="/learn/three-way-agreement/">13 Three-way agreement</a></li><li><a href="/learn/simulator-internals/">14 Inside the simulator</a></li><li><a href="/learn/when-they-disagree/">15 When they disagree</a></li></ol></li><li><h3>Module 6 · Coverage</h3><p>What the tests touched: lines, toggles, states, and what that number hides.</p><ol><li><a href="/learn/line-and-toggle/">16 Line and toggle</a></li><li><a href="/learn/fsm-coverage/">17 FSM coverage</a></li><li><a href="/learn/coverage-lies/">18 Coverage lies</a></li></ol></li><li><h3>Module 7 · Formal</h3><p>Assertions that hold every cycle, bounded search for a counterexample, and why a proof needs induction.</p><ol><li><a href="/learn/assertions/">19 Assertions</a></li><li><a href="/learn/bounded-model-checking/">20 Bounded model checking</a></li><li><a href="/learn/induction/">21 Induction</a></li></ol></li><li><h3>Module 8 · Mutation</h3><p>Break the design on purpose and count what the tests catch.</p><ol><li><a href="/learn/kill-the-mutant/">22 Kill the mutant</a></li><li><a href="/learn/surviving-mutants/">23 Surviving mutants</a></li><li><a href="/learn/test-strength/">24 Test strength</a></li></ol></li><li><h3>Module 9 · Sign-off</h3><p>One command, every receipt, a clean verdict you can show.</p><ol><li><a href="/learn/build-verify/">25 Build and verify</a></li><li><a href="/learn/board-selftest/">26 The board self-tests</a></li><li><a href="/learn/verify-capstone/">27 Capstone: verify a module whole</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/enums/index.html b/apps/website/public/learn/enums/index.html
index dd782ac313..17cf3edc1c 100644
--- a/apps/website/public/learn/enums/index.html
+++ b/apps/website/public/learn/enums/index.html
@@ -92,5 +92,6 @@ <h1>Enums</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · A spec file</h3><p>What a .t27 file holds: a module line, prose lines and comments.</p><ol><li><a href="/learn/what-a-spec-is/">01 What a spec is</a></li><li><a href="/learn/the-module-line/">02 The module line</a></li><li><a href="/learn/comments-and-prose/">03 Comments and prose lines</a></li></ol></li><li><h3>Module 2 · Values and types</h3><p>Constants, the width of a whole number, and true, false and text.</p><ol><li><a href="/learn/constants/">04 Constants</a></li><li><a href="/learn/whole-numbers/">05 Whole numbers and their widths</a></li><li><a href="/learn/true-false-and-text/">06 True, false and text</a></li></ol></li><li><h3>Module 3 · Arrays, trits and expressions</h3><p>Lists of values, the three values of a trit, and what operators compute.</p><ol><li><a href="/learn/arrays/">07 Arrays</a></li><li><a href="/learn/trits/">08 Trits: minus one, zero, one</a></li><li><a href="/learn/expressions/">09 Expressions</a></li></ol></li><li><h3>Module 4 · Tests and functions</h3><p>A test block, several small tests in one spec, and a function with typed inputs.</p><ol><li><a href="/learn/a-test-block/">10 A test block</a></li><li><a href="/learn/many-tests/">11 Many tests in one spec</a></li><li><a href="/learn/functions/">12 Functions</a></li></ol></li><li><h3>Module 5 · Inside a function</h3><p>Names with let and var, choices with if and switch, and loops.</p><ol><li><a href="/learn/local-names/">13 Local names: let and var</a></li><li><a href="/learn/if-and-else/">14 Choices: if, else and switch</a></li><li><a href="/learn/loops/">15 Loops: while and for</a></li></ol></li><li><h3>Module 6 · Visibility and your own types</h3><p>What pub shares with other modules, and the structs and enums you declare.</p><ol><li><a href="/learn/pub-or-private/">16 pub or private</a></li><li><a href="/learn/structs/">17 Structs</a></li><li class="on" aria-current="page"><a href="/learn/enums/">18 Enums</a></li></ol></li><li><h3>Module 7 · Modules, rules and gen-ts</h3><p>Reading another module, stating a rule that always holds, and a spec turned into TypeScript.</p><ol><li><a href="/learn/use-other-modules/">19 Using other modules</a></li><li><a href="/learn/invariants/">20 Invariants and benches</a></li><li><a href="/learn/gen-ts/">21 gen-ts: a spec becomes TypeScript</a></li></ol></li><li><h3>Module 8 · The compiler and tri</h3><p>The backends t27c writes, the stages it reads a spec in, and the tri command.</p><ol><li><a href="/learn/seven-backends/">22 Seven backends</a></li><li><a href="/learn/stages-of-t27c/">23 The stages of t27c</a></li><li><a href="/learn/tri-commands/">24 tri: the command line</a></li></ol></li><li><h3>Module 9 · Errors, a program, what next</h3><p>Reading a compiler error, one small complete program, and the course to take next.</p><ol><li><a href="/learn/reading-errors/">25 Reading compiler errors</a></li><li><a href="/learn/a-small-program/">26 A small complete program</a></li><li><a href="/learn/where-next/">27 Where to go next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/everything-in-one-bench/index.html b/apps/website/public/learn/everything-in-one-bench/index.html
index 69921db45f..889697c279 100644
--- a/apps/website/public/learn/everything-in-one-bench/index.html
+++ b/apps/website/public/learn/everything-in-one-bench/index.html
@@ -202,5 +202,6 @@ <h1>Everything in one bench</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a bus is</h3><p>Why a bus exists at all: a conversation on wires, framed and addressed, and who is allowed to talk.</p><ol><li><a href="/learn/wires-that-talk/">01 Wires that talk</a></li><li><a href="/learn/framing-a-conversation/">02 Framing a conversation</a></li><li><a href="/learn/master-and-servant/">03 Master and servant</a></li></ol></li><li><h3>Module 2 · UART</h3><p>The two-wire bus with no clock: the frame, the divisor that sets the rate, and the status a driver polls.</p><ol><li><a href="/learn/the-frame/">04 The frame: start, data, stop</a></li><li><a href="/learn/baud-and-the-divisor/">05 Baud and the divisor</a></li><li><a href="/learn/status-and-fifos/">06 Status and FIFOs</a></li></ol></li><li><h3>Module 3 · SPI</h3><p>The clocked conversation: four modes, a prescaler ladder, and a chip select per servant.</p><ol><li><a href="/learn/modes-cpol-cpha/">07 Modes: CPOL and CPHA</a></li><li><a href="/learn/the-prescaler-ladder/">08 The prescaler ladder</a></li><li><a href="/learn/chip-select-and-widths/">09 Chip select and widths</a></li></ol></li><li><h3>Module 4 · APB</h3><p>The register bus: PSEL and PENABLE, strobes and wait states, and how many address bits a peripheral count costs.</p><ol><li><a href="/learn/psel-and-penable/">10 PSEL and PENABLE</a></li><li><a href="/learn/strobes-and-waits/">11 Strobes and waits</a></li><li><a href="/learn/address-decode/">12 Address decode</a></li></ol></li><li><h3>Module 5 · AXI4</h3><p>The five channels: address, data and response in both directions, lite or full, bursts and IDs.</p><ol><li><a href="/learn/five-channels/">13 Five channels</a></li><li><a href="/learn/lite-or-full/">14 Lite or full</a></li><li><a href="/learn/bursts-and-ids/">15 Bursts and IDs</a></li></ol></li><li><h3>Module 6 · Memory</h3><p>What sits on the far side of every bus: memory maps, port kinds, and latency that a wait state must cover.</p><ol><li><a href="/learn/memory-maps/">16 Memory maps</a></li><li><a href="/learn/ports-and-kinds/">17 Ports and kinds</a></li><li><a href="/learn/latency/">18 Latency</a></li></ol></li><li><h3>Module 7 · Bridges</h3><p>Why designs grow more than one bus, and the packet bridge that moves work between them.</p><ol><li><a href="/learn/why-bridges/">19 Why bridges</a></li><li><a href="/learn/the-packet-bridge/">20 The packet bridge</a></li><li class="on" aria-current="page"><a href="/learn/everything-in-one-bench/">21 Everything in one bench</a></li></ol></li><li><h3>Module 8 · Ethernet</h3><p>Frames, the frame check sequence, RGMII timing, and the pre-registered steps of a real bring-up.</p><ol><li><a href="/learn/frames-and-the-fcs/">22 Frames and the FCS</a></li><li><a href="/learn/rgmii-timing/">23 RGMII timing</a></li><li><a href="/learn/bring-up-on-the-bench/">24 Bring-up on the bench</a></li></ol></li><li><h3>Module 9 · The bench</h3><p>The discipline that guards real hardware: who holds the IO, taking and giving the lock, and what runs next.</p><ol><li><a href="/learn/who-holds-the-io/">25 Who holds the IO</a></li><li><a href="/learn/claim-and-release/">26 Claim and release</a></li><li><a href="/learn/capstone-what-runs-next/">27 Capstone: what runs next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/expressions/index.html b/apps/website/public/learn/expressions/index.html
index 4f320dc75d..8a0b61467c 100644
--- a/apps/website/public/learn/expressions/index.html
+++ b/apps/website/public/learn/expressions/index.html
@@ -92,5 +92,6 @@ <h1>Expressions</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · A spec file</h3><p>What a .t27 file holds: a module line, prose lines and comments.</p><ol><li><a href="/learn/what-a-spec-is/">01 What a spec is</a></li><li><a href="/learn/the-module-line/">02 The module line</a></li><li><a href="/learn/comments-and-prose/">03 Comments and prose lines</a></li></ol></li><li><h3>Module 2 · Values and types</h3><p>Constants, the width of a whole number, and true, false and text.</p><ol><li><a href="/learn/constants/">04 Constants</a></li><li><a href="/learn/whole-numbers/">05 Whole numbers and their widths</a></li><li><a href="/learn/true-false-and-text/">06 True, false and text</a></li></ol></li><li><h3>Module 3 · Arrays, trits and expressions</h3><p>Lists of values, the three values of a trit, and what operators compute.</p><ol><li><a href="/learn/arrays/">07 Arrays</a></li><li><a href="/learn/trits/">08 Trits: minus one, zero, one</a></li><li class="on" aria-current="page"><a href="/learn/expressions/">09 Expressions</a></li></ol></li><li><h3>Module 4 · Tests and functions</h3><p>A test block, several small tests in one spec, and a function with typed inputs.</p><ol><li><a href="/learn/a-test-block/">10 A test block</a></li><li><a href="/learn/many-tests/">11 Many tests in one spec</a></li><li><a href="/learn/functions/">12 Functions</a></li></ol></li><li><h3>Module 5 · Inside a function</h3><p>Names with let and var, choices with if and switch, and loops.</p><ol><li><a href="/learn/local-names/">13 Local names: let and var</a></li><li><a href="/learn/if-and-else/">14 Choices: if, else and switch</a></li><li><a href="/learn/loops/">15 Loops: while and for</a></li></ol></li><li><h3>Module 6 · Visibility and your own types</h3><p>What pub shares with other modules, and the structs and enums you declare.</p><ol><li><a href="/learn/pub-or-private/">16 pub or private</a></li><li><a href="/learn/structs/">17 Structs</a></li><li><a href="/learn/enums/">18 Enums</a></li></ol></li><li><h3>Module 7 · Modules, rules and gen-ts</h3><p>Reading another module, stating a rule that always holds, and a spec turned into TypeScript.</p><ol><li><a href="/learn/use-other-modules/">19 Using other modules</a></li><li><a href="/learn/invariants/">20 Invariants and benches</a></li><li><a href="/learn/gen-ts/">21 gen-ts: a spec becomes TypeScript</a></li></ol></li><li><h3>Module 8 · The compiler and tri</h3><p>The backends t27c writes, the stages it reads a spec in, and the tri command.</p><ol><li><a href="/learn/seven-backends/">22 Seven backends</a></li><li><a href="/learn/stages-of-t27c/">23 The stages of t27c</a></li><li><a href="/learn/tri-commands/">24 tri: the command line</a></li></ol></li><li><h3>Module 9 · Errors, a program, what next</h3><p>Reading a compiler error, one small complete program, and the course to take next.</p><ol><li><a href="/learn/reading-errors/">25 Reading compiler errors</a></li><li><a href="/learn/a-small-program/">26 A small complete program</a></li><li><a href="/learn/where-next/">27 Where to go next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/false-path-and-max-delay/index.html b/apps/website/public/learn/false-path-and-max-delay/index.html
index 128fd2d176..79cc55b72e 100644
--- a/apps/website/public/learn/false-path-and-max-delay/index.html
+++ b/apps/website/public/learn/false-path-and-max-delay/index.html
@@ -448,5 +448,6 @@ <h1>False paths and max delay</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a clock is</h3><p>One edge, one world: what shares a clock edge shares a world, the period and the jitter of a real edge, and where the clock enters a board.</p><ol><li><a href="/learn/one-edge-one-world/">01 One edge, one world</a></li><li><a href="/learn/period-and-jitter/">02 A period, an edge, and jitter</a></li><li><a href="/learn/where-the-clock-enters/">03 Where the clock enters</a></li></ol></li><li><h3>Module 2 · Clock trees</h3><p>Skew and insertion delay, the global buffer network, and the trap of gating a clock with logic.</p><ol><li><a href="/learn/skew-and-insertion/">04 Skew and insertion delay</a></li><li><a href="/learn/global-buffers/">05 The global buffer</a></li><li><a href="/learn/gated-clock-trap/">06 The gated-clock trap</a></li></ol></li><li><h3>Module 3 · PLL and MMCM</h3><p>Multiply and divide one clock into another, move its phase in steps of the VCO, and which clocks the analyzer treats as related.</p><ol><li><a href="/learn/multiply-and-divide/">07 Multiply and divide</a></li><li><a href="/learn/phase-and-lock/">08 Phase and lock</a></li><li><a href="/learn/related-or-not/">09 Related or not</a></li></ol></li><li><h3>Module 4 · Resets</h3><p>Assert asynchronously, release synchronously: the three reset kinds, the release pipe, and the tree a reset grows.</p><ol><li><a href="/learn/sync-or-async/">10 Sync or async</a></li><li><a href="/learn/reset-release/">11 The release pipe</a></li><li><a href="/learn/reset-fanout/">12 Reset fanout</a></li></ol></li><li><h3>Module 5 · Metastability</h3><p>The setup-hold window, the mean time between failures in integer arithmetic, and the two flops that fix it.</p><ol><li><a href="/learn/setup-hold-window/">13 The setup-hold window</a></li><li><a href="/learn/mtbf/">14 Mean time between failures</a></li><li><a href="/learn/two-flop-sync/">15 The two-flop synchronizer</a></li></ol></li><li><h3>Module 6 · Crossing many bits</h3><p>Why a binary bus tears, why Gray code does not, and the handshake that moves a pulse between worlds.</p><ol><li><a href="/learn/why-binary-fails/">16 Why binary fails</a></li><li><a href="/learn/gray-code/">17 Gray code</a></li><li><a href="/learn/handshakes-and-pulses/">18 Handshakes and pulses</a></li></ol></li><li><h3>Module 7 · The asynchronous FIFO</h3><p>Pointers, flags and depth: the buffer that moves a stream between two clocks.</p><ol><li><a href="/learn/pointers-and-flags/">19 Pointers and flags</a></li><li><a href="/learn/depth-sizing/">20 Sizing the depth</a></li><li><a href="/learn/fifo-testbench/">21 The FIFO testbench</a></li></ol></li><li><h3>Module 8 · Constraints</h3><p>The lines that tell the analyzer what a clock is, which paths not to check, and what the pins must meet.</p><ol><li><a href="/learn/create-clock/">22 create_clock</a></li><li class="on" aria-current="page"><a href="/learn/false-path-and-max-delay/">23 False paths and max delay</a></li><li><a href="/learn/io-timing/">24 IO timing</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>A CDC report, one crossing captured at the flip-flops, and the bitstream diff that closes the course.</p><ol><li><a href="/learn/cdc-report/">25 The CDC report</a></li><li><a href="/learn/capture-a-crossing/">26 Capture a crossing</a></li><li><a href="/learn/capstone-diff/">27 Capstone: change it and diff it</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/fasm-to-frames/index.html b/apps/website/public/learn/fasm-to-frames/index.html
index 9d62519d04..54e1c254f7 100644
--- a/apps/website/public/learn/fasm-to-frames/index.html
+++ b/apps/website/public/learn/fasm-to-frames/index.html
@@ -110,5 +110,6 @@ <h1>From FASM to frames</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The chip</h3><p>What an FPGA is, which chip we use, and how its pins meet the board.</p><ol><li><a href="/learn/fpga-is-a-grid/">01 An FPGA is a grid you configure</a></li><li><a href="/learn/which-chip/">02 Which chip is this?</a></li><li><a href="/learn/pins-and-package/">03 Pins, balls and the package</a></li></ol></li><li><h3>Module 2 · Numbers in hardware</h3><p>Bits, trits and number formats, and what arithmetic costs in logic.</p><ol><li><a href="/learn/bits-and-trits/">04 Bits, trits and GF16</a></li><li><a href="/learn/what-arithmetic-costs/">05 What arithmetic costs</a></li><li><a href="/learn/how-much-fits/">06 How much fits on the chip</a></li></ol></li><li><h3>Module 3 · Your t27 program</h3><p>Write a spec, test it, and see why compiling is not the same as being right.</p><ol><li><a href="/learn/hello-t27/">07 Hello, t27</a></li><li><a href="/learn/tests-are-the-spec/">08 Tests are part of the spec</a></li><li><a href="/learn/compiles-is-not-correct/">09 Compiles is not the same as correct</a></li></ol></li><li><h3>Module 4 · Inside t27c</h3><p>How the compiler reads a spec and what it writes, including the native t27b.</p><ol><li><a href="/learn/compiler-stages/">10 The stages of t27c</a></li><li><a href="/learn/t27c-on-a-real-machine/">11 t27c on a real machine</a></li><li><a href="/learn/seven-backends-and-t27b/">12 Seven backends, and t27b</a></li></ol></li><li><h3>Module 5 · From spec to hardware</h3><p>The Verilog t27c writes, the cells it becomes, and what one LUT does.</p><ol><li><a href="/learn/spec-to-verilog/">13 From a spec to Verilog</a></li><li><a href="/learn/count-the-cells/">14 Count the cells</a></li><li><a href="/learn/one-lut-up-close/">15 One LUT, up close</a></li></ol></li><li><h3>Module 6 · Reading synthesis</h3><p>What yosys reports about your design, and which warnings matter.</p><ol><li><a href="/learn/synthesis-warnings/">16 Synthesis warnings that matter</a></li><li><a href="/learn/state-machines/">17 State machines</a></li><li><a href="/learn/where-the-luts-go/">18 Where the LUTs go</a></li></ol></li><li><h3>Module 7 · Place, route, timing</h3><p>Where the cells land on the die, and whether the clock is met.</p><ol><li><a href="/learn/placement/">19 Placement: where the cells land</a></li><li><a href="/learn/logic-depth/">20 Logic depth</a></li><li><a href="/learn/timing-slack/">21 Timing and slack</a></li></ol></li><li><h3>Module 8 · The bitstream</h3><p>How a routed design becomes the bits the chip loads, with no Vivado.</p><ol><li class="on" aria-current="page"><a href="/learn/fasm-to-frames/">22 From FASM to frames</a></li><li><a href="/learn/a-build-receipt/">23 A build, itemised</a></li><li><a href="/learn/read-a-bitstream/">24 Read a bitstream</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>Ask the chip who it is, load the bits, and check them.</p><ol><li><a href="/learn/jtag-asks-the-chip/">25 JTAG asks the chip</a></li><li><a href="/learn/same-bits-on-the-board/">26 The same bits, on the board</a></li><li><a href="/learn/capstone/">27 Capstone: out-guess yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/fifo-testbench/index.html b/apps/website/public/learn/fifo-testbench/index.html
index 8b7f9429ee..fc695debc3 100644
--- a/apps/website/public/learn/fifo-testbench/index.html
+++ b/apps/website/public/learn/fifo-testbench/index.html
@@ -434,5 +434,6 @@ <h1>The FIFO testbench</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a clock is</h3><p>One edge, one world: what shares a clock edge shares a world, the period and the jitter of a real edge, and where the clock enters a board.</p><ol><li><a href="/learn/one-edge-one-world/">01 One edge, one world</a></li><li><a href="/learn/period-and-jitter/">02 A period, an edge, and jitter</a></li><li><a href="/learn/where-the-clock-enters/">03 Where the clock enters</a></li></ol></li><li><h3>Module 2 · Clock trees</h3><p>Skew and insertion delay, the global buffer network, and the trap of gating a clock with logic.</p><ol><li><a href="/learn/skew-and-insertion/">04 Skew and insertion delay</a></li><li><a href="/learn/global-buffers/">05 The global buffer</a></li><li><a href="/learn/gated-clock-trap/">06 The gated-clock trap</a></li></ol></li><li><h3>Module 3 · PLL and MMCM</h3><p>Multiply and divide one clock into another, move its phase in steps of the VCO, and which clocks the analyzer treats as related.</p><ol><li><a href="/learn/multiply-and-divide/">07 Multiply and divide</a></li><li><a href="/learn/phase-and-lock/">08 Phase and lock</a></li><li><a href="/learn/related-or-not/">09 Related or not</a></li></ol></li><li><h3>Module 4 · Resets</h3><p>Assert asynchronously, release synchronously: the three reset kinds, the release pipe, and the tree a reset grows.</p><ol><li><a href="/learn/sync-or-async/">10 Sync or async</a></li><li><a href="/learn/reset-release/">11 The release pipe</a></li><li><a href="/learn/reset-fanout/">12 Reset fanout</a></li></ol></li><li><h3>Module 5 · Metastability</h3><p>The setup-hold window, the mean time between failures in integer arithmetic, and the two flops that fix it.</p><ol><li><a href="/learn/setup-hold-window/">13 The setup-hold window</a></li><li><a href="/learn/mtbf/">14 Mean time between failures</a></li><li><a href="/learn/two-flop-sync/">15 The two-flop synchronizer</a></li></ol></li><li><h3>Module 6 · Crossing many bits</h3><p>Why a binary bus tears, why Gray code does not, and the handshake that moves a pulse between worlds.</p><ol><li><a href="/learn/why-binary-fails/">16 Why binary fails</a></li><li><a href="/learn/gray-code/">17 Gray code</a></li><li><a href="/learn/handshakes-and-pulses/">18 Handshakes and pulses</a></li></ol></li><li><h3>Module 7 · The asynchronous FIFO</h3><p>Pointers, flags and depth: the buffer that moves a stream between two clocks.</p><ol><li><a href="/learn/pointers-and-flags/">19 Pointers and flags</a></li><li><a href="/learn/depth-sizing/">20 Sizing the depth</a></li><li class="on" aria-current="page"><a href="/learn/fifo-testbench/">21 The FIFO testbench</a></li></ol></li><li><h3>Module 8 · Constraints</h3><p>The lines that tell the analyzer what a clock is, which paths not to check, and what the pins must meet.</p><ol><li><a href="/learn/create-clock/">22 create_clock</a></li><li><a href="/learn/false-path-and-max-delay/">23 False paths and max delay</a></li><li><a href="/learn/io-timing/">24 IO timing</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>A CDC report, one crossing captured at the flip-flops, and the bitstream diff that closes the course.</p><ol><li><a href="/learn/cdc-report/">25 The CDC report</a></li><li><a href="/learn/capture-a-crossing/">26 Capture a crossing</a></li><li><a href="/learn/capstone-diff/">27 Capstone: change it and diff it</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/five-channels/index.html b/apps/website/public/learn/five-channels/index.html
index 18d71cde64..340a767897 100644
--- a/apps/website/public/learn/five-channels/index.html
+++ b/apps/website/public/learn/five-channels/index.html
@@ -463,5 +463,6 @@ <h1>Five channels</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a bus is</h3><p>Why a bus exists at all: a conversation on wires, framed and addressed, and who is allowed to talk.</p><ol><li><a href="/learn/wires-that-talk/">01 Wires that talk</a></li><li><a href="/learn/framing-a-conversation/">02 Framing a conversation</a></li><li><a href="/learn/master-and-servant/">03 Master and servant</a></li></ol></li><li><h3>Module 2 · UART</h3><p>The two-wire bus with no clock: the frame, the divisor that sets the rate, and the status a driver polls.</p><ol><li><a href="/learn/the-frame/">04 The frame: start, data, stop</a></li><li><a href="/learn/baud-and-the-divisor/">05 Baud and the divisor</a></li><li><a href="/learn/status-and-fifos/">06 Status and FIFOs</a></li></ol></li><li><h3>Module 3 · SPI</h3><p>The clocked conversation: four modes, a prescaler ladder, and a chip select per servant.</p><ol><li><a href="/learn/modes-cpol-cpha/">07 Modes: CPOL and CPHA</a></li><li><a href="/learn/the-prescaler-ladder/">08 The prescaler ladder</a></li><li><a href="/learn/chip-select-and-widths/">09 Chip select and widths</a></li></ol></li><li><h3>Module 4 · APB</h3><p>The register bus: PSEL and PENABLE, strobes and wait states, and how many address bits a peripheral count costs.</p><ol><li><a href="/learn/psel-and-penable/">10 PSEL and PENABLE</a></li><li><a href="/learn/strobes-and-waits/">11 Strobes and waits</a></li><li><a href="/learn/address-decode/">12 Address decode</a></li></ol></li><li><h3>Module 5 · AXI4</h3><p>The five channels: address, data and response in both directions, lite or full, bursts and IDs.</p><ol><li class="on" aria-current="page"><a href="/learn/five-channels/">13 Five channels</a></li><li><a href="/learn/lite-or-full/">14 Lite or full</a></li><li><a href="/learn/bursts-and-ids/">15 Bursts and IDs</a></li></ol></li><li><h3>Module 6 · Memory</h3><p>What sits on the far side of every bus: memory maps, port kinds, and latency that a wait state must cover.</p><ol><li><a href="/learn/memory-maps/">16 Memory maps</a></li><li><a href="/learn/ports-and-kinds/">17 Ports and kinds</a></li><li><a href="/learn/latency/">18 Latency</a></li></ol></li><li><h3>Module 7 · Bridges</h3><p>Why designs grow more than one bus, and the packet bridge that moves work between them.</p><ol><li><a href="/learn/why-bridges/">19 Why bridges</a></li><li><a href="/learn/the-packet-bridge/">20 The packet bridge</a></li><li><a href="/learn/everything-in-one-bench/">21 Everything in one bench</a></li></ol></li><li><h3>Module 8 · Ethernet</h3><p>Frames, the frame check sequence, RGMII timing, and the pre-registered steps of a real bring-up.</p><ol><li><a href="/learn/frames-and-the-fcs/">22 Frames and the FCS</a></li><li><a href="/learn/rgmii-timing/">23 RGMII timing</a></li><li><a href="/learn/bring-up-on-the-bench/">24 Bring-up on the bench</a></li></ol></li><li><h3>Module 9 · The bench</h3><p>The discipline that guards real hardware: who holds the IO, taking and giving the lock, and what runs next.</p><ol><li><a href="/learn/who-holds-the-io/">25 Who holds the IO</a></li><li><a href="/learn/claim-and-release/">26 Claim and release</a></li><li><a href="/learn/capstone-what-runs-next/">27 Capstone: what runs next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/five-rules-for-weights/index.html b/apps/website/public/learn/five-rules-for-weights/index.html
index a2b33ee2a2..5cc9f74640 100644
--- a/apps/website/public/learn/five-rules-for-weights/index.html
+++ b/apps/website/public/learn/five-rules-for-weights/index.html
@@ -117,5 +117,6 @@ <h1>Five rules for a weight alphabet</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Lab: our own research</h3><p>A number format of our own, an honest scoreboard, and a model's tables multiplied on the board.</p><ol><li><a href="/learn/t27-number-format/">01 T27: a number format of our own</a></li><li><a href="/learn/an-honest-scoreboard/">02 An honest scoreboard</a></li><li><a href="/learn/a-model-on-the-board/">03 A model's tables on the board</a></li></ol></li><li><h3>Module 2 · AI numbers: the MX block</h3><p>How AI chips keep weights in a few bits: one shared scale per block, the scale byte itself, and what one outlier does to its neighbours.</p><ol><li><a href="/learn/one-scale-per-block/">04 One scale for a block of weights</a></li><li><a href="/learn/scale-byte-native/">05 The scale byte on a real machine</a></li><li><a href="/learn/one-outlier/">06 One outlier, many zeros</a></li></ol></li><li><h3>Module 3 · Ternary weights</h3><p>Weights that are only minus, zero or plus a scale, the five rules a ternary alphabet must pass, and a test pass that checked nothing.</p><ol><li><a href="/learn/phi-scaled-ternary/">07 Weights of minus phi, zero and plus phi</a></li><li><a href="/learn/a-pass-that-checked-nothing/">08 A pass that checked nothing</a></li><li class="on" aria-current="page"><a href="/learn/five-rules-for-weights/">09 Five rules for a weight alphabet</a></li></ol></li><li><h3>Module 4 · The Ternary Network Float</h3><p>A rule the compiler enforces before any test runs, and a 17-bit float whose exponent is four balanced trits.</p><ol><li><a href="/learn/blocked-is-not-failed/">10 Blocked is not failed</a></li><li><a href="/learn/four-trit-exponent/">11 A float whose exponent is four trits</a></li><li><a href="/learn/negate-twice/">12 Negate twice, get the number back</a></li></ol></li><li><h3>Module 5 · Arithmetic on signed numbers</h3><p>Multiply two signed numbers, add them when their signs differ, and do both at once in a multiply-accumulate.</p><ol><li><a href="/learn/a-sign-is-one-bit/">13 A sign is one bit</a></li><li><a href="/learn/opposite-signs-subtract/">14 Opposite signs subtract</a></li><li><a href="/learn/multiply-then-add/">15 Multiply, then add</a></li></ol></li><li><h3>Module 6 · Parts of a neuron</h3><p>A ReLU that bends at zero, a power of two for softmax, and an argmax that names the answer.</p><ol><li><a href="/learn/a-bend-at-zero/">16 A bend at zero</a></li><li><a href="/learn/two-to-the-x/">17 Two to the x</a></li><li><a href="/learn/pick-the-largest/">18 Pick the largest</a></li></ol></li><li><h3>Module 7 · Learning from a mistake</h3><p>A loss that prices a wrong guess in bits, one step that moves a weight against its gradient, and the hidden layer that XOR needs.</p><ol><li><a href="/learn/a-loss-in-bits/">19 A loss in bits</a></li><li><a href="/learn/one-step-downhill/">20 One step downhill</a></li><li><a href="/learn/xor-needs-a-bend/">21 XOR needs a bend</a></li></ol></li><li><h3>Module 8 · BitNet: ternary networks</h3><p>A threshold that squeezes a sum back to three values, one neuron that becomes a different function when its weights change, and a neuron that reads its inputs 27 trits at a time.</p><ol><li><a href="/learn/back-to-three-values/">22 Back to three values</a></li><li><a href="/learn/same-neuron-new-weights/">23 Same neuron, new weights</a></li><li><a href="/learn/a-neuron-in-chunks/">24 A neuron in chunks</a></li></ol></li><li><h3>Module 9 · The ternary MAC as a chip</h3><p>The 27-trit dot product as wires with no register, the same sum added into a register on every clock, and a small whole network to close the course.</p><ol><li><a href="/learn/a-dot-product-in-wires/">25 A dot product made of wires</a></li><li><a href="/learn/add-it-up-every-clock/">26 Add it up on every clock</a></li><li><a href="/learn/a-whole-network/">27 A whole network</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/five-trits-a-byte/index.html b/apps/website/public/learn/five-trits-a-byte/index.html
index 946fac8ea0..496b601140 100644
--- a/apps/website/public/learn/five-trits-a-byte/index.html
+++ b/apps/website/public/learn/five-trits-a-byte/index.html
@@ -769,5 +769,6 @@ <h1>Five trits in a byte</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Three values</h3><p>Why three, how balanced ternary writes every number without a sign, and what flipping and cutting trits do.</p><ol><li><a href="/learn/why-three/">01 Why three</a></li><li><a href="/learn/balanced-ternary/">02 Balanced ternary</a></li><li><a href="/learn/flip-and-cut/">03 Flip to negate, cut to round</a></li></ol></li><li><h3>Module 2 · Logic with unknown</h3><p>Kleene's three-valued gates, two gates binary has no twin for, and addition as a pair of tables.</p><ol><li><a href="/learn/kleene-logic/">04 Three-valued logic</a></li><li><a href="/learn/consensus-and-any/">05 Consensus and any</a></li><li><a href="/learn/sum-and-carry/">06 Sum and carry</a></li></ol></li><li><h3>Module 3 · Adding trits</h3><p>The half adder, the full adder and a carry that ripples left, one place at a time.</p><ol><li><a href="/learn/half-adder/">07 The half adder</a></li><li><a href="/learn/full-adder/">08 The full adder</a></li><li><a href="/learn/ripple-carry/">09 Ripple carry</a></li></ol></li><li><h3>Module 4 · Multiplying without a multiplier</h3><p>Copy, drop or flip: a product by one trit, long multiplication, and a MAC that only adds.</p><ol><li><a href="/learn/times-a-trit/">10 Times one trit</a></li><li><a href="/learn/long-multiplication/">11 Long multiplication</a></li><li><a href="/learn/ternary-mac/">12 A ternary MAC</a></li></ol></li><li><h3>Module 5 · Trits in binary memory</h3><p>Two bits per trit, five trits per byte, and the bits a 27-trit word needs.</p><ol><li><a href="/learn/two-bits-per-trit/">13 Two bits per trit</a></li><li class="on" aria-current="page"><a href="/learn/five-trits-a-byte/">14 Five trits in a byte</a></li><li><a href="/learn/a-word-of-27/">15 A word of 27 trits</a></li></ol></li><li><h3>Module 6 · The TRI-27 instruction word</h3><p>The 32-bit word the Trinity emulator decodes: its fields, its 47 opcodes and its 15-bit immediate.</p><ol><li><a href="/learn/the-instruction-word/">16 The instruction word</a></li><li><a href="/learn/the-opcode-map/">17 The opcode map</a></li><li><a href="/learn/the-immediate/">18 The immediate</a></li></ol></li><li><h3>Module 7 · A ternary machine</h3><p>An ALU built from this course's adders, a three-way jump, and a program you can step.</p><ol><li><a href="/learn/a-ternary-alu/">19 A ternary ALU</a></li><li><a href="/learn/three-way-jump/">20 A three-way jump</a></li><li><a href="/learn/run-a-program/">21 Run a program</a></li></ol></li><li><h3>Module 8 · Three answers</h3><p>Compare at the highest differing trit, find a number in thirds, sort with three-way compares.</p><ol><li><a href="/learn/compare-at-the-top/">22 Compare at the top</a></li><li><a href="/learn/questions-in-thirds/">23 Questions in thirds</a></li><li><a href="/learn/sort-by-three/">24 Sort with three-way compares</a></li></ol></li><li><h3>Module 9 · Ternary neurons</h3><p>Weights of -1, 0, +1: one neuron, a detector and a small layer, with no multiplier anywhere.</p><ol><li><a href="/learn/one-ternary-neuron/">25 One ternary neuron</a></li><li><a href="/learn/a-bar-detector/">26 A bar detector</a></li><li><a href="/learn/a-tiny-layer/">27 A tiny layer</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/flip-and-cut/index.html b/apps/website/public/learn/flip-and-cut/index.html
index 727b14277b..db3c1a18ed 100644
--- a/apps/website/public/learn/flip-and-cut/index.html
+++ b/apps/website/public/learn/flip-and-cut/index.html
@@ -769,5 +769,6 @@ <h1>Flip to negate, cut to round</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Three values</h3><p>Why three, how balanced ternary writes every number without a sign, and what flipping and cutting trits do.</p><ol><li><a href="/learn/why-three/">01 Why three</a></li><li><a href="/learn/balanced-ternary/">02 Balanced ternary</a></li><li class="on" aria-current="page"><a href="/learn/flip-and-cut/">03 Flip to negate, cut to round</a></li></ol></li><li><h3>Module 2 · Logic with unknown</h3><p>Kleene's three-valued gates, two gates binary has no twin for, and addition as a pair of tables.</p><ol><li><a href="/learn/kleene-logic/">04 Three-valued logic</a></li><li><a href="/learn/consensus-and-any/">05 Consensus and any</a></li><li><a href="/learn/sum-and-carry/">06 Sum and carry</a></li></ol></li><li><h3>Module 3 · Adding trits</h3><p>The half adder, the full adder and a carry that ripples left, one place at a time.</p><ol><li><a href="/learn/half-adder/">07 The half adder</a></li><li><a href="/learn/full-adder/">08 The full adder</a></li><li><a href="/learn/ripple-carry/">09 Ripple carry</a></li></ol></li><li><h3>Module 4 · Multiplying without a multiplier</h3><p>Copy, drop or flip: a product by one trit, long multiplication, and a MAC that only adds.</p><ol><li><a href="/learn/times-a-trit/">10 Times one trit</a></li><li><a href="/learn/long-multiplication/">11 Long multiplication</a></li><li><a href="/learn/ternary-mac/">12 A ternary MAC</a></li></ol></li><li><h3>Module 5 · Trits in binary memory</h3><p>Two bits per trit, five trits per byte, and the bits a 27-trit word needs.</p><ol><li><a href="/learn/two-bits-per-trit/">13 Two bits per trit</a></li><li><a href="/learn/five-trits-a-byte/">14 Five trits in a byte</a></li><li><a href="/learn/a-word-of-27/">15 A word of 27 trits</a></li></ol></li><li><h3>Module 6 · The TRI-27 instruction word</h3><p>The 32-bit word the Trinity emulator decodes: its fields, its 47 opcodes and its 15-bit immediate.</p><ol><li><a href="/learn/the-instruction-word/">16 The instruction word</a></li><li><a href="/learn/the-opcode-map/">17 The opcode map</a></li><li><a href="/learn/the-immediate/">18 The immediate</a></li></ol></li><li><h3>Module 7 · A ternary machine</h3><p>An ALU built from this course's adders, a three-way jump, and a program you can step.</p><ol><li><a href="/learn/a-ternary-alu/">19 A ternary ALU</a></li><li><a href="/learn/three-way-jump/">20 A three-way jump</a></li><li><a href="/learn/run-a-program/">21 Run a program</a></li></ol></li><li><h3>Module 8 · Three answers</h3><p>Compare at the highest differing trit, find a number in thirds, sort with three-way compares.</p><ol><li><a href="/learn/compare-at-the-top/">22 Compare at the top</a></li><li><a href="/learn/questions-in-thirds/">23 Questions in thirds</a></li><li><a href="/learn/sort-by-three/">24 Sort with three-way compares</a></li></ol></li><li><h3>Module 9 · Ternary neurons</h3><p>Weights of -1, 0, +1: one neuron, a detector and a small layer, with no multiplier anywhere.</p><ol><li><a href="/learn/one-ternary-neuron/">25 One ternary neuron</a></li><li><a href="/learn/a-bar-detector/">26 A bar detector</a></li><li><a href="/learn/a-tiny-layer/">27 A tiny layer</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/folding-taps-onto-slices/index.html b/apps/website/public/learn/folding-taps-onto-slices/index.html
index 421dd5fda2..be6e16eca6 100644
--- a/apps/website/public/learn/folding-taps-onto-slices/index.html
+++ b/apps/website/public/learn/folding-taps-onto-slices/index.html
@@ -1122,5 +1122,6 @@ <h1>Folding taps onto slices</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Adding numbers</h3><p>Why addition is slow: the carry that walks up the word, the dedicated chain in every slice, prefix networks and carry-save trees.</p><ol><li><a href="/learn/the-carry-chain/">01 The carry chain</a></li><li><a href="/learn/prefix-adders/">02 Looking ahead: prefix adders</a></li><li><a href="/learn/carry-save/">03 Three numbers at once</a></li></ol></li><li><h3>Module 2 · Multiplying</h3><p>A product is a sum of shifted copies; Booth recoding halves them, and the DSP48E1 slice does the rest in one block.</p><ol><li><a href="/learn/shift-and-add/">04 Shift and add</a></li><li><a href="/learn/booths-trick/">05 Booth's trick</a></li><li><a href="/learn/the-dsp48e1-slice/">06 The DSP48E1 slice</a></li></ol></li><li><h3>Module 3 · Fixed point</h3><p>Where the binary point sits, what rounding does to a value and to its average, and what each bit of a quantizer buys.</p><ol><li><a href="/learn/q-formats/">07 Q formats</a></li><li><a href="/learn/rounding-and-saturation/">08 Rounding and saturation</a></li><li><a href="/learn/quantization-noise/">09 Quantization noise</a></li></ol></li><li><h3>Module 4 · Functions in hardware</h3><p>Sine, cosine, angle and length from shifts and adds, and when a table is the better answer.</p><ol><li><a href="/learn/rotating-by-shifts/">10 Rotating by shifts</a></li><li><a href="/learn/angle-and-magnitude/">11 Angle and magnitude</a></li><li><a href="/learn/tables-and-interpolation/">12 Tables and interpolation</a></li></ol></li><li><h3>Module 5 · Signals and sampling</h3><p>What sampling does to a frequency, how an oscillator is built from an adder, and what a DFT bin measures.</p><ol><li><a href="/learn/sampling-and-aliasing/">13 Sampling and aliasing</a></li><li><a href="/learn/the-nco/">14 A numerically controlled oscillator</a></li><li><a href="/learn/bins-of-a-spectrum/">15 Bins of a spectrum</a></li></ol></li><li><h3>Module 6 · FIR filters</h3><p>The moving average, a windowed-sinc design with integer taps, and folding the taps onto DSP slices.</p><ol><li><a href="/learn/the-moving-average/">16 The moving average</a></li><li><a href="/learn/designing-taps/">17 Designing taps</a></li><li class="on" aria-current="page"><a href="/learn/folding-taps-onto-slices/">18 Folding taps onto slices</a></li></ol></li><li><h3>Module 7 · Multirate</h3><p>Lowering the sample rate without folding noise in: decimation, the CIC filter and the polyphase form.</p><ol><li><a href="/learn/lowering-the-rate/">19 Lowering the rate</a></li><li><a href="/learn/the-cic-filter/">20 The CIC filter</a></li><li><a href="/learn/polyphase/">21 Polyphase</a></li></ol></li><li><h3>Module 8 · The FFT</h3><p>N log N instead of N^2: butterflies, the bit-reversed input order and the bits each stage adds.</p><ol><li><a href="/learn/the-butterfly/">22 The butterfly</a></li><li><a href="/learn/twiddles-and-bit-reversal/">23 Twiddles and bit reversal</a></li><li><a href="/learn/growth-through-the-stages/">24 Growth through the stages</a></li></ol></li><li><h3>Module 9 · On the bench</h3><p>From the arithmetic to the board: a correlator from the modem spec, a budget of slices and timing, and the capstone filter.</p><ol><li><a href="/learn/a-correlator-finds-the-frame/">25 A correlator finds the frame</a></li><li><a href="/learn/does-it-fit-and-close-timing/">26 Does it fit and close timing?</a></li><li><a href="/learn/capstone-a-filter-on-the-board/">27 Capstone: a filter on the board</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/four-trit-exponent/index.html b/apps/website/public/learn/four-trit-exponent/index.html
index 8f6f142c9e..91859a4c7f 100644
--- a/apps/website/public/learn/four-trit-exponent/index.html
+++ b/apps/website/public/learn/four-trit-exponent/index.html
@@ -142,5 +142,6 @@ <h1>A float whose exponent is four trits</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Lab: our own research</h3><p>A number format of our own, an honest scoreboard, and a model's tables multiplied on the board.</p><ol><li><a href="/learn/t27-number-format/">01 T27: a number format of our own</a></li><li><a href="/learn/an-honest-scoreboard/">02 An honest scoreboard</a></li><li><a href="/learn/a-model-on-the-board/">03 A model's tables on the board</a></li></ol></li><li><h3>Module 2 · AI numbers: the MX block</h3><p>How AI chips keep weights in a few bits: one shared scale per block, the scale byte itself, and what one outlier does to its neighbours.</p><ol><li><a href="/learn/one-scale-per-block/">04 One scale for a block of weights</a></li><li><a href="/learn/scale-byte-native/">05 The scale byte on a real machine</a></li><li><a href="/learn/one-outlier/">06 One outlier, many zeros</a></li></ol></li><li><h3>Module 3 · Ternary weights</h3><p>Weights that are only minus, zero or plus a scale, the five rules a ternary alphabet must pass, and a test pass that checked nothing.</p><ol><li><a href="/learn/phi-scaled-ternary/">07 Weights of minus phi, zero and plus phi</a></li><li><a href="/learn/a-pass-that-checked-nothing/">08 A pass that checked nothing</a></li><li><a href="/learn/five-rules-for-weights/">09 Five rules for a weight alphabet</a></li></ol></li><li><h3>Module 4 · The Ternary Network Float</h3><p>A rule the compiler enforces before any test runs, and a 17-bit float whose exponent is four balanced trits.</p><ol><li><a href="/learn/blocked-is-not-failed/">10 Blocked is not failed</a></li><li class="on" aria-current="page"><a href="/learn/four-trit-exponent/">11 A float whose exponent is four trits</a></li><li><a href="/learn/negate-twice/">12 Negate twice, get the number back</a></li></ol></li><li><h3>Module 5 · Arithmetic on signed numbers</h3><p>Multiply two signed numbers, add them when their signs differ, and do both at once in a multiply-accumulate.</p><ol><li><a href="/learn/a-sign-is-one-bit/">13 A sign is one bit</a></li><li><a href="/learn/opposite-signs-subtract/">14 Opposite signs subtract</a></li><li><a href="/learn/multiply-then-add/">15 Multiply, then add</a></li></ol></li><li><h3>Module 6 · Parts of a neuron</h3><p>A ReLU that bends at zero, a power of two for softmax, and an argmax that names the answer.</p><ol><li><a href="/learn/a-bend-at-zero/">16 A bend at zero</a></li><li><a href="/learn/two-to-the-x/">17 Two to the x</a></li><li><a href="/learn/pick-the-largest/">18 Pick the largest</a></li></ol></li><li><h3>Module 7 · Learning from a mistake</h3><p>A loss that prices a wrong guess in bits, one step that moves a weight against its gradient, and the hidden layer that XOR needs.</p><ol><li><a href="/learn/a-loss-in-bits/">19 A loss in bits</a></li><li><a href="/learn/one-step-downhill/">20 One step downhill</a></li><li><a href="/learn/xor-needs-a-bend/">21 XOR needs a bend</a></li></ol></li><li><h3>Module 8 · BitNet: ternary networks</h3><p>A threshold that squeezes a sum back to three values, one neuron that becomes a different function when its weights change, and a neuron that reads its inputs 27 trits at a time.</p><ol><li><a href="/learn/back-to-three-values/">22 Back to three values</a></li><li><a href="/learn/same-neuron-new-weights/">23 Same neuron, new weights</a></li><li><a href="/learn/a-neuron-in-chunks/">24 A neuron in chunks</a></li></ol></li><li><h3>Module 9 · The ternary MAC as a chip</h3><p>The 27-trit dot product as wires with no register, the same sum added into a register on every clock, and a small whole network to close the course.</p><ol><li><a href="/learn/a-dot-product-in-wires/">25 A dot product made of wires</a></li><li><a href="/learn/add-it-up-every-clock/">26 Add it up on every clock</a></li><li><a href="/learn/a-whole-network/">27 A whole network</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/fpga-is-a-grid/index.html b/apps/website/public/learn/fpga-is-a-grid/index.html
index 6d2727421f..194c756ef3 100644
--- a/apps/website/public/learn/fpga-is-a-grid/index.html
+++ b/apps/website/public/learn/fpga-is-a-grid/index.html
@@ -367,5 +367,6 @@ <h1>An FPGA is a grid you configure</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The chip</h3><p>What an FPGA is, which chip we use, and how its pins meet the board.</p><ol><li class="on" aria-current="page"><a href="/learn/fpga-is-a-grid/">01 An FPGA is a grid you configure</a></li><li><a href="/learn/which-chip/">02 Which chip is this?</a></li><li><a href="/learn/pins-and-package/">03 Pins, balls and the package</a></li></ol></li><li><h3>Module 2 · Numbers in hardware</h3><p>Bits, trits and number formats, and what arithmetic costs in logic.</p><ol><li><a href="/learn/bits-and-trits/">04 Bits, trits and GF16</a></li><li><a href="/learn/what-arithmetic-costs/">05 What arithmetic costs</a></li><li><a href="/learn/how-much-fits/">06 How much fits on the chip</a></li></ol></li><li><h3>Module 3 · Your t27 program</h3><p>Write a spec, test it, and see why compiling is not the same as being right.</p><ol><li><a href="/learn/hello-t27/">07 Hello, t27</a></li><li><a href="/learn/tests-are-the-spec/">08 Tests are part of the spec</a></li><li><a href="/learn/compiles-is-not-correct/">09 Compiles is not the same as correct</a></li></ol></li><li><h3>Module 4 · Inside t27c</h3><p>How the compiler reads a spec and what it writes, including the native t27b.</p><ol><li><a href="/learn/compiler-stages/">10 The stages of t27c</a></li><li><a href="/learn/t27c-on-a-real-machine/">11 t27c on a real machine</a></li><li><a href="/learn/seven-backends-and-t27b/">12 Seven backends, and t27b</a></li></ol></li><li><h3>Module 5 · From spec to hardware</h3><p>The Verilog t27c writes, the cells it becomes, and what one LUT does.</p><ol><li><a href="/learn/spec-to-verilog/">13 From a spec to Verilog</a></li><li><a href="/learn/count-the-cells/">14 Count the cells</a></li><li><a href="/learn/one-lut-up-close/">15 One LUT, up close</a></li></ol></li><li><h3>Module 6 · Reading synthesis</h3><p>What yosys reports about your design, and which warnings matter.</p><ol><li><a href="/learn/synthesis-warnings/">16 Synthesis warnings that matter</a></li><li><a href="/learn/state-machines/">17 State machines</a></li><li><a href="/learn/where-the-luts-go/">18 Where the LUTs go</a></li></ol></li><li><h3>Module 7 · Place, route, timing</h3><p>Where the cells land on the die, and whether the clock is met.</p><ol><li><a href="/learn/placement/">19 Placement: where the cells land</a></li><li><a href="/learn/logic-depth/">20 Logic depth</a></li><li><a href="/learn/timing-slack/">21 Timing and slack</a></li></ol></li><li><h3>Module 8 · The bitstream</h3><p>How a routed design becomes the bits the chip loads, with no Vivado.</p><ol><li><a href="/learn/fasm-to-frames/">22 From FASM to frames</a></li><li><a href="/learn/a-build-receipt/">23 A build, itemised</a></li><li><a href="/learn/read-a-bitstream/">24 Read a bitstream</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>Ask the chip who it is, load the bits, and check them.</p><ol><li><a href="/learn/jtag-asks-the-chip/">25 JTAG asks the chip</a></li><li><a href="/learn/same-bits-on-the-board/">26 The same bits, on the board</a></li><li><a href="/learn/capstone/">27 Capstone: out-guess yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/frames-and-the-fcs/index.html b/apps/website/public/learn/frames-and-the-fcs/index.html
index deca0e25d4..9078fa9987 100644
--- a/apps/website/public/learn/frames-and-the-fcs/index.html
+++ b/apps/website/public/learn/frames-and-the-fcs/index.html
@@ -299,5 +299,6 @@ <h1>Frames and the FCS</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a bus is</h3><p>Why a bus exists at all: a conversation on wires, framed and addressed, and who is allowed to talk.</p><ol><li><a href="/learn/wires-that-talk/">01 Wires that talk</a></li><li><a href="/learn/framing-a-conversation/">02 Framing a conversation</a></li><li><a href="/learn/master-and-servant/">03 Master and servant</a></li></ol></li><li><h3>Module 2 · UART</h3><p>The two-wire bus with no clock: the frame, the divisor that sets the rate, and the status a driver polls.</p><ol><li><a href="/learn/the-frame/">04 The frame: start, data, stop</a></li><li><a href="/learn/baud-and-the-divisor/">05 Baud and the divisor</a></li><li><a href="/learn/status-and-fifos/">06 Status and FIFOs</a></li></ol></li><li><h3>Module 3 · SPI</h3><p>The clocked conversation: four modes, a prescaler ladder, and a chip select per servant.</p><ol><li><a href="/learn/modes-cpol-cpha/">07 Modes: CPOL and CPHA</a></li><li><a href="/learn/the-prescaler-ladder/">08 The prescaler ladder</a></li><li><a href="/learn/chip-select-and-widths/">09 Chip select and widths</a></li></ol></li><li><h3>Module 4 · APB</h3><p>The register bus: PSEL and PENABLE, strobes and wait states, and how many address bits a peripheral count costs.</p><ol><li><a href="/learn/psel-and-penable/">10 PSEL and PENABLE</a></li><li><a href="/learn/strobes-and-waits/">11 Strobes and waits</a></li><li><a href="/learn/address-decode/">12 Address decode</a></li></ol></li><li><h3>Module 5 · AXI4</h3><p>The five channels: address, data and response in both directions, lite or full, bursts and IDs.</p><ol><li><a href="/learn/five-channels/">13 Five channels</a></li><li><a href="/learn/lite-or-full/">14 Lite or full</a></li><li><a href="/learn/bursts-and-ids/">15 Bursts and IDs</a></li></ol></li><li><h3>Module 6 · Memory</h3><p>What sits on the far side of every bus: memory maps, port kinds, and latency that a wait state must cover.</p><ol><li><a href="/learn/memory-maps/">16 Memory maps</a></li><li><a href="/learn/ports-and-kinds/">17 Ports and kinds</a></li><li><a href="/learn/latency/">18 Latency</a></li></ol></li><li><h3>Module 7 · Bridges</h3><p>Why designs grow more than one bus, and the packet bridge that moves work between them.</p><ol><li><a href="/learn/why-bridges/">19 Why bridges</a></li><li><a href="/learn/the-packet-bridge/">20 The packet bridge</a></li><li><a href="/learn/everything-in-one-bench/">21 Everything in one bench</a></li></ol></li><li><h3>Module 8 · Ethernet</h3><p>Frames, the frame check sequence, RGMII timing, and the pre-registered steps of a real bring-up.</p><ol><li class="on" aria-current="page"><a href="/learn/frames-and-the-fcs/">22 Frames and the FCS</a></li><li><a href="/learn/rgmii-timing/">23 RGMII timing</a></li><li><a href="/learn/bring-up-on-the-bench/">24 Bring-up on the bench</a></li></ol></li><li><h3>Module 9 · The bench</h3><p>The discipline that guards real hardware: who holds the IO, taking and giving the lock, and what runs next.</p><ol><li><a href="/learn/who-holds-the-io/">25 Who holds the IO</a></li><li><a href="/learn/claim-and-release/">26 Claim and release</a></li><li><a href="/learn/capstone-what-runs-next/">27 Capstone: what runs next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/framing-a-conversation/index.html b/apps/website/public/learn/framing-a-conversation/index.html
index ed4d72a288..3e0b26170c 100644
--- a/apps/website/public/learn/framing-a-conversation/index.html
+++ b/apps/website/public/learn/framing-a-conversation/index.html
@@ -476,5 +476,6 @@ <h1>Framing a conversation</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a bus is</h3><p>Why a bus exists at all: a conversation on wires, framed and addressed, and who is allowed to talk.</p><ol><li><a href="/learn/wires-that-talk/">01 Wires that talk</a></li><li class="on" aria-current="page"><a href="/learn/framing-a-conversation/">02 Framing a conversation</a></li><li><a href="/learn/master-and-servant/">03 Master and servant</a></li></ol></li><li><h3>Module 2 · UART</h3><p>The two-wire bus with no clock: the frame, the divisor that sets the rate, and the status a driver polls.</p><ol><li><a href="/learn/the-frame/">04 The frame: start, data, stop</a></li><li><a href="/learn/baud-and-the-divisor/">05 Baud and the divisor</a></li><li><a href="/learn/status-and-fifos/">06 Status and FIFOs</a></li></ol></li><li><h3>Module 3 · SPI</h3><p>The clocked conversation: four modes, a prescaler ladder, and a chip select per servant.</p><ol><li><a href="/learn/modes-cpol-cpha/">07 Modes: CPOL and CPHA</a></li><li><a href="/learn/the-prescaler-ladder/">08 The prescaler ladder</a></li><li><a href="/learn/chip-select-and-widths/">09 Chip select and widths</a></li></ol></li><li><h3>Module 4 · APB</h3><p>The register bus: PSEL and PENABLE, strobes and wait states, and how many address bits a peripheral count costs.</p><ol><li><a href="/learn/psel-and-penable/">10 PSEL and PENABLE</a></li><li><a href="/learn/strobes-and-waits/">11 Strobes and waits</a></li><li><a href="/learn/address-decode/">12 Address decode</a></li></ol></li><li><h3>Module 5 · AXI4</h3><p>The five channels: address, data and response in both directions, lite or full, bursts and IDs.</p><ol><li><a href="/learn/five-channels/">13 Five channels</a></li><li><a href="/learn/lite-or-full/">14 Lite or full</a></li><li><a href="/learn/bursts-and-ids/">15 Bursts and IDs</a></li></ol></li><li><h3>Module 6 · Memory</h3><p>What sits on the far side of every bus: memory maps, port kinds, and latency that a wait state must cover.</p><ol><li><a href="/learn/memory-maps/">16 Memory maps</a></li><li><a href="/learn/ports-and-kinds/">17 Ports and kinds</a></li><li><a href="/learn/latency/">18 Latency</a></li></ol></li><li><h3>Module 7 · Bridges</h3><p>Why designs grow more than one bus, and the packet bridge that moves work between them.</p><ol><li><a href="/learn/why-bridges/">19 Why bridges</a></li><li><a href="/learn/the-packet-bridge/">20 The packet bridge</a></li><li><a href="/learn/everything-in-one-bench/">21 Everything in one bench</a></li></ol></li><li><h3>Module 8 · Ethernet</h3><p>Frames, the frame check sequence, RGMII timing, and the pre-registered steps of a real bring-up.</p><ol><li><a href="/learn/frames-and-the-fcs/">22 Frames and the FCS</a></li><li><a href="/learn/rgmii-timing/">23 RGMII timing</a></li><li><a href="/learn/bring-up-on-the-bench/">24 Bring-up on the bench</a></li></ol></li><li><h3>Module 9 · The bench</h3><p>The discipline that guards real hardware: who holds the IO, taking and giving the lock, and what runs next.</p><ol><li><a href="/learn/who-holds-the-io/">25 Who holds the IO</a></li><li><a href="/learn/claim-and-release/">26 Claim and release</a></li><li><a href="/learn/capstone-what-runs-next/">27 Capstone: what runs next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/fsm-coverage/index.html b/apps/website/public/learn/fsm-coverage/index.html
index d95265180e..b32f8ebe3e 100644
--- a/apps/website/public/learn/fsm-coverage/index.html
+++ b/apps/website/public/learn/fsm-coverage/index.html
@@ -576,5 +576,6 @@ <h1>FSM coverage</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Why verify</h3><p>Designs that compile and are wrong, the model that decides, and the plan written before the code.</p><ol><li><a href="/learn/bugs-that-compile/">01 Bugs that compile</a></li><li><a href="/learn/golden-model/">02 The golden model</a></li><li><a href="/learn/plan-before-code/">03 Plan before code</a></li></ol></li><li><h3>Module 2 · Testbenches</h3><p>Stimulus, checks and a verdict, written as one spec beside the design it judges.</p><ol><li><a href="/learn/anatomy-of-a-tb/">04 Anatomy of a testbench</a></li><li><a href="/learn/self-checking/">05 Self-checking, not staring</a></li><li><a href="/learn/directed-vs-random/">06 Directed versus random</a></li></ol></li><li><h3>Module 3 · Waveforms</h3><p>A trace of every signal, read the way a hardware engineer reads it, and two runs compared.</p><ol><li><a href="/learn/vcd-format/">07 The VCD format</a></li><li><a href="/learn/debug-with-waves/">08 Debug with waves</a></li><li><a href="/learn/compare-traces/">09 Compare two traces</a></li></ol></li><li><h3>Module 4 · Conformance vectors</h3><p>Cases with the answer written beside them, kept where the compiler can reach them.</p><ol><li><a href="/learn/what-a-vector-is/">10 What a vector is</a></li><li><a href="/learn/golden-ruler-vectors/">11 Golden ruler vectors</a></li><li><a href="/learn/end-to-end-scenarios/">12 End-to-end scenarios</a></li></ol></li><li><h3>Module 5 · Cosimulation</h3><p>Spec, simulator and board agreeing on the bench Artix-7 XC7A200T, and what to do when they do not.</p><ol><li><a href="/learn/three-way-agreement/">13 Three-way agreement</a></li><li><a href="/learn/simulator-internals/">14 Inside the simulator</a></li><li><a href="/learn/when-they-disagree/">15 When they disagree</a></li></ol></li><li><h3>Module 6 · Coverage</h3><p>What the tests touched: lines, toggles, states, and what that number hides.</p><ol><li><a href="/learn/line-and-toggle/">16 Line and toggle</a></li><li class="on" aria-current="page"><a href="/learn/fsm-coverage/">17 FSM coverage</a></li><li><a href="/learn/coverage-lies/">18 Coverage lies</a></li></ol></li><li><h3>Module 7 · Formal</h3><p>Assertions that hold every cycle, bounded search for a counterexample, and why a proof needs induction.</p><ol><li><a href="/learn/assertions/">19 Assertions</a></li><li><a href="/learn/bounded-model-checking/">20 Bounded model checking</a></li><li><a href="/learn/induction/">21 Induction</a></li></ol></li><li><h3>Module 8 · Mutation</h3><p>Break the design on purpose and count what the tests catch.</p><ol><li><a href="/learn/kill-the-mutant/">22 Kill the mutant</a></li><li><a href="/learn/surviving-mutants/">23 Surviving mutants</a></li><li><a href="/learn/test-strength/">24 Test strength</a></li></ol></li><li><h3>Module 9 · Sign-off</h3><p>One command, every receipt, a clean verdict you can show.</p><ol><li><a href="/learn/build-verify/">25 Build and verify</a></li><li><a href="/learn/board-selftest/">26 The board self-tests</a></li><li><a href="/learn/verify-capstone/">27 Capstone: verify a module whole</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/full-adder/index.html b/apps/website/public/learn/full-adder/index.html
index 95ed570bc6..d42481a9b4 100644
--- a/apps/website/public/learn/full-adder/index.html
+++ b/apps/website/public/learn/full-adder/index.html
@@ -148,5 +148,6 @@ <h1>The full adder</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Three values</h3><p>Why three, how balanced ternary writes every number without a sign, and what flipping and cutting trits do.</p><ol><li><a href="/learn/why-three/">01 Why three</a></li><li><a href="/learn/balanced-ternary/">02 Balanced ternary</a></li><li><a href="/learn/flip-and-cut/">03 Flip to negate, cut to round</a></li></ol></li><li><h3>Module 2 · Logic with unknown</h3><p>Kleene's three-valued gates, two gates binary has no twin for, and addition as a pair of tables.</p><ol><li><a href="/learn/kleene-logic/">04 Three-valued logic</a></li><li><a href="/learn/consensus-and-any/">05 Consensus and any</a></li><li><a href="/learn/sum-and-carry/">06 Sum and carry</a></li></ol></li><li><h3>Module 3 · Adding trits</h3><p>The half adder, the full adder and a carry that ripples left, one place at a time.</p><ol><li><a href="/learn/half-adder/">07 The half adder</a></li><li class="on" aria-current="page"><a href="/learn/full-adder/">08 The full adder</a></li><li><a href="/learn/ripple-carry/">09 Ripple carry</a></li></ol></li><li><h3>Module 4 · Multiplying without a multiplier</h3><p>Copy, drop or flip: a product by one trit, long multiplication, and a MAC that only adds.</p><ol><li><a href="/learn/times-a-trit/">10 Times one trit</a></li><li><a href="/learn/long-multiplication/">11 Long multiplication</a></li><li><a href="/learn/ternary-mac/">12 A ternary MAC</a></li></ol></li><li><h3>Module 5 · Trits in binary memory</h3><p>Two bits per trit, five trits per byte, and the bits a 27-trit word needs.</p><ol><li><a href="/learn/two-bits-per-trit/">13 Two bits per trit</a></li><li><a href="/learn/five-trits-a-byte/">14 Five trits in a byte</a></li><li><a href="/learn/a-word-of-27/">15 A word of 27 trits</a></li></ol></li><li><h3>Module 6 · The TRI-27 instruction word</h3><p>The 32-bit word the Trinity emulator decodes: its fields, its 47 opcodes and its 15-bit immediate.</p><ol><li><a href="/learn/the-instruction-word/">16 The instruction word</a></li><li><a href="/learn/the-opcode-map/">17 The opcode map</a></li><li><a href="/learn/the-immediate/">18 The immediate</a></li></ol></li><li><h3>Module 7 · A ternary machine</h3><p>An ALU built from this course's adders, a three-way jump, and a program you can step.</p><ol><li><a href="/learn/a-ternary-alu/">19 A ternary ALU</a></li><li><a href="/learn/three-way-jump/">20 A three-way jump</a></li><li><a href="/learn/run-a-program/">21 Run a program</a></li></ol></li><li><h3>Module 8 · Three answers</h3><p>Compare at the highest differing trit, find a number in thirds, sort with three-way compares.</p><ol><li><a href="/learn/compare-at-the-top/">22 Compare at the top</a></li><li><a href="/learn/questions-in-thirds/">23 Questions in thirds</a></li><li><a href="/learn/sort-by-three/">24 Sort with three-way compares</a></li></ol></li><li><h3>Module 9 · Ternary neurons</h3><p>Weights of -1, 0, +1: one neuron, a detector and a small layer, with no multiplier anywhere.</p><ol><li><a href="/learn/one-ternary-neuron/">25 One ternary neuron</a></li><li><a href="/learn/a-bar-detector/">26 A bar detector</a></li><li><a href="/learn/a-tiny-layer/">27 A tiny layer</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/functions/index.html b/apps/website/public/learn/functions/index.html
index 8554096a3d..f1ec96f74f 100644
--- a/apps/website/public/learn/functions/index.html
+++ b/apps/website/public/learn/functions/index.html
@@ -89,5 +89,6 @@ <h1>Functions</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · A spec file</h3><p>What a .t27 file holds: a module line, prose lines and comments.</p><ol><li><a href="/learn/what-a-spec-is/">01 What a spec is</a></li><li><a href="/learn/the-module-line/">02 The module line</a></li><li><a href="/learn/comments-and-prose/">03 Comments and prose lines</a></li></ol></li><li><h3>Module 2 · Values and types</h3><p>Constants, the width of a whole number, and true, false and text.</p><ol><li><a href="/learn/constants/">04 Constants</a></li><li><a href="/learn/whole-numbers/">05 Whole numbers and their widths</a></li><li><a href="/learn/true-false-and-text/">06 True, false and text</a></li></ol></li><li><h3>Module 3 · Arrays, trits and expressions</h3><p>Lists of values, the three values of a trit, and what operators compute.</p><ol><li><a href="/learn/arrays/">07 Arrays</a></li><li><a href="/learn/trits/">08 Trits: minus one, zero, one</a></li><li><a href="/learn/expressions/">09 Expressions</a></li></ol></li><li><h3>Module 4 · Tests and functions</h3><p>A test block, several small tests in one spec, and a function with typed inputs.</p><ol><li><a href="/learn/a-test-block/">10 A test block</a></li><li><a href="/learn/many-tests/">11 Many tests in one spec</a></li><li class="on" aria-current="page"><a href="/learn/functions/">12 Functions</a></li></ol></li><li><h3>Module 5 · Inside a function</h3><p>Names with let and var, choices with if and switch, and loops.</p><ol><li><a href="/learn/local-names/">13 Local names: let and var</a></li><li><a href="/learn/if-and-else/">14 Choices: if, else and switch</a></li><li><a href="/learn/loops/">15 Loops: while and for</a></li></ol></li><li><h3>Module 6 · Visibility and your own types</h3><p>What pub shares with other modules, and the structs and enums you declare.</p><ol><li><a href="/learn/pub-or-private/">16 pub or private</a></li><li><a href="/learn/structs/">17 Structs</a></li><li><a href="/learn/enums/">18 Enums</a></li></ol></li><li><h3>Module 7 · Modules, rules and gen-ts</h3><p>Reading another module, stating a rule that always holds, and a spec turned into TypeScript.</p><ol><li><a href="/learn/use-other-modules/">19 Using other modules</a></li><li><a href="/learn/invariants/">20 Invariants and benches</a></li><li><a href="/learn/gen-ts/">21 gen-ts: a spec becomes TypeScript</a></li></ol></li><li><h3>Module 8 · The compiler and tri</h3><p>The backends t27c writes, the stages it reads a spec in, and the tri command.</p><ol><li><a href="/learn/seven-backends/">22 Seven backends</a></li><li><a href="/learn/stages-of-t27c/">23 The stages of t27c</a></li><li><a href="/learn/tri-commands/">24 tri: the command line</a></li></ol></li><li><h3>Module 9 · Errors, a program, what next</h3><p>Reading a compiler error, one small complete program, and the course to take next.</p><ol><li><a href="/learn/reading-errors/">25 Reading compiler errors</a></li><li><a href="/learn/a-small-program/">26 A small complete program</a></li><li><a href="/learn/where-next/">27 Where to go next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/gated-clock-trap/index.html b/apps/website/public/learn/gated-clock-trap/index.html
index 0aa793bec0..76022f224b 100644
--- a/apps/website/public/learn/gated-clock-trap/index.html
+++ b/apps/website/public/learn/gated-clock-trap/index.html
@@ -300,5 +300,6 @@ <h1>The gated-clock trap</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a clock is</h3><p>One edge, one world: what shares a clock edge shares a world, the period and the jitter of a real edge, and where the clock enters a board.</p><ol><li><a href="/learn/one-edge-one-world/">01 One edge, one world</a></li><li><a href="/learn/period-and-jitter/">02 A period, an edge, and jitter</a></li><li><a href="/learn/where-the-clock-enters/">03 Where the clock enters</a></li></ol></li><li><h3>Module 2 · Clock trees</h3><p>Skew and insertion delay, the global buffer network, and the trap of gating a clock with logic.</p><ol><li><a href="/learn/skew-and-insertion/">04 Skew and insertion delay</a></li><li><a href="/learn/global-buffers/">05 The global buffer</a></li><li class="on" aria-current="page"><a href="/learn/gated-clock-trap/">06 The gated-clock trap</a></li></ol></li><li><h3>Module 3 · PLL and MMCM</h3><p>Multiply and divide one clock into another, move its phase in steps of the VCO, and which clocks the analyzer treats as related.</p><ol><li><a href="/learn/multiply-and-divide/">07 Multiply and divide</a></li><li><a href="/learn/phase-and-lock/">08 Phase and lock</a></li><li><a href="/learn/related-or-not/">09 Related or not</a></li></ol></li><li><h3>Module 4 · Resets</h3><p>Assert asynchronously, release synchronously: the three reset kinds, the release pipe, and the tree a reset grows.</p><ol><li><a href="/learn/sync-or-async/">10 Sync or async</a></li><li><a href="/learn/reset-release/">11 The release pipe</a></li><li><a href="/learn/reset-fanout/">12 Reset fanout</a></li></ol></li><li><h3>Module 5 · Metastability</h3><p>The setup-hold window, the mean time between failures in integer arithmetic, and the two flops that fix it.</p><ol><li><a href="/learn/setup-hold-window/">13 The setup-hold window</a></li><li><a href="/learn/mtbf/">14 Mean time between failures</a></li><li><a href="/learn/two-flop-sync/">15 The two-flop synchronizer</a></li></ol></li><li><h3>Module 6 · Crossing many bits</h3><p>Why a binary bus tears, why Gray code does not, and the handshake that moves a pulse between worlds.</p><ol><li><a href="/learn/why-binary-fails/">16 Why binary fails</a></li><li><a href="/learn/gray-code/">17 Gray code</a></li><li><a href="/learn/handshakes-and-pulses/">18 Handshakes and pulses</a></li></ol></li><li><h3>Module 7 · The asynchronous FIFO</h3><p>Pointers, flags and depth: the buffer that moves a stream between two clocks.</p><ol><li><a href="/learn/pointers-and-flags/">19 Pointers and flags</a></li><li><a href="/learn/depth-sizing/">20 Sizing the depth</a></li><li><a href="/learn/fifo-testbench/">21 The FIFO testbench</a></li></ol></li><li><h3>Module 8 · Constraints</h3><p>The lines that tell the analyzer what a clock is, which paths not to check, and what the pins must meet.</p><ol><li><a href="/learn/create-clock/">22 create_clock</a></li><li><a href="/learn/false-path-and-max-delay/">23 False paths and max delay</a></li><li><a href="/learn/io-timing/">24 IO timing</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>A CDC report, one crossing captured at the flip-flops, and the bitstream diff that closes the course.</p><ol><li><a href="/learn/cdc-report/">25 The CDC report</a></li><li><a href="/learn/capture-a-crossing/">26 Capture a crossing</a></li><li><a href="/learn/capstone-diff/">27 Capstone: change it and diff it</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/gen-ts/index.html b/apps/website/public/learn/gen-ts/index.html
index 142daa4668..5ef8f49879 100644
--- a/apps/website/public/learn/gen-ts/index.html
+++ b/apps/website/public/learn/gen-ts/index.html
@@ -86,5 +86,6 @@ <h1>gen-ts: a spec becomes TypeScript</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · A spec file</h3><p>What a .t27 file holds: a module line, prose lines and comments.</p><ol><li><a href="/learn/what-a-spec-is/">01 What a spec is</a></li><li><a href="/learn/the-module-line/">02 The module line</a></li><li><a href="/learn/comments-and-prose/">03 Comments and prose lines</a></li></ol></li><li><h3>Module 2 · Values and types</h3><p>Constants, the width of a whole number, and true, false and text.</p><ol><li><a href="/learn/constants/">04 Constants</a></li><li><a href="/learn/whole-numbers/">05 Whole numbers and their widths</a></li><li><a href="/learn/true-false-and-text/">06 True, false and text</a></li></ol></li><li><h3>Module 3 · Arrays, trits and expressions</h3><p>Lists of values, the three values of a trit, and what operators compute.</p><ol><li><a href="/learn/arrays/">07 Arrays</a></li><li><a href="/learn/trits/">08 Trits: minus one, zero, one</a></li><li><a href="/learn/expressions/">09 Expressions</a></li></ol></li><li><h3>Module 4 · Tests and functions</h3><p>A test block, several small tests in one spec, and a function with typed inputs.</p><ol><li><a href="/learn/a-test-block/">10 A test block</a></li><li><a href="/learn/many-tests/">11 Many tests in one spec</a></li><li><a href="/learn/functions/">12 Functions</a></li></ol></li><li><h3>Module 5 · Inside a function</h3><p>Names with let and var, choices with if and switch, and loops.</p><ol><li><a href="/learn/local-names/">13 Local names: let and var</a></li><li><a href="/learn/if-and-else/">14 Choices: if, else and switch</a></li><li><a href="/learn/loops/">15 Loops: while and for</a></li></ol></li><li><h3>Module 6 · Visibility and your own types</h3><p>What pub shares with other modules, and the structs and enums you declare.</p><ol><li><a href="/learn/pub-or-private/">16 pub or private</a></li><li><a href="/learn/structs/">17 Structs</a></li><li><a href="/learn/enums/">18 Enums</a></li></ol></li><li><h3>Module 7 · Modules, rules and gen-ts</h3><p>Reading another module, stating a rule that always holds, and a spec turned into TypeScript.</p><ol><li><a href="/learn/use-other-modules/">19 Using other modules</a></li><li><a href="/learn/invariants/">20 Invariants and benches</a></li><li class="on" aria-current="page"><a href="/learn/gen-ts/">21 gen-ts: a spec becomes TypeScript</a></li></ol></li><li><h3>Module 8 · The compiler and tri</h3><p>The backends t27c writes, the stages it reads a spec in, and the tri command.</p><ol><li><a href="/learn/seven-backends/">22 Seven backends</a></li><li><a href="/learn/stages-of-t27c/">23 The stages of t27c</a></li><li><a href="/learn/tri-commands/">24 tri: the command line</a></li></ol></li><li><h3>Module 9 · Errors, a program, what next</h3><p>Reading a compiler error, one small complete program, and the course to take next.</p><ol><li><a href="/learn/reading-errors/">25 Reading compiler errors</a></li><li><a href="/learn/a-small-program/">26 A small complete program</a></li><li><a href="/learn/where-next/">27 Where to go next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/gf10-ten-bits/index.html b/apps/website/public/learn/gf10-ten-bits/index.html
index fc497bc5c8..06da57a705 100644
--- a/apps/website/public/learn/gf10-ten-bits/index.html
+++ b/apps/website/public/learn/gf10-ten-bits/index.html
@@ -154,5 +154,6 @@ <h1>GF10: ten bits</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The rule and its numbers</h3><p>One rule splits every width, the ratio it aims at, and the Lucas numbers behind the 3.</p><ol><li><a href="/learn/a-float-cut-by-phi/">01 A float cut by phi</a></li><li><a href="/learn/phi-as-a-ratio/">02 Phi as a ratio</a></li><li><a href="/learn/lucas-numbers-stay-whole/">03 Lucas numbers stay whole</a></li></ol></li><li><h3>Module 2 · Why phi, why three</h3><p>Why the split is phi, why base three, and how a spec checks GF16 keeps phi.</p><ol><li><a href="/learn/why-the-split-is-phi/">04 Why the split is phi</a></li><li><a href="/learn/why-base-three/">05 Why base three</a></li><li><a href="/learn/phi-in-sixteen-bits/">06 Phi in sixteen bits</a></li></ol></li><li><h3>Module 3 · The small rungs: GF4 to GF8</h3><p>GF4, GF6 and GF8, the fewest bits, where rounding to whole bits costs the most.</p><ol><li><a href="/learn/gf4-four-bits/">07 GF4: four bits</a></li><li><a href="/learn/gf6-six-bits/">08 GF6: six bits</a></li><li><a href="/learn/gf8-one-byte/">09 GF8: one byte</a></li></ol></li><li><h3>Module 4 · Ten to fourteen bits</h3><p>GF10, GF12 and GF14, and how the distance from 1 / phi moves as the word grows.</p><ol><li class="on" aria-current="page"><a href="/learn/gf10-ten-bits/">10 GF10: ten bits</a></li><li><a href="/learn/gf12-twelve-bits/">11 GF12: twelve bits</a></li><li><a href="/learn/gf14-fourteen-bits/">12 GF14: fourteen bits</a></li></ol></li><li><h3>Module 5 · GF16 at work</h3><p>The primary 16-bit format, a two-term dot product in GF-T16, then GF20 and GF24.</p><ol><li><a href="/learn/gf16-the-primary/">13 GF16: the primary format</a></li><li><a href="/learn/two-products-one-sum/">14 Two products, one sum</a></li><li><a href="/learn/gf24-twenty-four-bits/">15 GF20 and GF24</a></li></ol></li><li><h3>Module 6 · GF32 to GF64</h3><p>GF32 beside IEEE single, GF48 with no IEEE twin, GF64 beside IEEE double.</p><ol><li><a href="/learn/gf32-a-single/">16 GF32: a single</a></li><li><a href="/learn/gf48-forty-eight-bits/">17 GF48: forty-eight bits</a></li><li><a href="/learn/gf64-a-double/">18 GF64: a double</a></li></ol></li><li><h3>Module 7 · GF96 to GF256</h3><p>GF96, GF128 and GF256, where the specs hold the layout with invariants.</p><ol><li><a href="/learn/gf96-ninety-six-bits/">19 GF96: ninety-six bits</a></li><li><a href="/learn/gf128-a-quad/">20 GF128: a quad</a></li><li><a href="/learn/gf256-two-five-six/">21 GF256: two hundred fifty-six bits</a></li></ol></li><li><h3>Module 8 · The widest rungs, then trits</h3><p>GF512 and GF1024, the two widest rungs, then GF-T8, where the exponent moves to trits.</p><ol><li><a href="/learn/gf512-five-twelve/">22 GF512: five hundred twelve bits</a></li><li><a href="/learn/gf1024-one-kilobit/">23 GF1024: one kilobit</a></li><li><a href="/learn/gft8-exponent-in-trits/">24 GF-T8: an exponent in trits</a></li></ol></li><li><h3>Module 9 · More trits, then the decode</h3><p>GF-T16 and GF-T32, then why fixed fields decode in parallel and a posit does not.</p><ol><li><a href="/learn/gft16-six-trits/">25 GF-T16: six trits</a></li><li><a href="/learn/gft32-twelve-trits/">26 GF-T32: twelve trits</a></li><li><a href="/learn/decode-in-one-step/">27 Decode in one step</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/gf1024-one-kilobit/index.html b/apps/website/public/learn/gf1024-one-kilobit/index.html
index 4e33d7cdf3..d4dcb581d1 100644
--- a/apps/website/public/learn/gf1024-one-kilobit/index.html
+++ b/apps/website/public/learn/gf1024-one-kilobit/index.html
@@ -152,5 +152,6 @@ <h1>GF1024: one kilobit</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The rule and its numbers</h3><p>One rule splits every width, the ratio it aims at, and the Lucas numbers behind the 3.</p><ol><li><a href="/learn/a-float-cut-by-phi/">01 A float cut by phi</a></li><li><a href="/learn/phi-as-a-ratio/">02 Phi as a ratio</a></li><li><a href="/learn/lucas-numbers-stay-whole/">03 Lucas numbers stay whole</a></li></ol></li><li><h3>Module 2 · Why phi, why three</h3><p>Why the split is phi, why base three, and how a spec checks GF16 keeps phi.</p><ol><li><a href="/learn/why-the-split-is-phi/">04 Why the split is phi</a></li><li><a href="/learn/why-base-three/">05 Why base three</a></li><li><a href="/learn/phi-in-sixteen-bits/">06 Phi in sixteen bits</a></li></ol></li><li><h3>Module 3 · The small rungs: GF4 to GF8</h3><p>GF4, GF6 and GF8, the fewest bits, where rounding to whole bits costs the most.</p><ol><li><a href="/learn/gf4-four-bits/">07 GF4: four bits</a></li><li><a href="/learn/gf6-six-bits/">08 GF6: six bits</a></li><li><a href="/learn/gf8-one-byte/">09 GF8: one byte</a></li></ol></li><li><h3>Module 4 · Ten to fourteen bits</h3><p>GF10, GF12 and GF14, and how the distance from 1 / phi moves as the word grows.</p><ol><li><a href="/learn/gf10-ten-bits/">10 GF10: ten bits</a></li><li><a href="/learn/gf12-twelve-bits/">11 GF12: twelve bits</a></li><li><a href="/learn/gf14-fourteen-bits/">12 GF14: fourteen bits</a></li></ol></li><li><h3>Module 5 · GF16 at work</h3><p>The primary 16-bit format, a two-term dot product in GF-T16, then GF20 and GF24.</p><ol><li><a href="/learn/gf16-the-primary/">13 GF16: the primary format</a></li><li><a href="/learn/two-products-one-sum/">14 Two products, one sum</a></li><li><a href="/learn/gf24-twenty-four-bits/">15 GF20 and GF24</a></li></ol></li><li><h3>Module 6 · GF32 to GF64</h3><p>GF32 beside IEEE single, GF48 with no IEEE twin, GF64 beside IEEE double.</p><ol><li><a href="/learn/gf32-a-single/">16 GF32: a single</a></li><li><a href="/learn/gf48-forty-eight-bits/">17 GF48: forty-eight bits</a></li><li><a href="/learn/gf64-a-double/">18 GF64: a double</a></li></ol></li><li><h3>Module 7 · GF96 to GF256</h3><p>GF96, GF128 and GF256, where the specs hold the layout with invariants.</p><ol><li><a href="/learn/gf96-ninety-six-bits/">19 GF96: ninety-six bits</a></li><li><a href="/learn/gf128-a-quad/">20 GF128: a quad</a></li><li><a href="/learn/gf256-two-five-six/">21 GF256: two hundred fifty-six bits</a></li></ol></li><li><h3>Module 8 · The widest rungs, then trits</h3><p>GF512 and GF1024, the two widest rungs, then GF-T8, where the exponent moves to trits.</p><ol><li><a href="/learn/gf512-five-twelve/">22 GF512: five hundred twelve bits</a></li><li class="on" aria-current="page"><a href="/learn/gf1024-one-kilobit/">23 GF1024: one kilobit</a></li><li><a href="/learn/gft8-exponent-in-trits/">24 GF-T8: an exponent in trits</a></li></ol></li><li><h3>Module 9 · More trits, then the decode</h3><p>GF-T16 and GF-T32, then why fixed fields decode in parallel and a posit does not.</p><ol><li><a href="/learn/gft16-six-trits/">25 GF-T16: six trits</a></li><li><a href="/learn/gft32-twelve-trits/">26 GF-T32: twelve trits</a></li><li><a href="/learn/decode-in-one-step/">27 Decode in one step</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/gf12-twelve-bits/index.html b/apps/website/public/learn/gf12-twelve-bits/index.html
index b1f6dbbb96..d96fa82b8b 100644
--- a/apps/website/public/learn/gf12-twelve-bits/index.html
+++ b/apps/website/public/learn/gf12-twelve-bits/index.html
@@ -562,5 +562,6 @@ <h1>GF12: twelve bits</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The rule and its numbers</h3><p>One rule splits every width, the ratio it aims at, and the Lucas numbers behind the 3.</p><ol><li><a href="/learn/a-float-cut-by-phi/">01 A float cut by phi</a></li><li><a href="/learn/phi-as-a-ratio/">02 Phi as a ratio</a></li><li><a href="/learn/lucas-numbers-stay-whole/">03 Lucas numbers stay whole</a></li></ol></li><li><h3>Module 2 · Why phi, why three</h3><p>Why the split is phi, why base three, and how a spec checks GF16 keeps phi.</p><ol><li><a href="/learn/why-the-split-is-phi/">04 Why the split is phi</a></li><li><a href="/learn/why-base-three/">05 Why base three</a></li><li><a href="/learn/phi-in-sixteen-bits/">06 Phi in sixteen bits</a></li></ol></li><li><h3>Module 3 · The small rungs: GF4 to GF8</h3><p>GF4, GF6 and GF8, the fewest bits, where rounding to whole bits costs the most.</p><ol><li><a href="/learn/gf4-four-bits/">07 GF4: four bits</a></li><li><a href="/learn/gf6-six-bits/">08 GF6: six bits</a></li><li><a href="/learn/gf8-one-byte/">09 GF8: one byte</a></li></ol></li><li><h3>Module 4 · Ten to fourteen bits</h3><p>GF10, GF12 and GF14, and how the distance from 1 / phi moves as the word grows.</p><ol><li><a href="/learn/gf10-ten-bits/">10 GF10: ten bits</a></li><li class="on" aria-current="page"><a href="/learn/gf12-twelve-bits/">11 GF12: twelve bits</a></li><li><a href="/learn/gf14-fourteen-bits/">12 GF14: fourteen bits</a></li></ol></li><li><h3>Module 5 · GF16 at work</h3><p>The primary 16-bit format, a two-term dot product in GF-T16, then GF20 and GF24.</p><ol><li><a href="/learn/gf16-the-primary/">13 GF16: the primary format</a></li><li><a href="/learn/two-products-one-sum/">14 Two products, one sum</a></li><li><a href="/learn/gf24-twenty-four-bits/">15 GF20 and GF24</a></li></ol></li><li><h3>Module 6 · GF32 to GF64</h3><p>GF32 beside IEEE single, GF48 with no IEEE twin, GF64 beside IEEE double.</p><ol><li><a href="/learn/gf32-a-single/">16 GF32: a single</a></li><li><a href="/learn/gf48-forty-eight-bits/">17 GF48: forty-eight bits</a></li><li><a href="/learn/gf64-a-double/">18 GF64: a double</a></li></ol></li><li><h3>Module 7 · GF96 to GF256</h3><p>GF96, GF128 and GF256, where the specs hold the layout with invariants.</p><ol><li><a href="/learn/gf96-ninety-six-bits/">19 GF96: ninety-six bits</a></li><li><a href="/learn/gf128-a-quad/">20 GF128: a quad</a></li><li><a href="/learn/gf256-two-five-six/">21 GF256: two hundred fifty-six bits</a></li></ol></li><li><h3>Module 8 · The widest rungs, then trits</h3><p>GF512 and GF1024, the two widest rungs, then GF-T8, where the exponent moves to trits.</p><ol><li><a href="/learn/gf512-five-twelve/">22 GF512: five hundred twelve bits</a></li><li><a href="/learn/gf1024-one-kilobit/">23 GF1024: one kilobit</a></li><li><a href="/learn/gft8-exponent-in-trits/">24 GF-T8: an exponent in trits</a></li></ol></li><li><h3>Module 9 · More trits, then the decode</h3><p>GF-T16 and GF-T32, then why fixed fields decode in parallel and a posit does not.</p><ol><li><a href="/learn/gft16-six-trits/">25 GF-T16: six trits</a></li><li><a href="/learn/gft32-twelve-trits/">26 GF-T32: twelve trits</a></li><li><a href="/learn/decode-in-one-step/">27 Decode in one step</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/gf128-a-quad/index.html b/apps/website/public/learn/gf128-a-quad/index.html
index d0accf57f2..e8d9703069 100644
--- a/apps/website/public/learn/gf128-a-quad/index.html
+++ b/apps/website/public/learn/gf128-a-quad/index.html
@@ -146,5 +146,6 @@ <h1>GF128: a quad</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The rule and its numbers</h3><p>One rule splits every width, the ratio it aims at, and the Lucas numbers behind the 3.</p><ol><li><a href="/learn/a-float-cut-by-phi/">01 A float cut by phi</a></li><li><a href="/learn/phi-as-a-ratio/">02 Phi as a ratio</a></li><li><a href="/learn/lucas-numbers-stay-whole/">03 Lucas numbers stay whole</a></li></ol></li><li><h3>Module 2 · Why phi, why three</h3><p>Why the split is phi, why base three, and how a spec checks GF16 keeps phi.</p><ol><li><a href="/learn/why-the-split-is-phi/">04 Why the split is phi</a></li><li><a href="/learn/why-base-three/">05 Why base three</a></li><li><a href="/learn/phi-in-sixteen-bits/">06 Phi in sixteen bits</a></li></ol></li><li><h3>Module 3 · The small rungs: GF4 to GF8</h3><p>GF4, GF6 and GF8, the fewest bits, where rounding to whole bits costs the most.</p><ol><li><a href="/learn/gf4-four-bits/">07 GF4: four bits</a></li><li><a href="/learn/gf6-six-bits/">08 GF6: six bits</a></li><li><a href="/learn/gf8-one-byte/">09 GF8: one byte</a></li></ol></li><li><h3>Module 4 · Ten to fourteen bits</h3><p>GF10, GF12 and GF14, and how the distance from 1 / phi moves as the word grows.</p><ol><li><a href="/learn/gf10-ten-bits/">10 GF10: ten bits</a></li><li><a href="/learn/gf12-twelve-bits/">11 GF12: twelve bits</a></li><li><a href="/learn/gf14-fourteen-bits/">12 GF14: fourteen bits</a></li></ol></li><li><h3>Module 5 · GF16 at work</h3><p>The primary 16-bit format, a two-term dot product in GF-T16, then GF20 and GF24.</p><ol><li><a href="/learn/gf16-the-primary/">13 GF16: the primary format</a></li><li><a href="/learn/two-products-one-sum/">14 Two products, one sum</a></li><li><a href="/learn/gf24-twenty-four-bits/">15 GF20 and GF24</a></li></ol></li><li><h3>Module 6 · GF32 to GF64</h3><p>GF32 beside IEEE single, GF48 with no IEEE twin, GF64 beside IEEE double.</p><ol><li><a href="/learn/gf32-a-single/">16 GF32: a single</a></li><li><a href="/learn/gf48-forty-eight-bits/">17 GF48: forty-eight bits</a></li><li><a href="/learn/gf64-a-double/">18 GF64: a double</a></li></ol></li><li><h3>Module 7 · GF96 to GF256</h3><p>GF96, GF128 and GF256, where the specs hold the layout with invariants.</p><ol><li><a href="/learn/gf96-ninety-six-bits/">19 GF96: ninety-six bits</a></li><li class="on" aria-current="page"><a href="/learn/gf128-a-quad/">20 GF128: a quad</a></li><li><a href="/learn/gf256-two-five-six/">21 GF256: two hundred fifty-six bits</a></li></ol></li><li><h3>Module 8 · The widest rungs, then trits</h3><p>GF512 and GF1024, the two widest rungs, then GF-T8, where the exponent moves to trits.</p><ol><li><a href="/learn/gf512-five-twelve/">22 GF512: five hundred twelve bits</a></li><li><a href="/learn/gf1024-one-kilobit/">23 GF1024: one kilobit</a></li><li><a href="/learn/gft8-exponent-in-trits/">24 GF-T8: an exponent in trits</a></li></ol></li><li><h3>Module 9 · More trits, then the decode</h3><p>GF-T16 and GF-T32, then why fixed fields decode in parallel and a posit does not.</p><ol><li><a href="/learn/gft16-six-trits/">25 GF-T16: six trits</a></li><li><a href="/learn/gft32-twelve-trits/">26 GF-T32: twelve trits</a></li><li><a href="/learn/decode-in-one-step/">27 Decode in one step</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/gf14-fourteen-bits/index.html b/apps/website/public/learn/gf14-fourteen-bits/index.html
index 6a4e4f1771..fb01f03498 100644
--- a/apps/website/public/learn/gf14-fourteen-bits/index.html
+++ b/apps/website/public/learn/gf14-fourteen-bits/index.html
@@ -154,5 +154,6 @@ <h1>GF14: fourteen bits</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The rule and its numbers</h3><p>One rule splits every width, the ratio it aims at, and the Lucas numbers behind the 3.</p><ol><li><a href="/learn/a-float-cut-by-phi/">01 A float cut by phi</a></li><li><a href="/learn/phi-as-a-ratio/">02 Phi as a ratio</a></li><li><a href="/learn/lucas-numbers-stay-whole/">03 Lucas numbers stay whole</a></li></ol></li><li><h3>Module 2 · Why phi, why three</h3><p>Why the split is phi, why base three, and how a spec checks GF16 keeps phi.</p><ol><li><a href="/learn/why-the-split-is-phi/">04 Why the split is phi</a></li><li><a href="/learn/why-base-three/">05 Why base three</a></li><li><a href="/learn/phi-in-sixteen-bits/">06 Phi in sixteen bits</a></li></ol></li><li><h3>Module 3 · The small rungs: GF4 to GF8</h3><p>GF4, GF6 and GF8, the fewest bits, where rounding to whole bits costs the most.</p><ol><li><a href="/learn/gf4-four-bits/">07 GF4: four bits</a></li><li><a href="/learn/gf6-six-bits/">08 GF6: six bits</a></li><li><a href="/learn/gf8-one-byte/">09 GF8: one byte</a></li></ol></li><li><h3>Module 4 · Ten to fourteen bits</h3><p>GF10, GF12 and GF14, and how the distance from 1 / phi moves as the word grows.</p><ol><li><a href="/learn/gf10-ten-bits/">10 GF10: ten bits</a></li><li><a href="/learn/gf12-twelve-bits/">11 GF12: twelve bits</a></li><li class="on" aria-current="page"><a href="/learn/gf14-fourteen-bits/">12 GF14: fourteen bits</a></li></ol></li><li><h3>Module 5 · GF16 at work</h3><p>The primary 16-bit format, a two-term dot product in GF-T16, then GF20 and GF24.</p><ol><li><a href="/learn/gf16-the-primary/">13 GF16: the primary format</a></li><li><a href="/learn/two-products-one-sum/">14 Two products, one sum</a></li><li><a href="/learn/gf24-twenty-four-bits/">15 GF20 and GF24</a></li></ol></li><li><h3>Module 6 · GF32 to GF64</h3><p>GF32 beside IEEE single, GF48 with no IEEE twin, GF64 beside IEEE double.</p><ol><li><a href="/learn/gf32-a-single/">16 GF32: a single</a></li><li><a href="/learn/gf48-forty-eight-bits/">17 GF48: forty-eight bits</a></li><li><a href="/learn/gf64-a-double/">18 GF64: a double</a></li></ol></li><li><h3>Module 7 · GF96 to GF256</h3><p>GF96, GF128 and GF256, where the specs hold the layout with invariants.</p><ol><li><a href="/learn/gf96-ninety-six-bits/">19 GF96: ninety-six bits</a></li><li><a href="/learn/gf128-a-quad/">20 GF128: a quad</a></li><li><a href="/learn/gf256-two-five-six/">21 GF256: two hundred fifty-six bits</a></li></ol></li><li><h3>Module 8 · The widest rungs, then trits</h3><p>GF512 and GF1024, the two widest rungs, then GF-T8, where the exponent moves to trits.</p><ol><li><a href="/learn/gf512-five-twelve/">22 GF512: five hundred twelve bits</a></li><li><a href="/learn/gf1024-one-kilobit/">23 GF1024: one kilobit</a></li><li><a href="/learn/gft8-exponent-in-trits/">24 GF-T8: an exponent in trits</a></li></ol></li><li><h3>Module 9 · More trits, then the decode</h3><p>GF-T16 and GF-T32, then why fixed fields decode in parallel and a posit does not.</p><ol><li><a href="/learn/gft16-six-trits/">25 GF-T16: six trits</a></li><li><a href="/learn/gft32-twelve-trits/">26 GF-T32: twelve trits</a></li><li><a href="/learn/decode-in-one-step/">27 Decode in one step</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/gf16-the-primary/index.html b/apps/website/public/learn/gf16-the-primary/index.html
index 0e6b6d2c30..170edeae49 100644
--- a/apps/website/public/learn/gf16-the-primary/index.html
+++ b/apps/website/public/learn/gf16-the-primary/index.html
@@ -3863,5 +3863,6 @@ <h1>GF16: the primary format</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The rule and its numbers</h3><p>One rule splits every width, the ratio it aims at, and the Lucas numbers behind the 3.</p><ol><li><a href="/learn/a-float-cut-by-phi/">01 A float cut by phi</a></li><li><a href="/learn/phi-as-a-ratio/">02 Phi as a ratio</a></li><li><a href="/learn/lucas-numbers-stay-whole/">03 Lucas numbers stay whole</a></li></ol></li><li><h3>Module 2 · Why phi, why three</h3><p>Why the split is phi, why base three, and how a spec checks GF16 keeps phi.</p><ol><li><a href="/learn/why-the-split-is-phi/">04 Why the split is phi</a></li><li><a href="/learn/why-base-three/">05 Why base three</a></li><li><a href="/learn/phi-in-sixteen-bits/">06 Phi in sixteen bits</a></li></ol></li><li><h3>Module 3 · The small rungs: GF4 to GF8</h3><p>GF4, GF6 and GF8, the fewest bits, where rounding to whole bits costs the most.</p><ol><li><a href="/learn/gf4-four-bits/">07 GF4: four bits</a></li><li><a href="/learn/gf6-six-bits/">08 GF6: six bits</a></li><li><a href="/learn/gf8-one-byte/">09 GF8: one byte</a></li></ol></li><li><h3>Module 4 · Ten to fourteen bits</h3><p>GF10, GF12 and GF14, and how the distance from 1 / phi moves as the word grows.</p><ol><li><a href="/learn/gf10-ten-bits/">10 GF10: ten bits</a></li><li><a href="/learn/gf12-twelve-bits/">11 GF12: twelve bits</a></li><li><a href="/learn/gf14-fourteen-bits/">12 GF14: fourteen bits</a></li></ol></li><li><h3>Module 5 · GF16 at work</h3><p>The primary 16-bit format, a two-term dot product in GF-T16, then GF20 and GF24.</p><ol><li class="on" aria-current="page"><a href="/learn/gf16-the-primary/">13 GF16: the primary format</a></li><li><a href="/learn/two-products-one-sum/">14 Two products, one sum</a></li><li><a href="/learn/gf24-twenty-four-bits/">15 GF20 and GF24</a></li></ol></li><li><h3>Module 6 · GF32 to GF64</h3><p>GF32 beside IEEE single, GF48 with no IEEE twin, GF64 beside IEEE double.</p><ol><li><a href="/learn/gf32-a-single/">16 GF32: a single</a></li><li><a href="/learn/gf48-forty-eight-bits/">17 GF48: forty-eight bits</a></li><li><a href="/learn/gf64-a-double/">18 GF64: a double</a></li></ol></li><li><h3>Module 7 · GF96 to GF256</h3><p>GF96, GF128 and GF256, where the specs hold the layout with invariants.</p><ol><li><a href="/learn/gf96-ninety-six-bits/">19 GF96: ninety-six bits</a></li><li><a href="/learn/gf128-a-quad/">20 GF128: a quad</a></li><li><a href="/learn/gf256-two-five-six/">21 GF256: two hundred fifty-six bits</a></li></ol></li><li><h3>Module 8 · The widest rungs, then trits</h3><p>GF512 and GF1024, the two widest rungs, then GF-T8, where the exponent moves to trits.</p><ol><li><a href="/learn/gf512-five-twelve/">22 GF512: five hundred twelve bits</a></li><li><a href="/learn/gf1024-one-kilobit/">23 GF1024: one kilobit</a></li><li><a href="/learn/gft8-exponent-in-trits/">24 GF-T8: an exponent in trits</a></li></ol></li><li><h3>Module 9 · More trits, then the decode</h3><p>GF-T16 and GF-T32, then why fixed fields decode in parallel and a posit does not.</p><ol><li><a href="/learn/gft16-six-trits/">25 GF-T16: six trits</a></li><li><a href="/learn/gft32-twelve-trits/">26 GF-T32: twelve trits</a></li><li><a href="/learn/decode-in-one-step/">27 Decode in one step</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/gf24-twenty-four-bits/index.html b/apps/website/public/learn/gf24-twenty-four-bits/index.html
index 73be4e9da2..adba708c34 100644
--- a/apps/website/public/learn/gf24-twenty-four-bits/index.html
+++ b/apps/website/public/learn/gf24-twenty-four-bits/index.html
@@ -549,5 +549,6 @@ <h1>GF20 and GF24</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The rule and its numbers</h3><p>One rule splits every width, the ratio it aims at, and the Lucas numbers behind the 3.</p><ol><li><a href="/learn/a-float-cut-by-phi/">01 A float cut by phi</a></li><li><a href="/learn/phi-as-a-ratio/">02 Phi as a ratio</a></li><li><a href="/learn/lucas-numbers-stay-whole/">03 Lucas numbers stay whole</a></li></ol></li><li><h3>Module 2 · Why phi, why three</h3><p>Why the split is phi, why base three, and how a spec checks GF16 keeps phi.</p><ol><li><a href="/learn/why-the-split-is-phi/">04 Why the split is phi</a></li><li><a href="/learn/why-base-three/">05 Why base three</a></li><li><a href="/learn/phi-in-sixteen-bits/">06 Phi in sixteen bits</a></li></ol></li><li><h3>Module 3 · The small rungs: GF4 to GF8</h3><p>GF4, GF6 and GF8, the fewest bits, where rounding to whole bits costs the most.</p><ol><li><a href="/learn/gf4-four-bits/">07 GF4: four bits</a></li><li><a href="/learn/gf6-six-bits/">08 GF6: six bits</a></li><li><a href="/learn/gf8-one-byte/">09 GF8: one byte</a></li></ol></li><li><h3>Module 4 · Ten to fourteen bits</h3><p>GF10, GF12 and GF14, and how the distance from 1 / phi moves as the word grows.</p><ol><li><a href="/learn/gf10-ten-bits/">10 GF10: ten bits</a></li><li><a href="/learn/gf12-twelve-bits/">11 GF12: twelve bits</a></li><li><a href="/learn/gf14-fourteen-bits/">12 GF14: fourteen bits</a></li></ol></li><li><h3>Module 5 · GF16 at work</h3><p>The primary 16-bit format, a two-term dot product in GF-T16, then GF20 and GF24.</p><ol><li><a href="/learn/gf16-the-primary/">13 GF16: the primary format</a></li><li><a href="/learn/two-products-one-sum/">14 Two products, one sum</a></li><li class="on" aria-current="page"><a href="/learn/gf24-twenty-four-bits/">15 GF20 and GF24</a></li></ol></li><li><h3>Module 6 · GF32 to GF64</h3><p>GF32 beside IEEE single, GF48 with no IEEE twin, GF64 beside IEEE double.</p><ol><li><a href="/learn/gf32-a-single/">16 GF32: a single</a></li><li><a href="/learn/gf48-forty-eight-bits/">17 GF48: forty-eight bits</a></li><li><a href="/learn/gf64-a-double/">18 GF64: a double</a></li></ol></li><li><h3>Module 7 · GF96 to GF256</h3><p>GF96, GF128 and GF256, where the specs hold the layout with invariants.</p><ol><li><a href="/learn/gf96-ninety-six-bits/">19 GF96: ninety-six bits</a></li><li><a href="/learn/gf128-a-quad/">20 GF128: a quad</a></li><li><a href="/learn/gf256-two-five-six/">21 GF256: two hundred fifty-six bits</a></li></ol></li><li><h3>Module 8 · The widest rungs, then trits</h3><p>GF512 and GF1024, the two widest rungs, then GF-T8, where the exponent moves to trits.</p><ol><li><a href="/learn/gf512-five-twelve/">22 GF512: five hundred twelve bits</a></li><li><a href="/learn/gf1024-one-kilobit/">23 GF1024: one kilobit</a></li><li><a href="/learn/gft8-exponent-in-trits/">24 GF-T8: an exponent in trits</a></li></ol></li><li><h3>Module 9 · More trits, then the decode</h3><p>GF-T16 and GF-T32, then why fixed fields decode in parallel and a posit does not.</p><ol><li><a href="/learn/gft16-six-trits/">25 GF-T16: six trits</a></li><li><a href="/learn/gft32-twelve-trits/">26 GF-T32: twelve trits</a></li><li><a href="/learn/decode-in-one-step/">27 Decode in one step</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/gf256-two-five-six/index.html b/apps/website/public/learn/gf256-two-five-six/index.html
index d2788fe555..a4d2afd997 100644
--- a/apps/website/public/learn/gf256-two-five-six/index.html
+++ b/apps/website/public/learn/gf256-two-five-six/index.html
@@ -146,5 +146,6 @@ <h1>GF256: two hundred fifty-six bits</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The rule and its numbers</h3><p>One rule splits every width, the ratio it aims at, and the Lucas numbers behind the 3.</p><ol><li><a href="/learn/a-float-cut-by-phi/">01 A float cut by phi</a></li><li><a href="/learn/phi-as-a-ratio/">02 Phi as a ratio</a></li><li><a href="/learn/lucas-numbers-stay-whole/">03 Lucas numbers stay whole</a></li></ol></li><li><h3>Module 2 · Why phi, why three</h3><p>Why the split is phi, why base three, and how a spec checks GF16 keeps phi.</p><ol><li><a href="/learn/why-the-split-is-phi/">04 Why the split is phi</a></li><li><a href="/learn/why-base-three/">05 Why base three</a></li><li><a href="/learn/phi-in-sixteen-bits/">06 Phi in sixteen bits</a></li></ol></li><li><h3>Module 3 · The small rungs: GF4 to GF8</h3><p>GF4, GF6 and GF8, the fewest bits, where rounding to whole bits costs the most.</p><ol><li><a href="/learn/gf4-four-bits/">07 GF4: four bits</a></li><li><a href="/learn/gf6-six-bits/">08 GF6: six bits</a></li><li><a href="/learn/gf8-one-byte/">09 GF8: one byte</a></li></ol></li><li><h3>Module 4 · Ten to fourteen bits</h3><p>GF10, GF12 and GF14, and how the distance from 1 / phi moves as the word grows.</p><ol><li><a href="/learn/gf10-ten-bits/">10 GF10: ten bits</a></li><li><a href="/learn/gf12-twelve-bits/">11 GF12: twelve bits</a></li><li><a href="/learn/gf14-fourteen-bits/">12 GF14: fourteen bits</a></li></ol></li><li><h3>Module 5 · GF16 at work</h3><p>The primary 16-bit format, a two-term dot product in GF-T16, then GF20 and GF24.</p><ol><li><a href="/learn/gf16-the-primary/">13 GF16: the primary format</a></li><li><a href="/learn/two-products-one-sum/">14 Two products, one sum</a></li><li><a href="/learn/gf24-twenty-four-bits/">15 GF20 and GF24</a></li></ol></li><li><h3>Module 6 · GF32 to GF64</h3><p>GF32 beside IEEE single, GF48 with no IEEE twin, GF64 beside IEEE double.</p><ol><li><a href="/learn/gf32-a-single/">16 GF32: a single</a></li><li><a href="/learn/gf48-forty-eight-bits/">17 GF48: forty-eight bits</a></li><li><a href="/learn/gf64-a-double/">18 GF64: a double</a></li></ol></li><li><h3>Module 7 · GF96 to GF256</h3><p>GF96, GF128 and GF256, where the specs hold the layout with invariants.</p><ol><li><a href="/learn/gf96-ninety-six-bits/">19 GF96: ninety-six bits</a></li><li><a href="/learn/gf128-a-quad/">20 GF128: a quad</a></li><li class="on" aria-current="page"><a href="/learn/gf256-two-five-six/">21 GF256: two hundred fifty-six bits</a></li></ol></li><li><h3>Module 8 · The widest rungs, then trits</h3><p>GF512 and GF1024, the two widest rungs, then GF-T8, where the exponent moves to trits.</p><ol><li><a href="/learn/gf512-five-twelve/">22 GF512: five hundred twelve bits</a></li><li><a href="/learn/gf1024-one-kilobit/">23 GF1024: one kilobit</a></li><li><a href="/learn/gft8-exponent-in-trits/">24 GF-T8: an exponent in trits</a></li></ol></li><li><h3>Module 9 · More trits, then the decode</h3><p>GF-T16 and GF-T32, then why fixed fields decode in parallel and a posit does not.</p><ol><li><a href="/learn/gft16-six-trits/">25 GF-T16: six trits</a></li><li><a href="/learn/gft32-twelve-trits/">26 GF-T32: twelve trits</a></li><li><a href="/learn/decode-in-one-step/">27 Decode in one step</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/gf32-a-single/index.html b/apps/website/public/learn/gf32-a-single/index.html
index 72bf4a3eb0..2f44814e63 100644
--- a/apps/website/public/learn/gf32-a-single/index.html
+++ b/apps/website/public/learn/gf32-a-single/index.html
@@ -560,5 +560,6 @@ <h1>GF32: a single</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The rule and its numbers</h3><p>One rule splits every width, the ratio it aims at, and the Lucas numbers behind the 3.</p><ol><li><a href="/learn/a-float-cut-by-phi/">01 A float cut by phi</a></li><li><a href="/learn/phi-as-a-ratio/">02 Phi as a ratio</a></li><li><a href="/learn/lucas-numbers-stay-whole/">03 Lucas numbers stay whole</a></li></ol></li><li><h3>Module 2 · Why phi, why three</h3><p>Why the split is phi, why base three, and how a spec checks GF16 keeps phi.</p><ol><li><a href="/learn/why-the-split-is-phi/">04 Why the split is phi</a></li><li><a href="/learn/why-base-three/">05 Why base three</a></li><li><a href="/learn/phi-in-sixteen-bits/">06 Phi in sixteen bits</a></li></ol></li><li><h3>Module 3 · The small rungs: GF4 to GF8</h3><p>GF4, GF6 and GF8, the fewest bits, where rounding to whole bits costs the most.</p><ol><li><a href="/learn/gf4-four-bits/">07 GF4: four bits</a></li><li><a href="/learn/gf6-six-bits/">08 GF6: six bits</a></li><li><a href="/learn/gf8-one-byte/">09 GF8: one byte</a></li></ol></li><li><h3>Module 4 · Ten to fourteen bits</h3><p>GF10, GF12 and GF14, and how the distance from 1 / phi moves as the word grows.</p><ol><li><a href="/learn/gf10-ten-bits/">10 GF10: ten bits</a></li><li><a href="/learn/gf12-twelve-bits/">11 GF12: twelve bits</a></li><li><a href="/learn/gf14-fourteen-bits/">12 GF14: fourteen bits</a></li></ol></li><li><h3>Module 5 · GF16 at work</h3><p>The primary 16-bit format, a two-term dot product in GF-T16, then GF20 and GF24.</p><ol><li><a href="/learn/gf16-the-primary/">13 GF16: the primary format</a></li><li><a href="/learn/two-products-one-sum/">14 Two products, one sum</a></li><li><a href="/learn/gf24-twenty-four-bits/">15 GF20 and GF24</a></li></ol></li><li><h3>Module 6 · GF32 to GF64</h3><p>GF32 beside IEEE single, GF48 with no IEEE twin, GF64 beside IEEE double.</p><ol><li class="on" aria-current="page"><a href="/learn/gf32-a-single/">16 GF32: a single</a></li><li><a href="/learn/gf48-forty-eight-bits/">17 GF48: forty-eight bits</a></li><li><a href="/learn/gf64-a-double/">18 GF64: a double</a></li></ol></li><li><h3>Module 7 · GF96 to GF256</h3><p>GF96, GF128 and GF256, where the specs hold the layout with invariants.</p><ol><li><a href="/learn/gf96-ninety-six-bits/">19 GF96: ninety-six bits</a></li><li><a href="/learn/gf128-a-quad/">20 GF128: a quad</a></li><li><a href="/learn/gf256-two-five-six/">21 GF256: two hundred fifty-six bits</a></li></ol></li><li><h3>Module 8 · The widest rungs, then trits</h3><p>GF512 and GF1024, the two widest rungs, then GF-T8, where the exponent moves to trits.</p><ol><li><a href="/learn/gf512-five-twelve/">22 GF512: five hundred twelve bits</a></li><li><a href="/learn/gf1024-one-kilobit/">23 GF1024: one kilobit</a></li><li><a href="/learn/gft8-exponent-in-trits/">24 GF-T8: an exponent in trits</a></li></ol></li><li><h3>Module 9 · More trits, then the decode</h3><p>GF-T16 and GF-T32, then why fixed fields decode in parallel and a posit does not.</p><ol><li><a href="/learn/gft16-six-trits/">25 GF-T16: six trits</a></li><li><a href="/learn/gft32-twelve-trits/">26 GF-T32: twelve trits</a></li><li><a href="/learn/decode-in-one-step/">27 Decode in one step</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/gf4-four-bits/index.html b/apps/website/public/learn/gf4-four-bits/index.html
index 6c3356428b..a7fb7fdbdd 100644
--- a/apps/website/public/learn/gf4-four-bits/index.html
+++ b/apps/website/public/learn/gf4-four-bits/index.html
@@ -375,5 +375,6 @@ <h1>GF4: four bits</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The rule and its numbers</h3><p>One rule splits every width, the ratio it aims at, and the Lucas numbers behind the 3.</p><ol><li><a href="/learn/a-float-cut-by-phi/">01 A float cut by phi</a></li><li><a href="/learn/phi-as-a-ratio/">02 Phi as a ratio</a></li><li><a href="/learn/lucas-numbers-stay-whole/">03 Lucas numbers stay whole</a></li></ol></li><li><h3>Module 2 · Why phi, why three</h3><p>Why the split is phi, why base three, and how a spec checks GF16 keeps phi.</p><ol><li><a href="/learn/why-the-split-is-phi/">04 Why the split is phi</a></li><li><a href="/learn/why-base-three/">05 Why base three</a></li><li><a href="/learn/phi-in-sixteen-bits/">06 Phi in sixteen bits</a></li></ol></li><li><h3>Module 3 · The small rungs: GF4 to GF8</h3><p>GF4, GF6 and GF8, the fewest bits, where rounding to whole bits costs the most.</p><ol><li class="on" aria-current="page"><a href="/learn/gf4-four-bits/">07 GF4: four bits</a></li><li><a href="/learn/gf6-six-bits/">08 GF6: six bits</a></li><li><a href="/learn/gf8-one-byte/">09 GF8: one byte</a></li></ol></li><li><h3>Module 4 · Ten to fourteen bits</h3><p>GF10, GF12 and GF14, and how the distance from 1 / phi moves as the word grows.</p><ol><li><a href="/learn/gf10-ten-bits/">10 GF10: ten bits</a></li><li><a href="/learn/gf12-twelve-bits/">11 GF12: twelve bits</a></li><li><a href="/learn/gf14-fourteen-bits/">12 GF14: fourteen bits</a></li></ol></li><li><h3>Module 5 · GF16 at work</h3><p>The primary 16-bit format, a two-term dot product in GF-T16, then GF20 and GF24.</p><ol><li><a href="/learn/gf16-the-primary/">13 GF16: the primary format</a></li><li><a href="/learn/two-products-one-sum/">14 Two products, one sum</a></li><li><a href="/learn/gf24-twenty-four-bits/">15 GF20 and GF24</a></li></ol></li><li><h3>Module 6 · GF32 to GF64</h3><p>GF32 beside IEEE single, GF48 with no IEEE twin, GF64 beside IEEE double.</p><ol><li><a href="/learn/gf32-a-single/">16 GF32: a single</a></li><li><a href="/learn/gf48-forty-eight-bits/">17 GF48: forty-eight bits</a></li><li><a href="/learn/gf64-a-double/">18 GF64: a double</a></li></ol></li><li><h3>Module 7 · GF96 to GF256</h3><p>GF96, GF128 and GF256, where the specs hold the layout with invariants.</p><ol><li><a href="/learn/gf96-ninety-six-bits/">19 GF96: ninety-six bits</a></li><li><a href="/learn/gf128-a-quad/">20 GF128: a quad</a></li><li><a href="/learn/gf256-two-five-six/">21 GF256: two hundred fifty-six bits</a></li></ol></li><li><h3>Module 8 · The widest rungs, then trits</h3><p>GF512 and GF1024, the two widest rungs, then GF-T8, where the exponent moves to trits.</p><ol><li><a href="/learn/gf512-five-twelve/">22 GF512: five hundred twelve bits</a></li><li><a href="/learn/gf1024-one-kilobit/">23 GF1024: one kilobit</a></li><li><a href="/learn/gft8-exponent-in-trits/">24 GF-T8: an exponent in trits</a></li></ol></li><li><h3>Module 9 · More trits, then the decode</h3><p>GF-T16 and GF-T32, then why fixed fields decode in parallel and a posit does not.</p><ol><li><a href="/learn/gft16-six-trits/">25 GF-T16: six trits</a></li><li><a href="/learn/gft32-twelve-trits/">26 GF-T32: twelve trits</a></li><li><a href="/learn/decode-in-one-step/">27 Decode in one step</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/gf48-forty-eight-bits/index.html b/apps/website/public/learn/gf48-forty-eight-bits/index.html
index 691540f975..4dc406987c 100644
--- a/apps/website/public/learn/gf48-forty-eight-bits/index.html
+++ b/apps/website/public/learn/gf48-forty-eight-bits/index.html
@@ -154,5 +154,6 @@ <h1>GF48: forty-eight bits</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The rule and its numbers</h3><p>One rule splits every width, the ratio it aims at, and the Lucas numbers behind the 3.</p><ol><li><a href="/learn/a-float-cut-by-phi/">01 A float cut by phi</a></li><li><a href="/learn/phi-as-a-ratio/">02 Phi as a ratio</a></li><li><a href="/learn/lucas-numbers-stay-whole/">03 Lucas numbers stay whole</a></li></ol></li><li><h3>Module 2 · Why phi, why three</h3><p>Why the split is phi, why base three, and how a spec checks GF16 keeps phi.</p><ol><li><a href="/learn/why-the-split-is-phi/">04 Why the split is phi</a></li><li><a href="/learn/why-base-three/">05 Why base three</a></li><li><a href="/learn/phi-in-sixteen-bits/">06 Phi in sixteen bits</a></li></ol></li><li><h3>Module 3 · The small rungs: GF4 to GF8</h3><p>GF4, GF6 and GF8, the fewest bits, where rounding to whole bits costs the most.</p><ol><li><a href="/learn/gf4-four-bits/">07 GF4: four bits</a></li><li><a href="/learn/gf6-six-bits/">08 GF6: six bits</a></li><li><a href="/learn/gf8-one-byte/">09 GF8: one byte</a></li></ol></li><li><h3>Module 4 · Ten to fourteen bits</h3><p>GF10, GF12 and GF14, and how the distance from 1 / phi moves as the word grows.</p><ol><li><a href="/learn/gf10-ten-bits/">10 GF10: ten bits</a></li><li><a href="/learn/gf12-twelve-bits/">11 GF12: twelve bits</a></li><li><a href="/learn/gf14-fourteen-bits/">12 GF14: fourteen bits</a></li></ol></li><li><h3>Module 5 · GF16 at work</h3><p>The primary 16-bit format, a two-term dot product in GF-T16, then GF20 and GF24.</p><ol><li><a href="/learn/gf16-the-primary/">13 GF16: the primary format</a></li><li><a href="/learn/two-products-one-sum/">14 Two products, one sum</a></li><li><a href="/learn/gf24-twenty-four-bits/">15 GF20 and GF24</a></li></ol></li><li><h3>Module 6 · GF32 to GF64</h3><p>GF32 beside IEEE single, GF48 with no IEEE twin, GF64 beside IEEE double.</p><ol><li><a href="/learn/gf32-a-single/">16 GF32: a single</a></li><li class="on" aria-current="page"><a href="/learn/gf48-forty-eight-bits/">17 GF48: forty-eight bits</a></li><li><a href="/learn/gf64-a-double/">18 GF64: a double</a></li></ol></li><li><h3>Module 7 · GF96 to GF256</h3><p>GF96, GF128 and GF256, where the specs hold the layout with invariants.</p><ol><li><a href="/learn/gf96-ninety-six-bits/">19 GF96: ninety-six bits</a></li><li><a href="/learn/gf128-a-quad/">20 GF128: a quad</a></li><li><a href="/learn/gf256-two-five-six/">21 GF256: two hundred fifty-six bits</a></li></ol></li><li><h3>Module 8 · The widest rungs, then trits</h3><p>GF512 and GF1024, the two widest rungs, then GF-T8, where the exponent moves to trits.</p><ol><li><a href="/learn/gf512-five-twelve/">22 GF512: five hundred twelve bits</a></li><li><a href="/learn/gf1024-one-kilobit/">23 GF1024: one kilobit</a></li><li><a href="/learn/gft8-exponent-in-trits/">24 GF-T8: an exponent in trits</a></li></ol></li><li><h3>Module 9 · More trits, then the decode</h3><p>GF-T16 and GF-T32, then why fixed fields decode in parallel and a posit does not.</p><ol><li><a href="/learn/gft16-six-trits/">25 GF-T16: six trits</a></li><li><a href="/learn/gft32-twelve-trits/">26 GF-T32: twelve trits</a></li><li><a href="/learn/decode-in-one-step/">27 Decode in one step</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/gf512-five-twelve/index.html b/apps/website/public/learn/gf512-five-twelve/index.html
index 565c80379f..d110842e16 100644
--- a/apps/website/public/learn/gf512-five-twelve/index.html
+++ b/apps/website/public/learn/gf512-five-twelve/index.html
@@ -147,5 +147,6 @@ <h1>GF512: five hundred twelve bits</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The rule and its numbers</h3><p>One rule splits every width, the ratio it aims at, and the Lucas numbers behind the 3.</p><ol><li><a href="/learn/a-float-cut-by-phi/">01 A float cut by phi</a></li><li><a href="/learn/phi-as-a-ratio/">02 Phi as a ratio</a></li><li><a href="/learn/lucas-numbers-stay-whole/">03 Lucas numbers stay whole</a></li></ol></li><li><h3>Module 2 · Why phi, why three</h3><p>Why the split is phi, why base three, and how a spec checks GF16 keeps phi.</p><ol><li><a href="/learn/why-the-split-is-phi/">04 Why the split is phi</a></li><li><a href="/learn/why-base-three/">05 Why base three</a></li><li><a href="/learn/phi-in-sixteen-bits/">06 Phi in sixteen bits</a></li></ol></li><li><h3>Module 3 · The small rungs: GF4 to GF8</h3><p>GF4, GF6 and GF8, the fewest bits, where rounding to whole bits costs the most.</p><ol><li><a href="/learn/gf4-four-bits/">07 GF4: four bits</a></li><li><a href="/learn/gf6-six-bits/">08 GF6: six bits</a></li><li><a href="/learn/gf8-one-byte/">09 GF8: one byte</a></li></ol></li><li><h3>Module 4 · Ten to fourteen bits</h3><p>GF10, GF12 and GF14, and how the distance from 1 / phi moves as the word grows.</p><ol><li><a href="/learn/gf10-ten-bits/">10 GF10: ten bits</a></li><li><a href="/learn/gf12-twelve-bits/">11 GF12: twelve bits</a></li><li><a href="/learn/gf14-fourteen-bits/">12 GF14: fourteen bits</a></li></ol></li><li><h3>Module 5 · GF16 at work</h3><p>The primary 16-bit format, a two-term dot product in GF-T16, then GF20 and GF24.</p><ol><li><a href="/learn/gf16-the-primary/">13 GF16: the primary format</a></li><li><a href="/learn/two-products-one-sum/">14 Two products, one sum</a></li><li><a href="/learn/gf24-twenty-four-bits/">15 GF20 and GF24</a></li></ol></li><li><h3>Module 6 · GF32 to GF64</h3><p>GF32 beside IEEE single, GF48 with no IEEE twin, GF64 beside IEEE double.</p><ol><li><a href="/learn/gf32-a-single/">16 GF32: a single</a></li><li><a href="/learn/gf48-forty-eight-bits/">17 GF48: forty-eight bits</a></li><li><a href="/learn/gf64-a-double/">18 GF64: a double</a></li></ol></li><li><h3>Module 7 · GF96 to GF256</h3><p>GF96, GF128 and GF256, where the specs hold the layout with invariants.</p><ol><li><a href="/learn/gf96-ninety-six-bits/">19 GF96: ninety-six bits</a></li><li><a href="/learn/gf128-a-quad/">20 GF128: a quad</a></li><li><a href="/learn/gf256-two-five-six/">21 GF256: two hundred fifty-six bits</a></li></ol></li><li><h3>Module 8 · The widest rungs, then trits</h3><p>GF512 and GF1024, the two widest rungs, then GF-T8, where the exponent moves to trits.</p><ol><li class="on" aria-current="page"><a href="/learn/gf512-five-twelve/">22 GF512: five hundred twelve bits</a></li><li><a href="/learn/gf1024-one-kilobit/">23 GF1024: one kilobit</a></li><li><a href="/learn/gft8-exponent-in-trits/">24 GF-T8: an exponent in trits</a></li></ol></li><li><h3>Module 9 · More trits, then the decode</h3><p>GF-T16 and GF-T32, then why fixed fields decode in parallel and a posit does not.</p><ol><li><a href="/learn/gft16-six-trits/">25 GF-T16: six trits</a></li><li><a href="/learn/gft32-twelve-trits/">26 GF-T32: twelve trits</a></li><li><a href="/learn/decode-in-one-step/">27 Decode in one step</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/gf6-six-bits/index.html b/apps/website/public/learn/gf6-six-bits/index.html
index 6f87535c37..2b02ff66c9 100644
--- a/apps/website/public/learn/gf6-six-bits/index.html
+++ b/apps/website/public/learn/gf6-six-bits/index.html
@@ -154,5 +154,6 @@ <h1>GF6: six bits</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The rule and its numbers</h3><p>One rule splits every width, the ratio it aims at, and the Lucas numbers behind the 3.</p><ol><li><a href="/learn/a-float-cut-by-phi/">01 A float cut by phi</a></li><li><a href="/learn/phi-as-a-ratio/">02 Phi as a ratio</a></li><li><a href="/learn/lucas-numbers-stay-whole/">03 Lucas numbers stay whole</a></li></ol></li><li><h3>Module 2 · Why phi, why three</h3><p>Why the split is phi, why base three, and how a spec checks GF16 keeps phi.</p><ol><li><a href="/learn/why-the-split-is-phi/">04 Why the split is phi</a></li><li><a href="/learn/why-base-three/">05 Why base three</a></li><li><a href="/learn/phi-in-sixteen-bits/">06 Phi in sixteen bits</a></li></ol></li><li><h3>Module 3 · The small rungs: GF4 to GF8</h3><p>GF4, GF6 and GF8, the fewest bits, where rounding to whole bits costs the most.</p><ol><li><a href="/learn/gf4-four-bits/">07 GF4: four bits</a></li><li class="on" aria-current="page"><a href="/learn/gf6-six-bits/">08 GF6: six bits</a></li><li><a href="/learn/gf8-one-byte/">09 GF8: one byte</a></li></ol></li><li><h3>Module 4 · Ten to fourteen bits</h3><p>GF10, GF12 and GF14, and how the distance from 1 / phi moves as the word grows.</p><ol><li><a href="/learn/gf10-ten-bits/">10 GF10: ten bits</a></li><li><a href="/learn/gf12-twelve-bits/">11 GF12: twelve bits</a></li><li><a href="/learn/gf14-fourteen-bits/">12 GF14: fourteen bits</a></li></ol></li><li><h3>Module 5 · GF16 at work</h3><p>The primary 16-bit format, a two-term dot product in GF-T16, then GF20 and GF24.</p><ol><li><a href="/learn/gf16-the-primary/">13 GF16: the primary format</a></li><li><a href="/learn/two-products-one-sum/">14 Two products, one sum</a></li><li><a href="/learn/gf24-twenty-four-bits/">15 GF20 and GF24</a></li></ol></li><li><h3>Module 6 · GF32 to GF64</h3><p>GF32 beside IEEE single, GF48 with no IEEE twin, GF64 beside IEEE double.</p><ol><li><a href="/learn/gf32-a-single/">16 GF32: a single</a></li><li><a href="/learn/gf48-forty-eight-bits/">17 GF48: forty-eight bits</a></li><li><a href="/learn/gf64-a-double/">18 GF64: a double</a></li></ol></li><li><h3>Module 7 · GF96 to GF256</h3><p>GF96, GF128 and GF256, where the specs hold the layout with invariants.</p><ol><li><a href="/learn/gf96-ninety-six-bits/">19 GF96: ninety-six bits</a></li><li><a href="/learn/gf128-a-quad/">20 GF128: a quad</a></li><li><a href="/learn/gf256-two-five-six/">21 GF256: two hundred fifty-six bits</a></li></ol></li><li><h3>Module 8 · The widest rungs, then trits</h3><p>GF512 and GF1024, the two widest rungs, then GF-T8, where the exponent moves to trits.</p><ol><li><a href="/learn/gf512-five-twelve/">22 GF512: five hundred twelve bits</a></li><li><a href="/learn/gf1024-one-kilobit/">23 GF1024: one kilobit</a></li><li><a href="/learn/gft8-exponent-in-trits/">24 GF-T8: an exponent in trits</a></li></ol></li><li><h3>Module 9 · More trits, then the decode</h3><p>GF-T16 and GF-T32, then why fixed fields decode in parallel and a posit does not.</p><ol><li><a href="/learn/gft16-six-trits/">25 GF-T16: six trits</a></li><li><a href="/learn/gft32-twelve-trits/">26 GF-T32: twelve trits</a></li><li><a href="/learn/decode-in-one-step/">27 Decode in one step</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/gf64-a-double/index.html b/apps/website/public/learn/gf64-a-double/index.html
index 48d3f6f5fc..8a7d7a1171 100644
--- a/apps/website/public/learn/gf64-a-double/index.html
+++ b/apps/website/public/learn/gf64-a-double/index.html
@@ -298,5 +298,6 @@ <h1>GF64: a double</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The rule and its numbers</h3><p>One rule splits every width, the ratio it aims at, and the Lucas numbers behind the 3.</p><ol><li><a href="/learn/a-float-cut-by-phi/">01 A float cut by phi</a></li><li><a href="/learn/phi-as-a-ratio/">02 Phi as a ratio</a></li><li><a href="/learn/lucas-numbers-stay-whole/">03 Lucas numbers stay whole</a></li></ol></li><li><h3>Module 2 · Why phi, why three</h3><p>Why the split is phi, why base three, and how a spec checks GF16 keeps phi.</p><ol><li><a href="/learn/why-the-split-is-phi/">04 Why the split is phi</a></li><li><a href="/learn/why-base-three/">05 Why base three</a></li><li><a href="/learn/phi-in-sixteen-bits/">06 Phi in sixteen bits</a></li></ol></li><li><h3>Module 3 · The small rungs: GF4 to GF8</h3><p>GF4, GF6 and GF8, the fewest bits, where rounding to whole bits costs the most.</p><ol><li><a href="/learn/gf4-four-bits/">07 GF4: four bits</a></li><li><a href="/learn/gf6-six-bits/">08 GF6: six bits</a></li><li><a href="/learn/gf8-one-byte/">09 GF8: one byte</a></li></ol></li><li><h3>Module 4 · Ten to fourteen bits</h3><p>GF10, GF12 and GF14, and how the distance from 1 / phi moves as the word grows.</p><ol><li><a href="/learn/gf10-ten-bits/">10 GF10: ten bits</a></li><li><a href="/learn/gf12-twelve-bits/">11 GF12: twelve bits</a></li><li><a href="/learn/gf14-fourteen-bits/">12 GF14: fourteen bits</a></li></ol></li><li><h3>Module 5 · GF16 at work</h3><p>The primary 16-bit format, a two-term dot product in GF-T16, then GF20 and GF24.</p><ol><li><a href="/learn/gf16-the-primary/">13 GF16: the primary format</a></li><li><a href="/learn/two-products-one-sum/">14 Two products, one sum</a></li><li><a href="/learn/gf24-twenty-four-bits/">15 GF20 and GF24</a></li></ol></li><li><h3>Module 6 · GF32 to GF64</h3><p>GF32 beside IEEE single, GF48 with no IEEE twin, GF64 beside IEEE double.</p><ol><li><a href="/learn/gf32-a-single/">16 GF32: a single</a></li><li><a href="/learn/gf48-forty-eight-bits/">17 GF48: forty-eight bits</a></li><li class="on" aria-current="page"><a href="/learn/gf64-a-double/">18 GF64: a double</a></li></ol></li><li><h3>Module 7 · GF96 to GF256</h3><p>GF96, GF128 and GF256, where the specs hold the layout with invariants.</p><ol><li><a href="/learn/gf96-ninety-six-bits/">19 GF96: ninety-six bits</a></li><li><a href="/learn/gf128-a-quad/">20 GF128: a quad</a></li><li><a href="/learn/gf256-two-five-six/">21 GF256: two hundred fifty-six bits</a></li></ol></li><li><h3>Module 8 · The widest rungs, then trits</h3><p>GF512 and GF1024, the two widest rungs, then GF-T8, where the exponent moves to trits.</p><ol><li><a href="/learn/gf512-five-twelve/">22 GF512: five hundred twelve bits</a></li><li><a href="/learn/gf1024-one-kilobit/">23 GF1024: one kilobit</a></li><li><a href="/learn/gft8-exponent-in-trits/">24 GF-T8: an exponent in trits</a></li></ol></li><li><h3>Module 9 · More trits, then the decode</h3><p>GF-T16 and GF-T32, then why fixed fields decode in parallel and a posit does not.</p><ol><li><a href="/learn/gft16-six-trits/">25 GF-T16: six trits</a></li><li><a href="/learn/gft32-twelve-trits/">26 GF-T32: twelve trits</a></li><li><a href="/learn/decode-in-one-step/">27 Decode in one step</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/gf8-one-byte/index.html b/apps/website/public/learn/gf8-one-byte/index.html
index 56ed8d72bd..7a8e872fe4 100644
--- a/apps/website/public/learn/gf8-one-byte/index.html
+++ b/apps/website/public/learn/gf8-one-byte/index.html
@@ -598,5 +598,6 @@ <h1>GF8: one byte</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The rule and its numbers</h3><p>One rule splits every width, the ratio it aims at, and the Lucas numbers behind the 3.</p><ol><li><a href="/learn/a-float-cut-by-phi/">01 A float cut by phi</a></li><li><a href="/learn/phi-as-a-ratio/">02 Phi as a ratio</a></li><li><a href="/learn/lucas-numbers-stay-whole/">03 Lucas numbers stay whole</a></li></ol></li><li><h3>Module 2 · Why phi, why three</h3><p>Why the split is phi, why base three, and how a spec checks GF16 keeps phi.</p><ol><li><a href="/learn/why-the-split-is-phi/">04 Why the split is phi</a></li><li><a href="/learn/why-base-three/">05 Why base three</a></li><li><a href="/learn/phi-in-sixteen-bits/">06 Phi in sixteen bits</a></li></ol></li><li><h3>Module 3 · The small rungs: GF4 to GF8</h3><p>GF4, GF6 and GF8, the fewest bits, where rounding to whole bits costs the most.</p><ol><li><a href="/learn/gf4-four-bits/">07 GF4: four bits</a></li><li><a href="/learn/gf6-six-bits/">08 GF6: six bits</a></li><li class="on" aria-current="page"><a href="/learn/gf8-one-byte/">09 GF8: one byte</a></li></ol></li><li><h3>Module 4 · Ten to fourteen bits</h3><p>GF10, GF12 and GF14, and how the distance from 1 / phi moves as the word grows.</p><ol><li><a href="/learn/gf10-ten-bits/">10 GF10: ten bits</a></li><li><a href="/learn/gf12-twelve-bits/">11 GF12: twelve bits</a></li><li><a href="/learn/gf14-fourteen-bits/">12 GF14: fourteen bits</a></li></ol></li><li><h3>Module 5 · GF16 at work</h3><p>The primary 16-bit format, a two-term dot product in GF-T16, then GF20 and GF24.</p><ol><li><a href="/learn/gf16-the-primary/">13 GF16: the primary format</a></li><li><a href="/learn/two-products-one-sum/">14 Two products, one sum</a></li><li><a href="/learn/gf24-twenty-four-bits/">15 GF20 and GF24</a></li></ol></li><li><h3>Module 6 · GF32 to GF64</h3><p>GF32 beside IEEE single, GF48 with no IEEE twin, GF64 beside IEEE double.</p><ol><li><a href="/learn/gf32-a-single/">16 GF32: a single</a></li><li><a href="/learn/gf48-forty-eight-bits/">17 GF48: forty-eight bits</a></li><li><a href="/learn/gf64-a-double/">18 GF64: a double</a></li></ol></li><li><h3>Module 7 · GF96 to GF256</h3><p>GF96, GF128 and GF256, where the specs hold the layout with invariants.</p><ol><li><a href="/learn/gf96-ninety-six-bits/">19 GF96: ninety-six bits</a></li><li><a href="/learn/gf128-a-quad/">20 GF128: a quad</a></li><li><a href="/learn/gf256-two-five-six/">21 GF256: two hundred fifty-six bits</a></li></ol></li><li><h3>Module 8 · The widest rungs, then trits</h3><p>GF512 and GF1024, the two widest rungs, then GF-T8, where the exponent moves to trits.</p><ol><li><a href="/learn/gf512-five-twelve/">22 GF512: five hundred twelve bits</a></li><li><a href="/learn/gf1024-one-kilobit/">23 GF1024: one kilobit</a></li><li><a href="/learn/gft8-exponent-in-trits/">24 GF-T8: an exponent in trits</a></li></ol></li><li><h3>Module 9 · More trits, then the decode</h3><p>GF-T16 and GF-T32, then why fixed fields decode in parallel and a posit does not.</p><ol><li><a href="/learn/gft16-six-trits/">25 GF-T16: six trits</a></li><li><a href="/learn/gft32-twelve-trits/">26 GF-T32: twelve trits</a></li><li><a href="/learn/decode-in-one-step/">27 Decode in one step</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/gf96-ninety-six-bits/index.html b/apps/website/public/learn/gf96-ninety-six-bits/index.html
index a0e5260d5e..ce8b9ef532 100644
--- a/apps/website/public/learn/gf96-ninety-six-bits/index.html
+++ b/apps/website/public/learn/gf96-ninety-six-bits/index.html
@@ -146,5 +146,6 @@ <h1>GF96: ninety-six bits</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The rule and its numbers</h3><p>One rule splits every width, the ratio it aims at, and the Lucas numbers behind the 3.</p><ol><li><a href="/learn/a-float-cut-by-phi/">01 A float cut by phi</a></li><li><a href="/learn/phi-as-a-ratio/">02 Phi as a ratio</a></li><li><a href="/learn/lucas-numbers-stay-whole/">03 Lucas numbers stay whole</a></li></ol></li><li><h3>Module 2 · Why phi, why three</h3><p>Why the split is phi, why base three, and how a spec checks GF16 keeps phi.</p><ol><li><a href="/learn/why-the-split-is-phi/">04 Why the split is phi</a></li><li><a href="/learn/why-base-three/">05 Why base three</a></li><li><a href="/learn/phi-in-sixteen-bits/">06 Phi in sixteen bits</a></li></ol></li><li><h3>Module 3 · The small rungs: GF4 to GF8</h3><p>GF4, GF6 and GF8, the fewest bits, where rounding to whole bits costs the most.</p><ol><li><a href="/learn/gf4-four-bits/">07 GF4: four bits</a></li><li><a href="/learn/gf6-six-bits/">08 GF6: six bits</a></li><li><a href="/learn/gf8-one-byte/">09 GF8: one byte</a></li></ol></li><li><h3>Module 4 · Ten to fourteen bits</h3><p>GF10, GF12 and GF14, and how the distance from 1 / phi moves as the word grows.</p><ol><li><a href="/learn/gf10-ten-bits/">10 GF10: ten bits</a></li><li><a href="/learn/gf12-twelve-bits/">11 GF12: twelve bits</a></li><li><a href="/learn/gf14-fourteen-bits/">12 GF14: fourteen bits</a></li></ol></li><li><h3>Module 5 · GF16 at work</h3><p>The primary 16-bit format, a two-term dot product in GF-T16, then GF20 and GF24.</p><ol><li><a href="/learn/gf16-the-primary/">13 GF16: the primary format</a></li><li><a href="/learn/two-products-one-sum/">14 Two products, one sum</a></li><li><a href="/learn/gf24-twenty-four-bits/">15 GF20 and GF24</a></li></ol></li><li><h3>Module 6 · GF32 to GF64</h3><p>GF32 beside IEEE single, GF48 with no IEEE twin, GF64 beside IEEE double.</p><ol><li><a href="/learn/gf32-a-single/">16 GF32: a single</a></li><li><a href="/learn/gf48-forty-eight-bits/">17 GF48: forty-eight bits</a></li><li><a href="/learn/gf64-a-double/">18 GF64: a double</a></li></ol></li><li><h3>Module 7 · GF96 to GF256</h3><p>GF96, GF128 and GF256, where the specs hold the layout with invariants.</p><ol><li class="on" aria-current="page"><a href="/learn/gf96-ninety-six-bits/">19 GF96: ninety-six bits</a></li><li><a href="/learn/gf128-a-quad/">20 GF128: a quad</a></li><li><a href="/learn/gf256-two-five-six/">21 GF256: two hundred fifty-six bits</a></li></ol></li><li><h3>Module 8 · The widest rungs, then trits</h3><p>GF512 and GF1024, the two widest rungs, then GF-T8, where the exponent moves to trits.</p><ol><li><a href="/learn/gf512-five-twelve/">22 GF512: five hundred twelve bits</a></li><li><a href="/learn/gf1024-one-kilobit/">23 GF1024: one kilobit</a></li><li><a href="/learn/gft8-exponent-in-trits/">24 GF-T8: an exponent in trits</a></li></ol></li><li><h3>Module 9 · More trits, then the decode</h3><p>GF-T16 and GF-T32, then why fixed fields decode in parallel and a posit does not.</p><ol><li><a href="/learn/gft16-six-trits/">25 GF-T16: six trits</a></li><li><a href="/learn/gft32-twelve-trits/">26 GF-T32: twelve trits</a></li><li><a href="/learn/decode-in-one-step/">27 Decode in one step</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/gft16-six-trits/index.html b/apps/website/public/learn/gft16-six-trits/index.html
index 525813355d..19eba3ffe5 100644
--- a/apps/website/public/learn/gft16-six-trits/index.html
+++ b/apps/website/public/learn/gft16-six-trits/index.html
@@ -125,5 +125,6 @@ <h1>GF-T16: six trits</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The rule and its numbers</h3><p>One rule splits every width, the ratio it aims at, and the Lucas numbers behind the 3.</p><ol><li><a href="/learn/a-float-cut-by-phi/">01 A float cut by phi</a></li><li><a href="/learn/phi-as-a-ratio/">02 Phi as a ratio</a></li><li><a href="/learn/lucas-numbers-stay-whole/">03 Lucas numbers stay whole</a></li></ol></li><li><h3>Module 2 · Why phi, why three</h3><p>Why the split is phi, why base three, and how a spec checks GF16 keeps phi.</p><ol><li><a href="/learn/why-the-split-is-phi/">04 Why the split is phi</a></li><li><a href="/learn/why-base-three/">05 Why base three</a></li><li><a href="/learn/phi-in-sixteen-bits/">06 Phi in sixteen bits</a></li></ol></li><li><h3>Module 3 · The small rungs: GF4 to GF8</h3><p>GF4, GF6 and GF8, the fewest bits, where rounding to whole bits costs the most.</p><ol><li><a href="/learn/gf4-four-bits/">07 GF4: four bits</a></li><li><a href="/learn/gf6-six-bits/">08 GF6: six bits</a></li><li><a href="/learn/gf8-one-byte/">09 GF8: one byte</a></li></ol></li><li><h3>Module 4 · Ten to fourteen bits</h3><p>GF10, GF12 and GF14, and how the distance from 1 / phi moves as the word grows.</p><ol><li><a href="/learn/gf10-ten-bits/">10 GF10: ten bits</a></li><li><a href="/learn/gf12-twelve-bits/">11 GF12: twelve bits</a></li><li><a href="/learn/gf14-fourteen-bits/">12 GF14: fourteen bits</a></li></ol></li><li><h3>Module 5 · GF16 at work</h3><p>The primary 16-bit format, a two-term dot product in GF-T16, then GF20 and GF24.</p><ol><li><a href="/learn/gf16-the-primary/">13 GF16: the primary format</a></li><li><a href="/learn/two-products-one-sum/">14 Two products, one sum</a></li><li><a href="/learn/gf24-twenty-four-bits/">15 GF20 and GF24</a></li></ol></li><li><h3>Module 6 · GF32 to GF64</h3><p>GF32 beside IEEE single, GF48 with no IEEE twin, GF64 beside IEEE double.</p><ol><li><a href="/learn/gf32-a-single/">16 GF32: a single</a></li><li><a href="/learn/gf48-forty-eight-bits/">17 GF48: forty-eight bits</a></li><li><a href="/learn/gf64-a-double/">18 GF64: a double</a></li></ol></li><li><h3>Module 7 · GF96 to GF256</h3><p>GF96, GF128 and GF256, where the specs hold the layout with invariants.</p><ol><li><a href="/learn/gf96-ninety-six-bits/">19 GF96: ninety-six bits</a></li><li><a href="/learn/gf128-a-quad/">20 GF128: a quad</a></li><li><a href="/learn/gf256-two-five-six/">21 GF256: two hundred fifty-six bits</a></li></ol></li><li><h3>Module 8 · The widest rungs, then trits</h3><p>GF512 and GF1024, the two widest rungs, then GF-T8, where the exponent moves to trits.</p><ol><li><a href="/learn/gf512-five-twelve/">22 GF512: five hundred twelve bits</a></li><li><a href="/learn/gf1024-one-kilobit/">23 GF1024: one kilobit</a></li><li><a href="/learn/gft8-exponent-in-trits/">24 GF-T8: an exponent in trits</a></li></ol></li><li><h3>Module 9 · More trits, then the decode</h3><p>GF-T16 and GF-T32, then why fixed fields decode in parallel and a posit does not.</p><ol><li class="on" aria-current="page"><a href="/learn/gft16-six-trits/">25 GF-T16: six trits</a></li><li><a href="/learn/gft32-twelve-trits/">26 GF-T32: twelve trits</a></li><li><a href="/learn/decode-in-one-step/">27 Decode in one step</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/gft32-twelve-trits/index.html b/apps/website/public/learn/gft32-twelve-trits/index.html
index 8f14502f83..b848a1d30f 100644
--- a/apps/website/public/learn/gft32-twelve-trits/index.html
+++ b/apps/website/public/learn/gft32-twelve-trits/index.html
@@ -125,5 +125,6 @@ <h1>GF-T32: twelve trits</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The rule and its numbers</h3><p>One rule splits every width, the ratio it aims at, and the Lucas numbers behind the 3.</p><ol><li><a href="/learn/a-float-cut-by-phi/">01 A float cut by phi</a></li><li><a href="/learn/phi-as-a-ratio/">02 Phi as a ratio</a></li><li><a href="/learn/lucas-numbers-stay-whole/">03 Lucas numbers stay whole</a></li></ol></li><li><h3>Module 2 · Why phi, why three</h3><p>Why the split is phi, why base three, and how a spec checks GF16 keeps phi.</p><ol><li><a href="/learn/why-the-split-is-phi/">04 Why the split is phi</a></li><li><a href="/learn/why-base-three/">05 Why base three</a></li><li><a href="/learn/phi-in-sixteen-bits/">06 Phi in sixteen bits</a></li></ol></li><li><h3>Module 3 · The small rungs: GF4 to GF8</h3><p>GF4, GF6 and GF8, the fewest bits, where rounding to whole bits costs the most.</p><ol><li><a href="/learn/gf4-four-bits/">07 GF4: four bits</a></li><li><a href="/learn/gf6-six-bits/">08 GF6: six bits</a></li><li><a href="/learn/gf8-one-byte/">09 GF8: one byte</a></li></ol></li><li><h3>Module 4 · Ten to fourteen bits</h3><p>GF10, GF12 and GF14, and how the distance from 1 / phi moves as the word grows.</p><ol><li><a href="/learn/gf10-ten-bits/">10 GF10: ten bits</a></li><li><a href="/learn/gf12-twelve-bits/">11 GF12: twelve bits</a></li><li><a href="/learn/gf14-fourteen-bits/">12 GF14: fourteen bits</a></li></ol></li><li><h3>Module 5 · GF16 at work</h3><p>The primary 16-bit format, a two-term dot product in GF-T16, then GF20 and GF24.</p><ol><li><a href="/learn/gf16-the-primary/">13 GF16: the primary format</a></li><li><a href="/learn/two-products-one-sum/">14 Two products, one sum</a></li><li><a href="/learn/gf24-twenty-four-bits/">15 GF20 and GF24</a></li></ol></li><li><h3>Module 6 · GF32 to GF64</h3><p>GF32 beside IEEE single, GF48 with no IEEE twin, GF64 beside IEEE double.</p><ol><li><a href="/learn/gf32-a-single/">16 GF32: a single</a></li><li><a href="/learn/gf48-forty-eight-bits/">17 GF48: forty-eight bits</a></li><li><a href="/learn/gf64-a-double/">18 GF64: a double</a></li></ol></li><li><h3>Module 7 · GF96 to GF256</h3><p>GF96, GF128 and GF256, where the specs hold the layout with invariants.</p><ol><li><a href="/learn/gf96-ninety-six-bits/">19 GF96: ninety-six bits</a></li><li><a href="/learn/gf128-a-quad/">20 GF128: a quad</a></li><li><a href="/learn/gf256-two-five-six/">21 GF256: two hundred fifty-six bits</a></li></ol></li><li><h3>Module 8 · The widest rungs, then trits</h3><p>GF512 and GF1024, the two widest rungs, then GF-T8, where the exponent moves to trits.</p><ol><li><a href="/learn/gf512-five-twelve/">22 GF512: five hundred twelve bits</a></li><li><a href="/learn/gf1024-one-kilobit/">23 GF1024: one kilobit</a></li><li><a href="/learn/gft8-exponent-in-trits/">24 GF-T8: an exponent in trits</a></li></ol></li><li><h3>Module 9 · More trits, then the decode</h3><p>GF-T16 and GF-T32, then why fixed fields decode in parallel and a posit does not.</p><ol><li><a href="/learn/gft16-six-trits/">25 GF-T16: six trits</a></li><li class="on" aria-current="page"><a href="/learn/gft32-twelve-trits/">26 GF-T32: twelve trits</a></li><li><a href="/learn/decode-in-one-step/">27 Decode in one step</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/gft8-exponent-in-trits/index.html b/apps/website/public/learn/gft8-exponent-in-trits/index.html
index af3b8cce28..6424258010 100644
--- a/apps/website/public/learn/gft8-exponent-in-trits/index.html
+++ b/apps/website/public/learn/gft8-exponent-in-trits/index.html
@@ -125,5 +125,6 @@ <h1>GF-T8: an exponent in trits</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The rule and its numbers</h3><p>One rule splits every width, the ratio it aims at, and the Lucas numbers behind the 3.</p><ol><li><a href="/learn/a-float-cut-by-phi/">01 A float cut by phi</a></li><li><a href="/learn/phi-as-a-ratio/">02 Phi as a ratio</a></li><li><a href="/learn/lucas-numbers-stay-whole/">03 Lucas numbers stay whole</a></li></ol></li><li><h3>Module 2 · Why phi, why three</h3><p>Why the split is phi, why base three, and how a spec checks GF16 keeps phi.</p><ol><li><a href="/learn/why-the-split-is-phi/">04 Why the split is phi</a></li><li><a href="/learn/why-base-three/">05 Why base three</a></li><li><a href="/learn/phi-in-sixteen-bits/">06 Phi in sixteen bits</a></li></ol></li><li><h3>Module 3 · The small rungs: GF4 to GF8</h3><p>GF4, GF6 and GF8, the fewest bits, where rounding to whole bits costs the most.</p><ol><li><a href="/learn/gf4-four-bits/">07 GF4: four bits</a></li><li><a href="/learn/gf6-six-bits/">08 GF6: six bits</a></li><li><a href="/learn/gf8-one-byte/">09 GF8: one byte</a></li></ol></li><li><h3>Module 4 · Ten to fourteen bits</h3><p>GF10, GF12 and GF14, and how the distance from 1 / phi moves as the word grows.</p><ol><li><a href="/learn/gf10-ten-bits/">10 GF10: ten bits</a></li><li><a href="/learn/gf12-twelve-bits/">11 GF12: twelve bits</a></li><li><a href="/learn/gf14-fourteen-bits/">12 GF14: fourteen bits</a></li></ol></li><li><h3>Module 5 · GF16 at work</h3><p>The primary 16-bit format, a two-term dot product in GF-T16, then GF20 and GF24.</p><ol><li><a href="/learn/gf16-the-primary/">13 GF16: the primary format</a></li><li><a href="/learn/two-products-one-sum/">14 Two products, one sum</a></li><li><a href="/learn/gf24-twenty-four-bits/">15 GF20 and GF24</a></li></ol></li><li><h3>Module 6 · GF32 to GF64</h3><p>GF32 beside IEEE single, GF48 with no IEEE twin, GF64 beside IEEE double.</p><ol><li><a href="/learn/gf32-a-single/">16 GF32: a single</a></li><li><a href="/learn/gf48-forty-eight-bits/">17 GF48: forty-eight bits</a></li><li><a href="/learn/gf64-a-double/">18 GF64: a double</a></li></ol></li><li><h3>Module 7 · GF96 to GF256</h3><p>GF96, GF128 and GF256, where the specs hold the layout with invariants.</p><ol><li><a href="/learn/gf96-ninety-six-bits/">19 GF96: ninety-six bits</a></li><li><a href="/learn/gf128-a-quad/">20 GF128: a quad</a></li><li><a href="/learn/gf256-two-five-six/">21 GF256: two hundred fifty-six bits</a></li></ol></li><li><h3>Module 8 · The widest rungs, then trits</h3><p>GF512 and GF1024, the two widest rungs, then GF-T8, where the exponent moves to trits.</p><ol><li><a href="/learn/gf512-five-twelve/">22 GF512: five hundred twelve bits</a></li><li><a href="/learn/gf1024-one-kilobit/">23 GF1024: one kilobit</a></li><li class="on" aria-current="page"><a href="/learn/gft8-exponent-in-trits/">24 GF-T8: an exponent in trits</a></li></ol></li><li><h3>Module 9 · More trits, then the decode</h3><p>GF-T16 and GF-T32, then why fixed fields decode in parallel and a posit does not.</p><ol><li><a href="/learn/gft16-six-trits/">25 GF-T16: six trits</a></li><li><a href="/learn/gft32-twelve-trits/">26 GF-T32: twelve trits</a></li><li><a href="/learn/decode-in-one-step/">27 Decode in one step</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/global-buffers/index.html b/apps/website/public/learn/global-buffers/index.html
index 0e962f1cb3..9e9c13a5eb 100644
--- a/apps/website/public/learn/global-buffers/index.html
+++ b/apps/website/public/learn/global-buffers/index.html
@@ -294,5 +294,6 @@ <h1>The global buffer</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a clock is</h3><p>One edge, one world: what shares a clock edge shares a world, the period and the jitter of a real edge, and where the clock enters a board.</p><ol><li><a href="/learn/one-edge-one-world/">01 One edge, one world</a></li><li><a href="/learn/period-and-jitter/">02 A period, an edge, and jitter</a></li><li><a href="/learn/where-the-clock-enters/">03 Where the clock enters</a></li></ol></li><li><h3>Module 2 · Clock trees</h3><p>Skew and insertion delay, the global buffer network, and the trap of gating a clock with logic.</p><ol><li><a href="/learn/skew-and-insertion/">04 Skew and insertion delay</a></li><li class="on" aria-current="page"><a href="/learn/global-buffers/">05 The global buffer</a></li><li><a href="/learn/gated-clock-trap/">06 The gated-clock trap</a></li></ol></li><li><h3>Module 3 · PLL and MMCM</h3><p>Multiply and divide one clock into another, move its phase in steps of the VCO, and which clocks the analyzer treats as related.</p><ol><li><a href="/learn/multiply-and-divide/">07 Multiply and divide</a></li><li><a href="/learn/phase-and-lock/">08 Phase and lock</a></li><li><a href="/learn/related-or-not/">09 Related or not</a></li></ol></li><li><h3>Module 4 · Resets</h3><p>Assert asynchronously, release synchronously: the three reset kinds, the release pipe, and the tree a reset grows.</p><ol><li><a href="/learn/sync-or-async/">10 Sync or async</a></li><li><a href="/learn/reset-release/">11 The release pipe</a></li><li><a href="/learn/reset-fanout/">12 Reset fanout</a></li></ol></li><li><h3>Module 5 · Metastability</h3><p>The setup-hold window, the mean time between failures in integer arithmetic, and the two flops that fix it.</p><ol><li><a href="/learn/setup-hold-window/">13 The setup-hold window</a></li><li><a href="/learn/mtbf/">14 Mean time between failures</a></li><li><a href="/learn/two-flop-sync/">15 The two-flop synchronizer</a></li></ol></li><li><h3>Module 6 · Crossing many bits</h3><p>Why a binary bus tears, why Gray code does not, and the handshake that moves a pulse between worlds.</p><ol><li><a href="/learn/why-binary-fails/">16 Why binary fails</a></li><li><a href="/learn/gray-code/">17 Gray code</a></li><li><a href="/learn/handshakes-and-pulses/">18 Handshakes and pulses</a></li></ol></li><li><h3>Module 7 · The asynchronous FIFO</h3><p>Pointers, flags and depth: the buffer that moves a stream between two clocks.</p><ol><li><a href="/learn/pointers-and-flags/">19 Pointers and flags</a></li><li><a href="/learn/depth-sizing/">20 Sizing the depth</a></li><li><a href="/learn/fifo-testbench/">21 The FIFO testbench</a></li></ol></li><li><h3>Module 8 · Constraints</h3><p>The lines that tell the analyzer what a clock is, which paths not to check, and what the pins must meet.</p><ol><li><a href="/learn/create-clock/">22 create_clock</a></li><li><a href="/learn/false-path-and-max-delay/">23 False paths and max delay</a></li><li><a href="/learn/io-timing/">24 IO timing</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>A CDC report, one crossing captured at the flip-flops, and the bitstream diff that closes the course.</p><ol><li><a href="/learn/cdc-report/">25 The CDC report</a></li><li><a href="/learn/capture-a-crossing/">26 Capture a crossing</a></li><li><a href="/learn/capstone-diff/">27 Capstone: change it and diff it</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/golden-model/index.html b/apps/website/public/learn/golden-model/index.html
index ea202ef49e..b3909f4590 100644
--- a/apps/website/public/learn/golden-model/index.html
+++ b/apps/website/public/learn/golden-model/index.html
@@ -360,5 +360,6 @@ <h1>The golden model</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Why verify</h3><p>Designs that compile and are wrong, the model that decides, and the plan written before the code.</p><ol><li><a href="/learn/bugs-that-compile/">01 Bugs that compile</a></li><li class="on" aria-current="page"><a href="/learn/golden-model/">02 The golden model</a></li><li><a href="/learn/plan-before-code/">03 Plan before code</a></li></ol></li><li><h3>Module 2 · Testbenches</h3><p>Stimulus, checks and a verdict, written as one spec beside the design it judges.</p><ol><li><a href="/learn/anatomy-of-a-tb/">04 Anatomy of a testbench</a></li><li><a href="/learn/self-checking/">05 Self-checking, not staring</a></li><li><a href="/learn/directed-vs-random/">06 Directed versus random</a></li></ol></li><li><h3>Module 3 · Waveforms</h3><p>A trace of every signal, read the way a hardware engineer reads it, and two runs compared.</p><ol><li><a href="/learn/vcd-format/">07 The VCD format</a></li><li><a href="/learn/debug-with-waves/">08 Debug with waves</a></li><li><a href="/learn/compare-traces/">09 Compare two traces</a></li></ol></li><li><h3>Module 4 · Conformance vectors</h3><p>Cases with the answer written beside them, kept where the compiler can reach them.</p><ol><li><a href="/learn/what-a-vector-is/">10 What a vector is</a></li><li><a href="/learn/golden-ruler-vectors/">11 Golden ruler vectors</a></li><li><a href="/learn/end-to-end-scenarios/">12 End-to-end scenarios</a></li></ol></li><li><h3>Module 5 · Cosimulation</h3><p>Spec, simulator and board agreeing on the bench Artix-7 XC7A200T, and what to do when they do not.</p><ol><li><a href="/learn/three-way-agreement/">13 Three-way agreement</a></li><li><a href="/learn/simulator-internals/">14 Inside the simulator</a></li><li><a href="/learn/when-they-disagree/">15 When they disagree</a></li></ol></li><li><h3>Module 6 · Coverage</h3><p>What the tests touched: lines, toggles, states, and what that number hides.</p><ol><li><a href="/learn/line-and-toggle/">16 Line and toggle</a></li><li><a href="/learn/fsm-coverage/">17 FSM coverage</a></li><li><a href="/learn/coverage-lies/">18 Coverage lies</a></li></ol></li><li><h3>Module 7 · Formal</h3><p>Assertions that hold every cycle, bounded search for a counterexample, and why a proof needs induction.</p><ol><li><a href="/learn/assertions/">19 Assertions</a></li><li><a href="/learn/bounded-model-checking/">20 Bounded model checking</a></li><li><a href="/learn/induction/">21 Induction</a></li></ol></li><li><h3>Module 8 · Mutation</h3><p>Break the design on purpose and count what the tests catch.</p><ol><li><a href="/learn/kill-the-mutant/">22 Kill the mutant</a></li><li><a href="/learn/surviving-mutants/">23 Surviving mutants</a></li><li><a href="/learn/test-strength/">24 Test strength</a></li></ol></li><li><h3>Module 9 · Sign-off</h3><p>One command, every receipt, a clean verdict you can show.</p><ol><li><a href="/learn/build-verify/">25 Build and verify</a></li><li><a href="/learn/board-selftest/">26 The board self-tests</a></li><li><a href="/learn/verify-capstone/">27 Capstone: verify a module whole</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/golden-ruler-vectors/index.html b/apps/website/public/learn/golden-ruler-vectors/index.html
index 3ef6bb57ff..aeb84dc0ba 100644
--- a/apps/website/public/learn/golden-ruler-vectors/index.html
+++ b/apps/website/public/learn/golden-ruler-vectors/index.html
@@ -565,5 +565,6 @@ <h1>Golden ruler vectors</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Why verify</h3><p>Designs that compile and are wrong, the model that decides, and the plan written before the code.</p><ol><li><a href="/learn/bugs-that-compile/">01 Bugs that compile</a></li><li><a href="/learn/golden-model/">02 The golden model</a></li><li><a href="/learn/plan-before-code/">03 Plan before code</a></li></ol></li><li><h3>Module 2 · Testbenches</h3><p>Stimulus, checks and a verdict, written as one spec beside the design it judges.</p><ol><li><a href="/learn/anatomy-of-a-tb/">04 Anatomy of a testbench</a></li><li><a href="/learn/self-checking/">05 Self-checking, not staring</a></li><li><a href="/learn/directed-vs-random/">06 Directed versus random</a></li></ol></li><li><h3>Module 3 · Waveforms</h3><p>A trace of every signal, read the way a hardware engineer reads it, and two runs compared.</p><ol><li><a href="/learn/vcd-format/">07 The VCD format</a></li><li><a href="/learn/debug-with-waves/">08 Debug with waves</a></li><li><a href="/learn/compare-traces/">09 Compare two traces</a></li></ol></li><li><h3>Module 4 · Conformance vectors</h3><p>Cases with the answer written beside them, kept where the compiler can reach them.</p><ol><li><a href="/learn/what-a-vector-is/">10 What a vector is</a></li><li class="on" aria-current="page"><a href="/learn/golden-ruler-vectors/">11 Golden ruler vectors</a></li><li><a href="/learn/end-to-end-scenarios/">12 End-to-end scenarios</a></li></ol></li><li><h3>Module 5 · Cosimulation</h3><p>Spec, simulator and board agreeing on the bench Artix-7 XC7A200T, and what to do when they do not.</p><ol><li><a href="/learn/three-way-agreement/">13 Three-way agreement</a></li><li><a href="/learn/simulator-internals/">14 Inside the simulator</a></li><li><a href="/learn/when-they-disagree/">15 When they disagree</a></li></ol></li><li><h3>Module 6 · Coverage</h3><p>What the tests touched: lines, toggles, states, and what that number hides.</p><ol><li><a href="/learn/line-and-toggle/">16 Line and toggle</a></li><li><a href="/learn/fsm-coverage/">17 FSM coverage</a></li><li><a href="/learn/coverage-lies/">18 Coverage lies</a></li></ol></li><li><h3>Module 7 · Formal</h3><p>Assertions that hold every cycle, bounded search for a counterexample, and why a proof needs induction.</p><ol><li><a href="/learn/assertions/">19 Assertions</a></li><li><a href="/learn/bounded-model-checking/">20 Bounded model checking</a></li><li><a href="/learn/induction/">21 Induction</a></li></ol></li><li><h3>Module 8 · Mutation</h3><p>Break the design on purpose and count what the tests catch.</p><ol><li><a href="/learn/kill-the-mutant/">22 Kill the mutant</a></li><li><a href="/learn/surviving-mutants/">23 Surviving mutants</a></li><li><a href="/learn/test-strength/">24 Test strength</a></li></ol></li><li><h3>Module 9 · Sign-off</h3><p>One command, every receipt, a clean verdict you can show.</p><ol><li><a href="/learn/build-verify/">25 Build and verify</a></li><li><a href="/learn/board-selftest/">26 The board self-tests</a></li><li><a href="/learn/verify-capstone/">27 Capstone: verify a module whole</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/goldenfloat/index.html b/apps/website/public/learn/goldenfloat/index.html
index 93c89ac7f8..520744b9ee 100644
--- a/apps/website/public/learn/goldenfloat/index.html
+++ b/apps/website/public/learn/goldenfloat/index.html
@@ -67,5 +67,6 @@ <h1>GoldenFloat with t27: one rule, seventeen float widths</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The rule and its numbers</h3><p>One rule splits every width, the ratio it aims at, and the Lucas numbers behind the 3.</p><ol><li><a href="/learn/a-float-cut-by-phi/">01 A float cut by phi</a></li><li><a href="/learn/phi-as-a-ratio/">02 Phi as a ratio</a></li><li><a href="/learn/lucas-numbers-stay-whole/">03 Lucas numbers stay whole</a></li></ol></li><li><h3>Module 2 · Why phi, why three</h3><p>Why the split is phi, why base three, and how a spec checks GF16 keeps phi.</p><ol><li><a href="/learn/why-the-split-is-phi/">04 Why the split is phi</a></li><li><a href="/learn/why-base-three/">05 Why base three</a></li><li><a href="/learn/phi-in-sixteen-bits/">06 Phi in sixteen bits</a></li></ol></li><li><h3>Module 3 · The small rungs: GF4 to GF8</h3><p>GF4, GF6 and GF8, the fewest bits, where rounding to whole bits costs the most.</p><ol><li><a href="/learn/gf4-four-bits/">07 GF4: four bits</a></li><li><a href="/learn/gf6-six-bits/">08 GF6: six bits</a></li><li><a href="/learn/gf8-one-byte/">09 GF8: one byte</a></li></ol></li><li><h3>Module 4 · Ten to fourteen bits</h3><p>GF10, GF12 and GF14, and how the distance from 1 / phi moves as the word grows.</p><ol><li><a href="/learn/gf10-ten-bits/">10 GF10: ten bits</a></li><li><a href="/learn/gf12-twelve-bits/">11 GF12: twelve bits</a></li><li><a href="/learn/gf14-fourteen-bits/">12 GF14: fourteen bits</a></li></ol></li><li><h3>Module 5 · GF16 at work</h3><p>The primary 16-bit format, a two-term dot product in GF-T16, then GF20 and GF24.</p><ol><li><a href="/learn/gf16-the-primary/">13 GF16: the primary format</a></li><li><a href="/learn/two-products-one-sum/">14 Two products, one sum</a></li><li><a href="/learn/gf24-twenty-four-bits/">15 GF20 and GF24</a></li></ol></li><li><h3>Module 6 · GF32 to GF64</h3><p>GF32 beside IEEE single, GF48 with no IEEE twin, GF64 beside IEEE double.</p><ol><li><a href="/learn/gf32-a-single/">16 GF32: a single</a></li><li><a href="/learn/gf48-forty-eight-bits/">17 GF48: forty-eight bits</a></li><li><a href="/learn/gf64-a-double/">18 GF64: a double</a></li></ol></li><li><h3>Module 7 · GF96 to GF256</h3><p>GF96, GF128 and GF256, where the specs hold the layout with invariants.</p><ol><li><a href="/learn/gf96-ninety-six-bits/">19 GF96: ninety-six bits</a></li><li><a href="/learn/gf128-a-quad/">20 GF128: a quad</a></li><li><a href="/learn/gf256-two-five-six/">21 GF256: two hundred fifty-six bits</a></li></ol></li><li><h3>Module 8 · The widest rungs, then trits</h3><p>GF512 and GF1024, the two widest rungs, then GF-T8, where the exponent moves to trits.</p><ol><li><a href="/learn/gf512-five-twelve/">22 GF512: five hundred twelve bits</a></li><li><a href="/learn/gf1024-one-kilobit/">23 GF1024: one kilobit</a></li><li><a href="/learn/gft8-exponent-in-trits/">24 GF-T8: an exponent in trits</a></li></ol></li><li><h3>Module 9 · More trits, then the decode</h3><p>GF-T16 and GF-T32, then why fixed fields decode in parallel and a posit does not.</p><ol><li><a href="/learn/gft16-six-trits/">25 GF-T16: six trits</a></li><li><a href="/learn/gft32-twelve-trits/">26 GF-T32: twelve trits</a></li><li><a href="/learn/decode-in-one-step/">27 Decode in one step</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/gray-code/index.html b/apps/website/public/learn/gray-code/index.html
index ac95e300c2..ca7275406c 100644
--- a/apps/website/public/learn/gray-code/index.html
+++ b/apps/website/public/learn/gray-code/index.html
@@ -434,5 +434,6 @@ <h1>Gray code</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a clock is</h3><p>One edge, one world: what shares a clock edge shares a world, the period and the jitter of a real edge, and where the clock enters a board.</p><ol><li><a href="/learn/one-edge-one-world/">01 One edge, one world</a></li><li><a href="/learn/period-and-jitter/">02 A period, an edge, and jitter</a></li><li><a href="/learn/where-the-clock-enters/">03 Where the clock enters</a></li></ol></li><li><h3>Module 2 · Clock trees</h3><p>Skew and insertion delay, the global buffer network, and the trap of gating a clock with logic.</p><ol><li><a href="/learn/skew-and-insertion/">04 Skew and insertion delay</a></li><li><a href="/learn/global-buffers/">05 The global buffer</a></li><li><a href="/learn/gated-clock-trap/">06 The gated-clock trap</a></li></ol></li><li><h3>Module 3 · PLL and MMCM</h3><p>Multiply and divide one clock into another, move its phase in steps of the VCO, and which clocks the analyzer treats as related.</p><ol><li><a href="/learn/multiply-and-divide/">07 Multiply and divide</a></li><li><a href="/learn/phase-and-lock/">08 Phase and lock</a></li><li><a href="/learn/related-or-not/">09 Related or not</a></li></ol></li><li><h3>Module 4 · Resets</h3><p>Assert asynchronously, release synchronously: the three reset kinds, the release pipe, and the tree a reset grows.</p><ol><li><a href="/learn/sync-or-async/">10 Sync or async</a></li><li><a href="/learn/reset-release/">11 The release pipe</a></li><li><a href="/learn/reset-fanout/">12 Reset fanout</a></li></ol></li><li><h3>Module 5 · Metastability</h3><p>The setup-hold window, the mean time between failures in integer arithmetic, and the two flops that fix it.</p><ol><li><a href="/learn/setup-hold-window/">13 The setup-hold window</a></li><li><a href="/learn/mtbf/">14 Mean time between failures</a></li><li><a href="/learn/two-flop-sync/">15 The two-flop synchronizer</a></li></ol></li><li><h3>Module 6 · Crossing many bits</h3><p>Why a binary bus tears, why Gray code does not, and the handshake that moves a pulse between worlds.</p><ol><li><a href="/learn/why-binary-fails/">16 Why binary fails</a></li><li class="on" aria-current="page"><a href="/learn/gray-code/">17 Gray code</a></li><li><a href="/learn/handshakes-and-pulses/">18 Handshakes and pulses</a></li></ol></li><li><h3>Module 7 · The asynchronous FIFO</h3><p>Pointers, flags and depth: the buffer that moves a stream between two clocks.</p><ol><li><a href="/learn/pointers-and-flags/">19 Pointers and flags</a></li><li><a href="/learn/depth-sizing/">20 Sizing the depth</a></li><li><a href="/learn/fifo-testbench/">21 The FIFO testbench</a></li></ol></li><li><h3>Module 8 · Constraints</h3><p>The lines that tell the analyzer what a clock is, which paths not to check, and what the pins must meet.</p><ol><li><a href="/learn/create-clock/">22 create_clock</a></li><li><a href="/learn/false-path-and-max-delay/">23 False paths and max delay</a></li><li><a href="/learn/io-timing/">24 IO timing</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>A CDC report, one crossing captured at the flip-flops, and the bitstream diff that closes the course.</p><ol><li><a href="/learn/cdc-report/">25 The CDC report</a></li><li><a href="/learn/capture-a-crossing/">26 Capture a crossing</a></li><li><a href="/learn/capstone-diff/">27 Capstone: change it and diff it</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/growth-through-the-stages/index.html b/apps/website/public/learn/growth-through-the-stages/index.html
index f1e1d75de1..6412971106 100644
--- a/apps/website/public/learn/growth-through-the-stages/index.html
+++ b/apps/website/public/learn/growth-through-the-stages/index.html
@@ -1122,5 +1122,6 @@ <h1>Growth through the stages</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Adding numbers</h3><p>Why addition is slow: the carry that walks up the word, the dedicated chain in every slice, prefix networks and carry-save trees.</p><ol><li><a href="/learn/the-carry-chain/">01 The carry chain</a></li><li><a href="/learn/prefix-adders/">02 Looking ahead: prefix adders</a></li><li><a href="/learn/carry-save/">03 Three numbers at once</a></li></ol></li><li><h3>Module 2 · Multiplying</h3><p>A product is a sum of shifted copies; Booth recoding halves them, and the DSP48E1 slice does the rest in one block.</p><ol><li><a href="/learn/shift-and-add/">04 Shift and add</a></li><li><a href="/learn/booths-trick/">05 Booth's trick</a></li><li><a href="/learn/the-dsp48e1-slice/">06 The DSP48E1 slice</a></li></ol></li><li><h3>Module 3 · Fixed point</h3><p>Where the binary point sits, what rounding does to a value and to its average, and what each bit of a quantizer buys.</p><ol><li><a href="/learn/q-formats/">07 Q formats</a></li><li><a href="/learn/rounding-and-saturation/">08 Rounding and saturation</a></li><li><a href="/learn/quantization-noise/">09 Quantization noise</a></li></ol></li><li><h3>Module 4 · Functions in hardware</h3><p>Sine, cosine, angle and length from shifts and adds, and when a table is the better answer.</p><ol><li><a href="/learn/rotating-by-shifts/">10 Rotating by shifts</a></li><li><a href="/learn/angle-and-magnitude/">11 Angle and magnitude</a></li><li><a href="/learn/tables-and-interpolation/">12 Tables and interpolation</a></li></ol></li><li><h3>Module 5 · Signals and sampling</h3><p>What sampling does to a frequency, how an oscillator is built from an adder, and what a DFT bin measures.</p><ol><li><a href="/learn/sampling-and-aliasing/">13 Sampling and aliasing</a></li><li><a href="/learn/the-nco/">14 A numerically controlled oscillator</a></li><li><a href="/learn/bins-of-a-spectrum/">15 Bins of a spectrum</a></li></ol></li><li><h3>Module 6 · FIR filters</h3><p>The moving average, a windowed-sinc design with integer taps, and folding the taps onto DSP slices.</p><ol><li><a href="/learn/the-moving-average/">16 The moving average</a></li><li><a href="/learn/designing-taps/">17 Designing taps</a></li><li><a href="/learn/folding-taps-onto-slices/">18 Folding taps onto slices</a></li></ol></li><li><h3>Module 7 · Multirate</h3><p>Lowering the sample rate without folding noise in: decimation, the CIC filter and the polyphase form.</p><ol><li><a href="/learn/lowering-the-rate/">19 Lowering the rate</a></li><li><a href="/learn/the-cic-filter/">20 The CIC filter</a></li><li><a href="/learn/polyphase/">21 Polyphase</a></li></ol></li><li><h3>Module 8 · The FFT</h3><p>N log N instead of N^2: butterflies, the bit-reversed input order and the bits each stage adds.</p><ol><li><a href="/learn/the-butterfly/">22 The butterfly</a></li><li><a href="/learn/twiddles-and-bit-reversal/">23 Twiddles and bit reversal</a></li><li class="on" aria-current="page"><a href="/learn/growth-through-the-stages/">24 Growth through the stages</a></li></ol></li><li><h3>Module 9 · On the bench</h3><p>From the arithmetic to the board: a correlator from the modem spec, a budget of slices and timing, and the capstone filter.</p><ol><li><a href="/learn/a-correlator-finds-the-frame/">25 A correlator finds the frame</a></li><li><a href="/learn/does-it-fit-and-close-timing/">26 Does it fit and close timing?</a></li><li><a href="/learn/capstone-a-filter-on-the-board/">27 Capstone: a filter on the board</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/half-adder/index.html b/apps/website/public/learn/half-adder/index.html
index 8c151336e8..d15c9b115f 100644
--- a/apps/website/public/learn/half-adder/index.html
+++ b/apps/website/public/learn/half-adder/index.html
@@ -769,5 +769,6 @@ <h1>The half adder</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Three values</h3><p>Why three, how balanced ternary writes every number without a sign, and what flipping and cutting trits do.</p><ol><li><a href="/learn/why-three/">01 Why three</a></li><li><a href="/learn/balanced-ternary/">02 Balanced ternary</a></li><li><a href="/learn/flip-and-cut/">03 Flip to negate, cut to round</a></li></ol></li><li><h3>Module 2 · Logic with unknown</h3><p>Kleene's three-valued gates, two gates binary has no twin for, and addition as a pair of tables.</p><ol><li><a href="/learn/kleene-logic/">04 Three-valued logic</a></li><li><a href="/learn/consensus-and-any/">05 Consensus and any</a></li><li><a href="/learn/sum-and-carry/">06 Sum and carry</a></li></ol></li><li><h3>Module 3 · Adding trits</h3><p>The half adder, the full adder and a carry that ripples left, one place at a time.</p><ol><li class="on" aria-current="page"><a href="/learn/half-adder/">07 The half adder</a></li><li><a href="/learn/full-adder/">08 The full adder</a></li><li><a href="/learn/ripple-carry/">09 Ripple carry</a></li></ol></li><li><h3>Module 4 · Multiplying without a multiplier</h3><p>Copy, drop or flip: a product by one trit, long multiplication, and a MAC that only adds.</p><ol><li><a href="/learn/times-a-trit/">10 Times one trit</a></li><li><a href="/learn/long-multiplication/">11 Long multiplication</a></li><li><a href="/learn/ternary-mac/">12 A ternary MAC</a></li></ol></li><li><h3>Module 5 · Trits in binary memory</h3><p>Two bits per trit, five trits per byte, and the bits a 27-trit word needs.</p><ol><li><a href="/learn/two-bits-per-trit/">13 Two bits per trit</a></li><li><a href="/learn/five-trits-a-byte/">14 Five trits in a byte</a></li><li><a href="/learn/a-word-of-27/">15 A word of 27 trits</a></li></ol></li><li><h3>Module 6 · The TRI-27 instruction word</h3><p>The 32-bit word the Trinity emulator decodes: its fields, its 47 opcodes and its 15-bit immediate.</p><ol><li><a href="/learn/the-instruction-word/">16 The instruction word</a></li><li><a href="/learn/the-opcode-map/">17 The opcode map</a></li><li><a href="/learn/the-immediate/">18 The immediate</a></li></ol></li><li><h3>Module 7 · A ternary machine</h3><p>An ALU built from this course's adders, a three-way jump, and a program you can step.</p><ol><li><a href="/learn/a-ternary-alu/">19 A ternary ALU</a></li><li><a href="/learn/three-way-jump/">20 A three-way jump</a></li><li><a href="/learn/run-a-program/">21 Run a program</a></li></ol></li><li><h3>Module 8 · Three answers</h3><p>Compare at the highest differing trit, find a number in thirds, sort with three-way compares.</p><ol><li><a href="/learn/compare-at-the-top/">22 Compare at the top</a></li><li><a href="/learn/questions-in-thirds/">23 Questions in thirds</a></li><li><a href="/learn/sort-by-three/">24 Sort with three-way compares</a></li></ol></li><li><h3>Module 9 · Ternary neurons</h3><p>Weights of -1, 0, +1: one neuron, a detector and a small layer, with no multiplier anywhere.</p><ol><li><a href="/learn/one-ternary-neuron/">25 One ternary neuron</a></li><li><a href="/learn/a-bar-detector/">26 A bar detector</a></li><li><a href="/learn/a-tiny-layer/">27 A tiny layer</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/handshakes-and-pulses/index.html b/apps/website/public/learn/handshakes-and-pulses/index.html
index e7d8e85733..c518d0a9e9 100644
--- a/apps/website/public/learn/handshakes-and-pulses/index.html
+++ b/apps/website/public/learn/handshakes-and-pulses/index.html
@@ -300,5 +300,6 @@ <h1>Handshakes and pulses</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a clock is</h3><p>One edge, one world: what shares a clock edge shares a world, the period and the jitter of a real edge, and where the clock enters a board.</p><ol><li><a href="/learn/one-edge-one-world/">01 One edge, one world</a></li><li><a href="/learn/period-and-jitter/">02 A period, an edge, and jitter</a></li><li><a href="/learn/where-the-clock-enters/">03 Where the clock enters</a></li></ol></li><li><h3>Module 2 · Clock trees</h3><p>Skew and insertion delay, the global buffer network, and the trap of gating a clock with logic.</p><ol><li><a href="/learn/skew-and-insertion/">04 Skew and insertion delay</a></li><li><a href="/learn/global-buffers/">05 The global buffer</a></li><li><a href="/learn/gated-clock-trap/">06 The gated-clock trap</a></li></ol></li><li><h3>Module 3 · PLL and MMCM</h3><p>Multiply and divide one clock into another, move its phase in steps of the VCO, and which clocks the analyzer treats as related.</p><ol><li><a href="/learn/multiply-and-divide/">07 Multiply and divide</a></li><li><a href="/learn/phase-and-lock/">08 Phase and lock</a></li><li><a href="/learn/related-or-not/">09 Related or not</a></li></ol></li><li><h3>Module 4 · Resets</h3><p>Assert asynchronously, release synchronously: the three reset kinds, the release pipe, and the tree a reset grows.</p><ol><li><a href="/learn/sync-or-async/">10 Sync or async</a></li><li><a href="/learn/reset-release/">11 The release pipe</a></li><li><a href="/learn/reset-fanout/">12 Reset fanout</a></li></ol></li><li><h3>Module 5 · Metastability</h3><p>The setup-hold window, the mean time between failures in integer arithmetic, and the two flops that fix it.</p><ol><li><a href="/learn/setup-hold-window/">13 The setup-hold window</a></li><li><a href="/learn/mtbf/">14 Mean time between failures</a></li><li><a href="/learn/two-flop-sync/">15 The two-flop synchronizer</a></li></ol></li><li><h3>Module 6 · Crossing many bits</h3><p>Why a binary bus tears, why Gray code does not, and the handshake that moves a pulse between worlds.</p><ol><li><a href="/learn/why-binary-fails/">16 Why binary fails</a></li><li><a href="/learn/gray-code/">17 Gray code</a></li><li class="on" aria-current="page"><a href="/learn/handshakes-and-pulses/">18 Handshakes and pulses</a></li></ol></li><li><h3>Module 7 · The asynchronous FIFO</h3><p>Pointers, flags and depth: the buffer that moves a stream between two clocks.</p><ol><li><a href="/learn/pointers-and-flags/">19 Pointers and flags</a></li><li><a href="/learn/depth-sizing/">20 Sizing the depth</a></li><li><a href="/learn/fifo-testbench/">21 The FIFO testbench</a></li></ol></li><li><h3>Module 8 · Constraints</h3><p>The lines that tell the analyzer what a clock is, which paths not to check, and what the pins must meet.</p><ol><li><a href="/learn/create-clock/">22 create_clock</a></li><li><a href="/learn/false-path-and-max-delay/">23 False paths and max delay</a></li><li><a href="/learn/io-timing/">24 IO timing</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>A CDC report, one crossing captured at the flip-flops, and the bitstream diff that closes the course.</p><ol><li><a href="/learn/cdc-report/">25 The CDC report</a></li><li><a href="/learn/capture-a-crossing/">26 Capture a crossing</a></li><li><a href="/learn/capstone-diff/">27 Capstone: change it and diff it</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/hello-t27/index.html b/apps/website/public/learn/hello-t27/index.html
index 5c5317dc33..dc428ed9ea 100644
--- a/apps/website/public/learn/hello-t27/index.html
+++ b/apps/website/public/learn/hello-t27/index.html
@@ -153,5 +153,6 @@ <h1>Hello, t27</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The chip</h3><p>What an FPGA is, which chip we use, and how its pins meet the board.</p><ol><li><a href="/learn/fpga-is-a-grid/">01 An FPGA is a grid you configure</a></li><li><a href="/learn/which-chip/">02 Which chip is this?</a></li><li><a href="/learn/pins-and-package/">03 Pins, balls and the package</a></li></ol></li><li><h3>Module 2 · Numbers in hardware</h3><p>Bits, trits and number formats, and what arithmetic costs in logic.</p><ol><li><a href="/learn/bits-and-trits/">04 Bits, trits and GF16</a></li><li><a href="/learn/what-arithmetic-costs/">05 What arithmetic costs</a></li><li><a href="/learn/how-much-fits/">06 How much fits on the chip</a></li></ol></li><li><h3>Module 3 · Your t27 program</h3><p>Write a spec, test it, and see why compiling is not the same as being right.</p><ol><li class="on" aria-current="page"><a href="/learn/hello-t27/">07 Hello, t27</a></li><li><a href="/learn/tests-are-the-spec/">08 Tests are part of the spec</a></li><li><a href="/learn/compiles-is-not-correct/">09 Compiles is not the same as correct</a></li></ol></li><li><h3>Module 4 · Inside t27c</h3><p>How the compiler reads a spec and what it writes, including the native t27b.</p><ol><li><a href="/learn/compiler-stages/">10 The stages of t27c</a></li><li><a href="/learn/t27c-on-a-real-machine/">11 t27c on a real machine</a></li><li><a href="/learn/seven-backends-and-t27b/">12 Seven backends, and t27b</a></li></ol></li><li><h3>Module 5 · From spec to hardware</h3><p>The Verilog t27c writes, the cells it becomes, and what one LUT does.</p><ol><li><a href="/learn/spec-to-verilog/">13 From a spec to Verilog</a></li><li><a href="/learn/count-the-cells/">14 Count the cells</a></li><li><a href="/learn/one-lut-up-close/">15 One LUT, up close</a></li></ol></li><li><h3>Module 6 · Reading synthesis</h3><p>What yosys reports about your design, and which warnings matter.</p><ol><li><a href="/learn/synthesis-warnings/">16 Synthesis warnings that matter</a></li><li><a href="/learn/state-machines/">17 State machines</a></li><li><a href="/learn/where-the-luts-go/">18 Where the LUTs go</a></li></ol></li><li><h3>Module 7 · Place, route, timing</h3><p>Where the cells land on the die, and whether the clock is met.</p><ol><li><a href="/learn/placement/">19 Placement: where the cells land</a></li><li><a href="/learn/logic-depth/">20 Logic depth</a></li><li><a href="/learn/timing-slack/">21 Timing and slack</a></li></ol></li><li><h3>Module 8 · The bitstream</h3><p>How a routed design becomes the bits the chip loads, with no Vivado.</p><ol><li><a href="/learn/fasm-to-frames/">22 From FASM to frames</a></li><li><a href="/learn/a-build-receipt/">23 A build, itemised</a></li><li><a href="/learn/read-a-bitstream/">24 Read a bitstream</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>Ask the chip who it is, load the bits, and check them.</p><ol><li><a href="/learn/jtag-asks-the-chip/">25 JTAG asks the chip</a></li><li><a href="/learn/same-bits-on-the-board/">26 The same bits, on the board</a></li><li><a href="/learn/capstone/">27 Capstone: out-guess yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/how-much-fits/index.html b/apps/website/public/learn/how-much-fits/index.html
index 97038eaf90..043c22d03f 100644
--- a/apps/website/public/learn/how-much-fits/index.html
+++ b/apps/website/public/learn/how-much-fits/index.html
@@ -441,5 +441,6 @@ <h1>How much fits on the chip</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The chip</h3><p>What an FPGA is, which chip we use, and how its pins meet the board.</p><ol><li><a href="/learn/fpga-is-a-grid/">01 An FPGA is a grid you configure</a></li><li><a href="/learn/which-chip/">02 Which chip is this?</a></li><li><a href="/learn/pins-and-package/">03 Pins, balls and the package</a></li></ol></li><li><h3>Module 2 · Numbers in hardware</h3><p>Bits, trits and number formats, and what arithmetic costs in logic.</p><ol><li><a href="/learn/bits-and-trits/">04 Bits, trits and GF16</a></li><li><a href="/learn/what-arithmetic-costs/">05 What arithmetic costs</a></li><li class="on" aria-current="page"><a href="/learn/how-much-fits/">06 How much fits on the chip</a></li></ol></li><li><h3>Module 3 · Your t27 program</h3><p>Write a spec, test it, and see why compiling is not the same as being right.</p><ol><li><a href="/learn/hello-t27/">07 Hello, t27</a></li><li><a href="/learn/tests-are-the-spec/">08 Tests are part of the spec</a></li><li><a href="/learn/compiles-is-not-correct/">09 Compiles is not the same as correct</a></li></ol></li><li><h3>Module 4 · Inside t27c</h3><p>How the compiler reads a spec and what it writes, including the native t27b.</p><ol><li><a href="/learn/compiler-stages/">10 The stages of t27c</a></li><li><a href="/learn/t27c-on-a-real-machine/">11 t27c on a real machine</a></li><li><a href="/learn/seven-backends-and-t27b/">12 Seven backends, and t27b</a></li></ol></li><li><h3>Module 5 · From spec to hardware</h3><p>The Verilog t27c writes, the cells it becomes, and what one LUT does.</p><ol><li><a href="/learn/spec-to-verilog/">13 From a spec to Verilog</a></li><li><a href="/learn/count-the-cells/">14 Count the cells</a></li><li><a href="/learn/one-lut-up-close/">15 One LUT, up close</a></li></ol></li><li><h3>Module 6 · Reading synthesis</h3><p>What yosys reports about your design, and which warnings matter.</p><ol><li><a href="/learn/synthesis-warnings/">16 Synthesis warnings that matter</a></li><li><a href="/learn/state-machines/">17 State machines</a></li><li><a href="/learn/where-the-luts-go/">18 Where the LUTs go</a></li></ol></li><li><h3>Module 7 · Place, route, timing</h3><p>Where the cells land on the die, and whether the clock is met.</p><ol><li><a href="/learn/placement/">19 Placement: where the cells land</a></li><li><a href="/learn/logic-depth/">20 Logic depth</a></li><li><a href="/learn/timing-slack/">21 Timing and slack</a></li></ol></li><li><h3>Module 8 · The bitstream</h3><p>How a routed design becomes the bits the chip loads, with no Vivado.</p><ol><li><a href="/learn/fasm-to-frames/">22 From FASM to frames</a></li><li><a href="/learn/a-build-receipt/">23 A build, itemised</a></li><li><a href="/learn/read-a-bitstream/">24 Read a bitstream</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>Ask the chip who it is, load the bits, and check them.</p><ol><li><a href="/learn/jtag-asks-the-chip/">25 JTAG asks the chip</a></li><li><a href="/learn/same-bits-on-the-board/">26 The same bits, on the board</a></li><li><a href="/learn/capstone/">27 Capstone: out-guess yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/if-and-else/index.html b/apps/website/public/learn/if-and-else/index.html
index 2af7b5e0eb..4f329514b6 100644
--- a/apps/website/public/learn/if-and-else/index.html
+++ b/apps/website/public/learn/if-and-else/index.html
@@ -100,5 +100,6 @@ <h1>Choices: if, else and switch</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · A spec file</h3><p>What a .t27 file holds: a module line, prose lines and comments.</p><ol><li><a href="/learn/what-a-spec-is/">01 What a spec is</a></li><li><a href="/learn/the-module-line/">02 The module line</a></li><li><a href="/learn/comments-and-prose/">03 Comments and prose lines</a></li></ol></li><li><h3>Module 2 · Values and types</h3><p>Constants, the width of a whole number, and true, false and text.</p><ol><li><a href="/learn/constants/">04 Constants</a></li><li><a href="/learn/whole-numbers/">05 Whole numbers and their widths</a></li><li><a href="/learn/true-false-and-text/">06 True, false and text</a></li></ol></li><li><h3>Module 3 · Arrays, trits and expressions</h3><p>Lists of values, the three values of a trit, and what operators compute.</p><ol><li><a href="/learn/arrays/">07 Arrays</a></li><li><a href="/learn/trits/">08 Trits: minus one, zero, one</a></li><li><a href="/learn/expressions/">09 Expressions</a></li></ol></li><li><h3>Module 4 · Tests and functions</h3><p>A test block, several small tests in one spec, and a function with typed inputs.</p><ol><li><a href="/learn/a-test-block/">10 A test block</a></li><li><a href="/learn/many-tests/">11 Many tests in one spec</a></li><li><a href="/learn/functions/">12 Functions</a></li></ol></li><li><h3>Module 5 · Inside a function</h3><p>Names with let and var, choices with if and switch, and loops.</p><ol><li><a href="/learn/local-names/">13 Local names: let and var</a></li><li class="on" aria-current="page"><a href="/learn/if-and-else/">14 Choices: if, else and switch</a></li><li><a href="/learn/loops/">15 Loops: while and for</a></li></ol></li><li><h3>Module 6 · Visibility and your own types</h3><p>What pub shares with other modules, and the structs and enums you declare.</p><ol><li><a href="/learn/pub-or-private/">16 pub or private</a></li><li><a href="/learn/structs/">17 Structs</a></li><li><a href="/learn/enums/">18 Enums</a></li></ol></li><li><h3>Module 7 · Modules, rules and gen-ts</h3><p>Reading another module, stating a rule that always holds, and a spec turned into TypeScript.</p><ol><li><a href="/learn/use-other-modules/">19 Using other modules</a></li><li><a href="/learn/invariants/">20 Invariants and benches</a></li><li><a href="/learn/gen-ts/">21 gen-ts: a spec becomes TypeScript</a></li></ol></li><li><h3>Module 8 · The compiler and tri</h3><p>The backends t27c writes, the stages it reads a spec in, and the tri command.</p><ol><li><a href="/learn/seven-backends/">22 Seven backends</a></li><li><a href="/learn/stages-of-t27c/">23 The stages of t27c</a></li><li><a href="/learn/tri-commands/">24 tri: the command line</a></li></ol></li><li><h3>Module 9 · Errors, a program, what next</h3><p>Reading a compiler error, one small complete program, and the course to take next.</p><ol><li><a href="/learn/reading-errors/">25 Reading compiler errors</a></li><li><a href="/learn/a-small-program/">26 A small complete program</a></li><li><a href="/learn/where-next/">27 Where to go next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/index.html b/apps/website/public/learn/index.html
index 6c82050c77..8abd0a8a08 100644
--- a/apps/website/public/learn/index.html
+++ b/apps/website/public/learn/index.html
@@ -67,5 +67,6 @@ <h1>t27 basics: the language in 27 lessons</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · A spec file</h3><p>What a .t27 file holds: a module line, prose lines and comments.</p><ol><li><a href="/learn/what-a-spec-is/">01 What a spec is</a></li><li><a href="/learn/the-module-line/">02 The module line</a></li><li><a href="/learn/comments-and-prose/">03 Comments and prose lines</a></li></ol></li><li><h3>Module 2 · Values and types</h3><p>Constants, the width of a whole number, and true, false and text.</p><ol><li><a href="/learn/constants/">04 Constants</a></li><li><a href="/learn/whole-numbers/">05 Whole numbers and their widths</a></li><li><a href="/learn/true-false-and-text/">06 True, false and text</a></li></ol></li><li><h3>Module 3 · Arrays, trits and expressions</h3><p>Lists of values, the three values of a trit, and what operators compute.</p><ol><li><a href="/learn/arrays/">07 Arrays</a></li><li><a href="/learn/trits/">08 Trits: minus one, zero, one</a></li><li><a href="/learn/expressions/">09 Expressions</a></li></ol></li><li><h3>Module 4 · Tests and functions</h3><p>A test block, several small tests in one spec, and a function with typed inputs.</p><ol><li><a href="/learn/a-test-block/">10 A test block</a></li><li><a href="/learn/many-tests/">11 Many tests in one spec</a></li><li><a href="/learn/functions/">12 Functions</a></li></ol></li><li><h3>Module 5 · Inside a function</h3><p>Names with let and var, choices with if and switch, and loops.</p><ol><li><a href="/learn/local-names/">13 Local names: let and var</a></li><li><a href="/learn/if-and-else/">14 Choices: if, else and switch</a></li><li><a href="/learn/loops/">15 Loops: while and for</a></li></ol></li><li><h3>Module 6 · Visibility and your own types</h3><p>What pub shares with other modules, and the structs and enums you declare.</p><ol><li><a href="/learn/pub-or-private/">16 pub or private</a></li><li><a href="/learn/structs/">17 Structs</a></li><li><a href="/learn/enums/">18 Enums</a></li></ol></li><li><h3>Module 7 · Modules, rules and gen-ts</h3><p>Reading another module, stating a rule that always holds, and a spec turned into TypeScript.</p><ol><li><a href="/learn/use-other-modules/">19 Using other modules</a></li><li><a href="/learn/invariants/">20 Invariants and benches</a></li><li><a href="/learn/gen-ts/">21 gen-ts: a spec becomes TypeScript</a></li></ol></li><li><h3>Module 8 · The compiler and tri</h3><p>The backends t27c writes, the stages it reads a spec in, and the tri command.</p><ol><li><a href="/learn/seven-backends/">22 Seven backends</a></li><li><a href="/learn/stages-of-t27c/">23 The stages of t27c</a></li><li><a href="/learn/tri-commands/">24 tri: the command line</a></li></ol></li><li><h3>Module 9 · Errors, a program, what next</h3><p>Reading a compiler error, one small complete program, and the course to take next.</p><ol><li><a href="/learn/reading-errors/">25 Reading compiler errors</a></li><li><a href="/learn/a-small-program/">26 A small complete program</a></li><li><a href="/learn/where-next/">27 Where to go next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/induction/index.html b/apps/website/public/learn/induction/index.html
index 4624bd9123..d391c554ab 100644
--- a/apps/website/public/learn/induction/index.html
+++ b/apps/website/public/learn/induction/index.html
@@ -207,5 +207,6 @@ <h1>Induction</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Why verify</h3><p>Designs that compile and are wrong, the model that decides, and the plan written before the code.</p><ol><li><a href="/learn/bugs-that-compile/">01 Bugs that compile</a></li><li><a href="/learn/golden-model/">02 The golden model</a></li><li><a href="/learn/plan-before-code/">03 Plan before code</a></li></ol></li><li><h3>Module 2 · Testbenches</h3><p>Stimulus, checks and a verdict, written as one spec beside the design it judges.</p><ol><li><a href="/learn/anatomy-of-a-tb/">04 Anatomy of a testbench</a></li><li><a href="/learn/self-checking/">05 Self-checking, not staring</a></li><li><a href="/learn/directed-vs-random/">06 Directed versus random</a></li></ol></li><li><h3>Module 3 · Waveforms</h3><p>A trace of every signal, read the way a hardware engineer reads it, and two runs compared.</p><ol><li><a href="/learn/vcd-format/">07 The VCD format</a></li><li><a href="/learn/debug-with-waves/">08 Debug with waves</a></li><li><a href="/learn/compare-traces/">09 Compare two traces</a></li></ol></li><li><h3>Module 4 · Conformance vectors</h3><p>Cases with the answer written beside them, kept where the compiler can reach them.</p><ol><li><a href="/learn/what-a-vector-is/">10 What a vector is</a></li><li><a href="/learn/golden-ruler-vectors/">11 Golden ruler vectors</a></li><li><a href="/learn/end-to-end-scenarios/">12 End-to-end scenarios</a></li></ol></li><li><h3>Module 5 · Cosimulation</h3><p>Spec, simulator and board agreeing on the bench Artix-7 XC7A200T, and what to do when they do not.</p><ol><li><a href="/learn/three-way-agreement/">13 Three-way agreement</a></li><li><a href="/learn/simulator-internals/">14 Inside the simulator</a></li><li><a href="/learn/when-they-disagree/">15 When they disagree</a></li></ol></li><li><h3>Module 6 · Coverage</h3><p>What the tests touched: lines, toggles, states, and what that number hides.</p><ol><li><a href="/learn/line-and-toggle/">16 Line and toggle</a></li><li><a href="/learn/fsm-coverage/">17 FSM coverage</a></li><li><a href="/learn/coverage-lies/">18 Coverage lies</a></li></ol></li><li><h3>Module 7 · Formal</h3><p>Assertions that hold every cycle, bounded search for a counterexample, and why a proof needs induction.</p><ol><li><a href="/learn/assertions/">19 Assertions</a></li><li><a href="/learn/bounded-model-checking/">20 Bounded model checking</a></li><li class="on" aria-current="page"><a href="/learn/induction/">21 Induction</a></li></ol></li><li><h3>Module 8 · Mutation</h3><p>Break the design on purpose and count what the tests catch.</p><ol><li><a href="/learn/kill-the-mutant/">22 Kill the mutant</a></li><li><a href="/learn/surviving-mutants/">23 Surviving mutants</a></li><li><a href="/learn/test-strength/">24 Test strength</a></li></ol></li><li><h3>Module 9 · Sign-off</h3><p>One command, every receipt, a clean verdict you can show.</p><ol><li><a href="/learn/build-verify/">25 Build and verify</a></li><li><a href="/learn/board-selftest/">26 The board self-tests</a></li><li><a href="/learn/verify-capstone/">27 Capstone: verify a module whole</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/invariants/index.html b/apps/website/public/learn/invariants/index.html
index 849a4c907d..80aa1f2f88 100644
--- a/apps/website/public/learn/invariants/index.html
+++ b/apps/website/public/learn/invariants/index.html
@@ -84,5 +84,6 @@ <h1>Invariants and benches</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · A spec file</h3><p>What a .t27 file holds: a module line, prose lines and comments.</p><ol><li><a href="/learn/what-a-spec-is/">01 What a spec is</a></li><li><a href="/learn/the-module-line/">02 The module line</a></li><li><a href="/learn/comments-and-prose/">03 Comments and prose lines</a></li></ol></li><li><h3>Module 2 · Values and types</h3><p>Constants, the width of a whole number, and true, false and text.</p><ol><li><a href="/learn/constants/">04 Constants</a></li><li><a href="/learn/whole-numbers/">05 Whole numbers and their widths</a></li><li><a href="/learn/true-false-and-text/">06 True, false and text</a></li></ol></li><li><h3>Module 3 · Arrays, trits and expressions</h3><p>Lists of values, the three values of a trit, and what operators compute.</p><ol><li><a href="/learn/arrays/">07 Arrays</a></li><li><a href="/learn/trits/">08 Trits: minus one, zero, one</a></li><li><a href="/learn/expressions/">09 Expressions</a></li></ol></li><li><h3>Module 4 · Tests and functions</h3><p>A test block, several small tests in one spec, and a function with typed inputs.</p><ol><li><a href="/learn/a-test-block/">10 A test block</a></li><li><a href="/learn/many-tests/">11 Many tests in one spec</a></li><li><a href="/learn/functions/">12 Functions</a></li></ol></li><li><h3>Module 5 · Inside a function</h3><p>Names with let and var, choices with if and switch, and loops.</p><ol><li><a href="/learn/local-names/">13 Local names: let and var</a></li><li><a href="/learn/if-and-else/">14 Choices: if, else and switch</a></li><li><a href="/learn/loops/">15 Loops: while and for</a></li></ol></li><li><h3>Module 6 · Visibility and your own types</h3><p>What pub shares with other modules, and the structs and enums you declare.</p><ol><li><a href="/learn/pub-or-private/">16 pub or private</a></li><li><a href="/learn/structs/">17 Structs</a></li><li><a href="/learn/enums/">18 Enums</a></li></ol></li><li><h3>Module 7 · Modules, rules and gen-ts</h3><p>Reading another module, stating a rule that always holds, and a spec turned into TypeScript.</p><ol><li><a href="/learn/use-other-modules/">19 Using other modules</a></li><li class="on" aria-current="page"><a href="/learn/invariants/">20 Invariants and benches</a></li><li><a href="/learn/gen-ts/">21 gen-ts: a spec becomes TypeScript</a></li></ol></li><li><h3>Module 8 · The compiler and tri</h3><p>The backends t27c writes, the stages it reads a spec in, and the tri command.</p><ol><li><a href="/learn/seven-backends/">22 Seven backends</a></li><li><a href="/learn/stages-of-t27c/">23 The stages of t27c</a></li><li><a href="/learn/tri-commands/">24 tri: the command line</a></li></ol></li><li><h3>Module 9 · Errors, a program, what next</h3><p>Reading a compiler error, one small complete program, and the course to take next.</p><ol><li><a href="/learn/reading-errors/">25 Reading compiler errors</a></li><li><a href="/learn/a-small-program/">26 A small complete program</a></li><li><a href="/learn/where-next/">27 Where to go next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/io-timing/index.html b/apps/website/public/learn/io-timing/index.html
index 539689c07b..32b76c96bb 100644
--- a/apps/website/public/learn/io-timing/index.html
+++ b/apps/website/public/learn/io-timing/index.html
@@ -364,5 +364,6 @@ <h1>IO timing</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a clock is</h3><p>One edge, one world: what shares a clock edge shares a world, the period and the jitter of a real edge, and where the clock enters a board.</p><ol><li><a href="/learn/one-edge-one-world/">01 One edge, one world</a></li><li><a href="/learn/period-and-jitter/">02 A period, an edge, and jitter</a></li><li><a href="/learn/where-the-clock-enters/">03 Where the clock enters</a></li></ol></li><li><h3>Module 2 · Clock trees</h3><p>Skew and insertion delay, the global buffer network, and the trap of gating a clock with logic.</p><ol><li><a href="/learn/skew-and-insertion/">04 Skew and insertion delay</a></li><li><a href="/learn/global-buffers/">05 The global buffer</a></li><li><a href="/learn/gated-clock-trap/">06 The gated-clock trap</a></li></ol></li><li><h3>Module 3 · PLL and MMCM</h3><p>Multiply and divide one clock into another, move its phase in steps of the VCO, and which clocks the analyzer treats as related.</p><ol><li><a href="/learn/multiply-and-divide/">07 Multiply and divide</a></li><li><a href="/learn/phase-and-lock/">08 Phase and lock</a></li><li><a href="/learn/related-or-not/">09 Related or not</a></li></ol></li><li><h3>Module 4 · Resets</h3><p>Assert asynchronously, release synchronously: the three reset kinds, the release pipe, and the tree a reset grows.</p><ol><li><a href="/learn/sync-or-async/">10 Sync or async</a></li><li><a href="/learn/reset-release/">11 The release pipe</a></li><li><a href="/learn/reset-fanout/">12 Reset fanout</a></li></ol></li><li><h3>Module 5 · Metastability</h3><p>The setup-hold window, the mean time between failures in integer arithmetic, and the two flops that fix it.</p><ol><li><a href="/learn/setup-hold-window/">13 The setup-hold window</a></li><li><a href="/learn/mtbf/">14 Mean time between failures</a></li><li><a href="/learn/two-flop-sync/">15 The two-flop synchronizer</a></li></ol></li><li><h3>Module 6 · Crossing many bits</h3><p>Why a binary bus tears, why Gray code does not, and the handshake that moves a pulse between worlds.</p><ol><li><a href="/learn/why-binary-fails/">16 Why binary fails</a></li><li><a href="/learn/gray-code/">17 Gray code</a></li><li><a href="/learn/handshakes-and-pulses/">18 Handshakes and pulses</a></li></ol></li><li><h3>Module 7 · The asynchronous FIFO</h3><p>Pointers, flags and depth: the buffer that moves a stream between two clocks.</p><ol><li><a href="/learn/pointers-and-flags/">19 Pointers and flags</a></li><li><a href="/learn/depth-sizing/">20 Sizing the depth</a></li><li><a href="/learn/fifo-testbench/">21 The FIFO testbench</a></li></ol></li><li><h3>Module 8 · Constraints</h3><p>The lines that tell the analyzer what a clock is, which paths not to check, and what the pins must meet.</p><ol><li><a href="/learn/create-clock/">22 create_clock</a></li><li><a href="/learn/false-path-and-max-delay/">23 False paths and max delay</a></li><li class="on" aria-current="page"><a href="/learn/io-timing/">24 IO timing</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>A CDC report, one crossing captured at the flip-flops, and the bitstream diff that closes the course.</p><ol><li><a href="/learn/cdc-report/">25 The CDC report</a></li><li><a href="/learn/capture-a-crossing/">26 Capture a crossing</a></li><li><a href="/learn/capstone-diff/">27 Capstone: change it and diff it</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/jtag-asks-the-chip/index.html b/apps/website/public/learn/jtag-asks-the-chip/index.html
index 19c634407c..2f2dfa1eda 100644
--- a/apps/website/public/learn/jtag-asks-the-chip/index.html
+++ b/apps/website/public/learn/jtag-asks-the-chip/index.html
@@ -110,5 +110,6 @@ <h1>JTAG asks the chip</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The chip</h3><p>What an FPGA is, which chip we use, and how its pins meet the board.</p><ol><li><a href="/learn/fpga-is-a-grid/">01 An FPGA is a grid you configure</a></li><li><a href="/learn/which-chip/">02 Which chip is this?</a></li><li><a href="/learn/pins-and-package/">03 Pins, balls and the package</a></li></ol></li><li><h3>Module 2 · Numbers in hardware</h3><p>Bits, trits and number formats, and what arithmetic costs in logic.</p><ol><li><a href="/learn/bits-and-trits/">04 Bits, trits and GF16</a></li><li><a href="/learn/what-arithmetic-costs/">05 What arithmetic costs</a></li><li><a href="/learn/how-much-fits/">06 How much fits on the chip</a></li></ol></li><li><h3>Module 3 · Your t27 program</h3><p>Write a spec, test it, and see why compiling is not the same as being right.</p><ol><li><a href="/learn/hello-t27/">07 Hello, t27</a></li><li><a href="/learn/tests-are-the-spec/">08 Tests are part of the spec</a></li><li><a href="/learn/compiles-is-not-correct/">09 Compiles is not the same as correct</a></li></ol></li><li><h3>Module 4 · Inside t27c</h3><p>How the compiler reads a spec and what it writes, including the native t27b.</p><ol><li><a href="/learn/compiler-stages/">10 The stages of t27c</a></li><li><a href="/learn/t27c-on-a-real-machine/">11 t27c on a real machine</a></li><li><a href="/learn/seven-backends-and-t27b/">12 Seven backends, and t27b</a></li></ol></li><li><h3>Module 5 · From spec to hardware</h3><p>The Verilog t27c writes, the cells it becomes, and what one LUT does.</p><ol><li><a href="/learn/spec-to-verilog/">13 From a spec to Verilog</a></li><li><a href="/learn/count-the-cells/">14 Count the cells</a></li><li><a href="/learn/one-lut-up-close/">15 One LUT, up close</a></li></ol></li><li><h3>Module 6 · Reading synthesis</h3><p>What yosys reports about your design, and which warnings matter.</p><ol><li><a href="/learn/synthesis-warnings/">16 Synthesis warnings that matter</a></li><li><a href="/learn/state-machines/">17 State machines</a></li><li><a href="/learn/where-the-luts-go/">18 Where the LUTs go</a></li></ol></li><li><h3>Module 7 · Place, route, timing</h3><p>Where the cells land on the die, and whether the clock is met.</p><ol><li><a href="/learn/placement/">19 Placement: where the cells land</a></li><li><a href="/learn/logic-depth/">20 Logic depth</a></li><li><a href="/learn/timing-slack/">21 Timing and slack</a></li></ol></li><li><h3>Module 8 · The bitstream</h3><p>How a routed design becomes the bits the chip loads, with no Vivado.</p><ol><li><a href="/learn/fasm-to-frames/">22 From FASM to frames</a></li><li><a href="/learn/a-build-receipt/">23 A build, itemised</a></li><li><a href="/learn/read-a-bitstream/">24 Read a bitstream</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>Ask the chip who it is, load the bits, and check them.</p><ol><li class="on" aria-current="page"><a href="/learn/jtag-asks-the-chip/">25 JTAG asks the chip</a></li><li><a href="/learn/same-bits-on-the-board/">26 The same bits, on the board</a></li><li><a href="/learn/capstone/">27 Capstone: out-guess yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/kill-the-mutant/index.html b/apps/website/public/learn/kill-the-mutant/index.html
index 44e054896e..54c9734a25 100644
--- a/apps/website/public/learn/kill-the-mutant/index.html
+++ b/apps/website/public/learn/kill-the-mutant/index.html
@@ -110,5 +110,6 @@ <h1>Kill the mutant</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Why verify</h3><p>Designs that compile and are wrong, the model that decides, and the plan written before the code.</p><ol><li><a href="/learn/bugs-that-compile/">01 Bugs that compile</a></li><li><a href="/learn/golden-model/">02 The golden model</a></li><li><a href="/learn/plan-before-code/">03 Plan before code</a></li></ol></li><li><h3>Module 2 · Testbenches</h3><p>Stimulus, checks and a verdict, written as one spec beside the design it judges.</p><ol><li><a href="/learn/anatomy-of-a-tb/">04 Anatomy of a testbench</a></li><li><a href="/learn/self-checking/">05 Self-checking, not staring</a></li><li><a href="/learn/directed-vs-random/">06 Directed versus random</a></li></ol></li><li><h3>Module 3 · Waveforms</h3><p>A trace of every signal, read the way a hardware engineer reads it, and two runs compared.</p><ol><li><a href="/learn/vcd-format/">07 The VCD format</a></li><li><a href="/learn/debug-with-waves/">08 Debug with waves</a></li><li><a href="/learn/compare-traces/">09 Compare two traces</a></li></ol></li><li><h3>Module 4 · Conformance vectors</h3><p>Cases with the answer written beside them, kept where the compiler can reach them.</p><ol><li><a href="/learn/what-a-vector-is/">10 What a vector is</a></li><li><a href="/learn/golden-ruler-vectors/">11 Golden ruler vectors</a></li><li><a href="/learn/end-to-end-scenarios/">12 End-to-end scenarios</a></li></ol></li><li><h3>Module 5 · Cosimulation</h3><p>Spec, simulator and board agreeing on the bench Artix-7 XC7A200T, and what to do when they do not.</p><ol><li><a href="/learn/three-way-agreement/">13 Three-way agreement</a></li><li><a href="/learn/simulator-internals/">14 Inside the simulator</a></li><li><a href="/learn/when-they-disagree/">15 When they disagree</a></li></ol></li><li><h3>Module 6 · Coverage</h3><p>What the tests touched: lines, toggles, states, and what that number hides.</p><ol><li><a href="/learn/line-and-toggle/">16 Line and toggle</a></li><li><a href="/learn/fsm-coverage/">17 FSM coverage</a></li><li><a href="/learn/coverage-lies/">18 Coverage lies</a></li></ol></li><li><h3>Module 7 · Formal</h3><p>Assertions that hold every cycle, bounded search for a counterexample, and why a proof needs induction.</p><ol><li><a href="/learn/assertions/">19 Assertions</a></li><li><a href="/learn/bounded-model-checking/">20 Bounded model checking</a></li><li><a href="/learn/induction/">21 Induction</a></li></ol></li><li><h3>Module 8 · Mutation</h3><p>Break the design on purpose and count what the tests catch.</p><ol><li class="on" aria-current="page"><a href="/learn/kill-the-mutant/">22 Kill the mutant</a></li><li><a href="/learn/surviving-mutants/">23 Surviving mutants</a></li><li><a href="/learn/test-strength/">24 Test strength</a></li></ol></li><li><h3>Module 9 · Sign-off</h3><p>One command, every receipt, a clean verdict you can show.</p><ol><li><a href="/learn/build-verify/">25 Build and verify</a></li><li><a href="/learn/board-selftest/">26 The board self-tests</a></li><li><a href="/learn/verify-capstone/">27 Capstone: verify a module whole</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/kleene-logic/index.html b/apps/website/public/learn/kleene-logic/index.html
index 6aeccaf661..d5701fbdfd 100644
--- a/apps/website/public/learn/kleene-logic/index.html
+++ b/apps/website/public/learn/kleene-logic/index.html
@@ -769,5 +769,6 @@ <h1>Three-valued logic</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Three values</h3><p>Why three, how balanced ternary writes every number without a sign, and what flipping and cutting trits do.</p><ol><li><a href="/learn/why-three/">01 Why three</a></li><li><a href="/learn/balanced-ternary/">02 Balanced ternary</a></li><li><a href="/learn/flip-and-cut/">03 Flip to negate, cut to round</a></li></ol></li><li><h3>Module 2 · Logic with unknown</h3><p>Kleene's three-valued gates, two gates binary has no twin for, and addition as a pair of tables.</p><ol><li class="on" aria-current="page"><a href="/learn/kleene-logic/">04 Three-valued logic</a></li><li><a href="/learn/consensus-and-any/">05 Consensus and any</a></li><li><a href="/learn/sum-and-carry/">06 Sum and carry</a></li></ol></li><li><h3>Module 3 · Adding trits</h3><p>The half adder, the full adder and a carry that ripples left, one place at a time.</p><ol><li><a href="/learn/half-adder/">07 The half adder</a></li><li><a href="/learn/full-adder/">08 The full adder</a></li><li><a href="/learn/ripple-carry/">09 Ripple carry</a></li></ol></li><li><h3>Module 4 · Multiplying without a multiplier</h3><p>Copy, drop or flip: a product by one trit, long multiplication, and a MAC that only adds.</p><ol><li><a href="/learn/times-a-trit/">10 Times one trit</a></li><li><a href="/learn/long-multiplication/">11 Long multiplication</a></li><li><a href="/learn/ternary-mac/">12 A ternary MAC</a></li></ol></li><li><h3>Module 5 · Trits in binary memory</h3><p>Two bits per trit, five trits per byte, and the bits a 27-trit word needs.</p><ol><li><a href="/learn/two-bits-per-trit/">13 Two bits per trit</a></li><li><a href="/learn/five-trits-a-byte/">14 Five trits in a byte</a></li><li><a href="/learn/a-word-of-27/">15 A word of 27 trits</a></li></ol></li><li><h3>Module 6 · The TRI-27 instruction word</h3><p>The 32-bit word the Trinity emulator decodes: its fields, its 47 opcodes and its 15-bit immediate.</p><ol><li><a href="/learn/the-instruction-word/">16 The instruction word</a></li><li><a href="/learn/the-opcode-map/">17 The opcode map</a></li><li><a href="/learn/the-immediate/">18 The immediate</a></li></ol></li><li><h3>Module 7 · A ternary machine</h3><p>An ALU built from this course's adders, a three-way jump, and a program you can step.</p><ol><li><a href="/learn/a-ternary-alu/">19 A ternary ALU</a></li><li><a href="/learn/three-way-jump/">20 A three-way jump</a></li><li><a href="/learn/run-a-program/">21 Run a program</a></li></ol></li><li><h3>Module 8 · Three answers</h3><p>Compare at the highest differing trit, find a number in thirds, sort with three-way compares.</p><ol><li><a href="/learn/compare-at-the-top/">22 Compare at the top</a></li><li><a href="/learn/questions-in-thirds/">23 Questions in thirds</a></li><li><a href="/learn/sort-by-three/">24 Sort with three-way compares</a></li></ol></li><li><h3>Module 9 · Ternary neurons</h3><p>Weights of -1, 0, +1: one neuron, a detector and a small layer, with no multiplier anywhere.</p><ol><li><a href="/learn/one-ternary-neuron/">25 One ternary neuron</a></li><li><a href="/learn/a-bar-detector/">26 A bar detector</a></li><li><a href="/learn/a-tiny-layer/">27 A tiny layer</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/latency/index.html b/apps/website/public/learn/latency/index.html
index 7358001997..80ff120158 100644
--- a/apps/website/public/learn/latency/index.html
+++ b/apps/website/public/learn/latency/index.html
@@ -238,5 +238,6 @@ <h1>Latency</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a bus is</h3><p>Why a bus exists at all: a conversation on wires, framed and addressed, and who is allowed to talk.</p><ol><li><a href="/learn/wires-that-talk/">01 Wires that talk</a></li><li><a href="/learn/framing-a-conversation/">02 Framing a conversation</a></li><li><a href="/learn/master-and-servant/">03 Master and servant</a></li></ol></li><li><h3>Module 2 · UART</h3><p>The two-wire bus with no clock: the frame, the divisor that sets the rate, and the status a driver polls.</p><ol><li><a href="/learn/the-frame/">04 The frame: start, data, stop</a></li><li><a href="/learn/baud-and-the-divisor/">05 Baud and the divisor</a></li><li><a href="/learn/status-and-fifos/">06 Status and FIFOs</a></li></ol></li><li><h3>Module 3 · SPI</h3><p>The clocked conversation: four modes, a prescaler ladder, and a chip select per servant.</p><ol><li><a href="/learn/modes-cpol-cpha/">07 Modes: CPOL and CPHA</a></li><li><a href="/learn/the-prescaler-ladder/">08 The prescaler ladder</a></li><li><a href="/learn/chip-select-and-widths/">09 Chip select and widths</a></li></ol></li><li><h3>Module 4 · APB</h3><p>The register bus: PSEL and PENABLE, strobes and wait states, and how many address bits a peripheral count costs.</p><ol><li><a href="/learn/psel-and-penable/">10 PSEL and PENABLE</a></li><li><a href="/learn/strobes-and-waits/">11 Strobes and waits</a></li><li><a href="/learn/address-decode/">12 Address decode</a></li></ol></li><li><h3>Module 5 · AXI4</h3><p>The five channels: address, data and response in both directions, lite or full, bursts and IDs.</p><ol><li><a href="/learn/five-channels/">13 Five channels</a></li><li><a href="/learn/lite-or-full/">14 Lite or full</a></li><li><a href="/learn/bursts-and-ids/">15 Bursts and IDs</a></li></ol></li><li><h3>Module 6 · Memory</h3><p>What sits on the far side of every bus: memory maps, port kinds, and latency that a wait state must cover.</p><ol><li><a href="/learn/memory-maps/">16 Memory maps</a></li><li><a href="/learn/ports-and-kinds/">17 Ports and kinds</a></li><li class="on" aria-current="page"><a href="/learn/latency/">18 Latency</a></li></ol></li><li><h3>Module 7 · Bridges</h3><p>Why designs grow more than one bus, and the packet bridge that moves work between them.</p><ol><li><a href="/learn/why-bridges/">19 Why bridges</a></li><li><a href="/learn/the-packet-bridge/">20 The packet bridge</a></li><li><a href="/learn/everything-in-one-bench/">21 Everything in one bench</a></li></ol></li><li><h3>Module 8 · Ethernet</h3><p>Frames, the frame check sequence, RGMII timing, and the pre-registered steps of a real bring-up.</p><ol><li><a href="/learn/frames-and-the-fcs/">22 Frames and the FCS</a></li><li><a href="/learn/rgmii-timing/">23 RGMII timing</a></li><li><a href="/learn/bring-up-on-the-bench/">24 Bring-up on the bench</a></li></ol></li><li><h3>Module 9 · The bench</h3><p>The discipline that guards real hardware: who holds the IO, taking and giving the lock, and what runs next.</p><ol><li><a href="/learn/who-holds-the-io/">25 Who holds the IO</a></li><li><a href="/learn/claim-and-release/">26 Claim and release</a></li><li><a href="/learn/capstone-what-runs-next/">27 Capstone: what runs next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/line-and-toggle/index.html b/apps/website/public/learn/line-and-toggle/index.html
index 59c4fd41d0..ea3b88ad2e 100644
--- a/apps/website/public/learn/line-and-toggle/index.html
+++ b/apps/website/public/learn/line-and-toggle/index.html
@@ -576,5 +576,6 @@ <h1>Line and toggle</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Why verify</h3><p>Designs that compile and are wrong, the model that decides, and the plan written before the code.</p><ol><li><a href="/learn/bugs-that-compile/">01 Bugs that compile</a></li><li><a href="/learn/golden-model/">02 The golden model</a></li><li><a href="/learn/plan-before-code/">03 Plan before code</a></li></ol></li><li><h3>Module 2 · Testbenches</h3><p>Stimulus, checks and a verdict, written as one spec beside the design it judges.</p><ol><li><a href="/learn/anatomy-of-a-tb/">04 Anatomy of a testbench</a></li><li><a href="/learn/self-checking/">05 Self-checking, not staring</a></li><li><a href="/learn/directed-vs-random/">06 Directed versus random</a></li></ol></li><li><h3>Module 3 · Waveforms</h3><p>A trace of every signal, read the way a hardware engineer reads it, and two runs compared.</p><ol><li><a href="/learn/vcd-format/">07 The VCD format</a></li><li><a href="/learn/debug-with-waves/">08 Debug with waves</a></li><li><a href="/learn/compare-traces/">09 Compare two traces</a></li></ol></li><li><h3>Module 4 · Conformance vectors</h3><p>Cases with the answer written beside them, kept where the compiler can reach them.</p><ol><li><a href="/learn/what-a-vector-is/">10 What a vector is</a></li><li><a href="/learn/golden-ruler-vectors/">11 Golden ruler vectors</a></li><li><a href="/learn/end-to-end-scenarios/">12 End-to-end scenarios</a></li></ol></li><li><h3>Module 5 · Cosimulation</h3><p>Spec, simulator and board agreeing on the bench Artix-7 XC7A200T, and what to do when they do not.</p><ol><li><a href="/learn/three-way-agreement/">13 Three-way agreement</a></li><li><a href="/learn/simulator-internals/">14 Inside the simulator</a></li><li><a href="/learn/when-they-disagree/">15 When they disagree</a></li></ol></li><li><h3>Module 6 · Coverage</h3><p>What the tests touched: lines, toggles, states, and what that number hides.</p><ol><li class="on" aria-current="page"><a href="/learn/line-and-toggle/">16 Line and toggle</a></li><li><a href="/learn/fsm-coverage/">17 FSM coverage</a></li><li><a href="/learn/coverage-lies/">18 Coverage lies</a></li></ol></li><li><h3>Module 7 · Formal</h3><p>Assertions that hold every cycle, bounded search for a counterexample, and why a proof needs induction.</p><ol><li><a href="/learn/assertions/">19 Assertions</a></li><li><a href="/learn/bounded-model-checking/">20 Bounded model checking</a></li><li><a href="/learn/induction/">21 Induction</a></li></ol></li><li><h3>Module 8 · Mutation</h3><p>Break the design on purpose and count what the tests catch.</p><ol><li><a href="/learn/kill-the-mutant/">22 Kill the mutant</a></li><li><a href="/learn/surviving-mutants/">23 Surviving mutants</a></li><li><a href="/learn/test-strength/">24 Test strength</a></li></ol></li><li><h3>Module 9 · Sign-off</h3><p>One command, every receipt, a clean verdict you can show.</p><ol><li><a href="/learn/build-verify/">25 Build and verify</a></li><li><a href="/learn/board-selftest/">26 The board self-tests</a></li><li><a href="/learn/verify-capstone/">27 Capstone: verify a module whole</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/lite-or-full/index.html b/apps/website/public/learn/lite-or-full/index.html
index d4dc141d57..022a2f06e1 100644
--- a/apps/website/public/learn/lite-or-full/index.html
+++ b/apps/website/public/learn/lite-or-full/index.html
@@ -463,5 +463,6 @@ <h1>Lite or full</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a bus is</h3><p>Why a bus exists at all: a conversation on wires, framed and addressed, and who is allowed to talk.</p><ol><li><a href="/learn/wires-that-talk/">01 Wires that talk</a></li><li><a href="/learn/framing-a-conversation/">02 Framing a conversation</a></li><li><a href="/learn/master-and-servant/">03 Master and servant</a></li></ol></li><li><h3>Module 2 · UART</h3><p>The two-wire bus with no clock: the frame, the divisor that sets the rate, and the status a driver polls.</p><ol><li><a href="/learn/the-frame/">04 The frame: start, data, stop</a></li><li><a href="/learn/baud-and-the-divisor/">05 Baud and the divisor</a></li><li><a href="/learn/status-and-fifos/">06 Status and FIFOs</a></li></ol></li><li><h3>Module 3 · SPI</h3><p>The clocked conversation: four modes, a prescaler ladder, and a chip select per servant.</p><ol><li><a href="/learn/modes-cpol-cpha/">07 Modes: CPOL and CPHA</a></li><li><a href="/learn/the-prescaler-ladder/">08 The prescaler ladder</a></li><li><a href="/learn/chip-select-and-widths/">09 Chip select and widths</a></li></ol></li><li><h3>Module 4 · APB</h3><p>The register bus: PSEL and PENABLE, strobes and wait states, and how many address bits a peripheral count costs.</p><ol><li><a href="/learn/psel-and-penable/">10 PSEL and PENABLE</a></li><li><a href="/learn/strobes-and-waits/">11 Strobes and waits</a></li><li><a href="/learn/address-decode/">12 Address decode</a></li></ol></li><li><h3>Module 5 · AXI4</h3><p>The five channels: address, data and response in both directions, lite or full, bursts and IDs.</p><ol><li><a href="/learn/five-channels/">13 Five channels</a></li><li class="on" aria-current="page"><a href="/learn/lite-or-full/">14 Lite or full</a></li><li><a href="/learn/bursts-and-ids/">15 Bursts and IDs</a></li></ol></li><li><h3>Module 6 · Memory</h3><p>What sits on the far side of every bus: memory maps, port kinds, and latency that a wait state must cover.</p><ol><li><a href="/learn/memory-maps/">16 Memory maps</a></li><li><a href="/learn/ports-and-kinds/">17 Ports and kinds</a></li><li><a href="/learn/latency/">18 Latency</a></li></ol></li><li><h3>Module 7 · Bridges</h3><p>Why designs grow more than one bus, and the packet bridge that moves work between them.</p><ol><li><a href="/learn/why-bridges/">19 Why bridges</a></li><li><a href="/learn/the-packet-bridge/">20 The packet bridge</a></li><li><a href="/learn/everything-in-one-bench/">21 Everything in one bench</a></li></ol></li><li><h3>Module 8 · Ethernet</h3><p>Frames, the frame check sequence, RGMII timing, and the pre-registered steps of a real bring-up.</p><ol><li><a href="/learn/frames-and-the-fcs/">22 Frames and the FCS</a></li><li><a href="/learn/rgmii-timing/">23 RGMII timing</a></li><li><a href="/learn/bring-up-on-the-bench/">24 Bring-up on the bench</a></li></ol></li><li><h3>Module 9 · The bench</h3><p>The discipline that guards real hardware: who holds the IO, taking and giving the lock, and what runs next.</p><ol><li><a href="/learn/who-holds-the-io/">25 Who holds the IO</a></li><li><a href="/learn/claim-and-release/">26 Claim and release</a></li><li><a href="/learn/capstone-what-runs-next/">27 Capstone: what runs next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/local-names/index.html b/apps/website/public/learn/local-names/index.html
index 1c68b8cc4e..e4ac2b0a26 100644
--- a/apps/website/public/learn/local-names/index.html
+++ b/apps/website/public/learn/local-names/index.html
@@ -94,5 +94,6 @@ <h1>Local names: let and var</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · A spec file</h3><p>What a .t27 file holds: a module line, prose lines and comments.</p><ol><li><a href="/learn/what-a-spec-is/">01 What a spec is</a></li><li><a href="/learn/the-module-line/">02 The module line</a></li><li><a href="/learn/comments-and-prose/">03 Comments and prose lines</a></li></ol></li><li><h3>Module 2 · Values and types</h3><p>Constants, the width of a whole number, and true, false and text.</p><ol><li><a href="/learn/constants/">04 Constants</a></li><li><a href="/learn/whole-numbers/">05 Whole numbers and their widths</a></li><li><a href="/learn/true-false-and-text/">06 True, false and text</a></li></ol></li><li><h3>Module 3 · Arrays, trits and expressions</h3><p>Lists of values, the three values of a trit, and what operators compute.</p><ol><li><a href="/learn/arrays/">07 Arrays</a></li><li><a href="/learn/trits/">08 Trits: minus one, zero, one</a></li><li><a href="/learn/expressions/">09 Expressions</a></li></ol></li><li><h3>Module 4 · Tests and functions</h3><p>A test block, several small tests in one spec, and a function with typed inputs.</p><ol><li><a href="/learn/a-test-block/">10 A test block</a></li><li><a href="/learn/many-tests/">11 Many tests in one spec</a></li><li><a href="/learn/functions/">12 Functions</a></li></ol></li><li><h3>Module 5 · Inside a function</h3><p>Names with let and var, choices with if and switch, and loops.</p><ol><li class="on" aria-current="page"><a href="/learn/local-names/">13 Local names: let and var</a></li><li><a href="/learn/if-and-else/">14 Choices: if, else and switch</a></li><li><a href="/learn/loops/">15 Loops: while and for</a></li></ol></li><li><h3>Module 6 · Visibility and your own types</h3><p>What pub shares with other modules, and the structs and enums you declare.</p><ol><li><a href="/learn/pub-or-private/">16 pub or private</a></li><li><a href="/learn/structs/">17 Structs</a></li><li><a href="/learn/enums/">18 Enums</a></li></ol></li><li><h3>Module 7 · Modules, rules and gen-ts</h3><p>Reading another module, stating a rule that always holds, and a spec turned into TypeScript.</p><ol><li><a href="/learn/use-other-modules/">19 Using other modules</a></li><li><a href="/learn/invariants/">20 Invariants and benches</a></li><li><a href="/learn/gen-ts/">21 gen-ts: a spec becomes TypeScript</a></li></ol></li><li><h3>Module 8 · The compiler and tri</h3><p>The backends t27c writes, the stages it reads a spec in, and the tri command.</p><ol><li><a href="/learn/seven-backends/">22 Seven backends</a></li><li><a href="/learn/stages-of-t27c/">23 The stages of t27c</a></li><li><a href="/learn/tri-commands/">24 tri: the command line</a></li></ol></li><li><h3>Module 9 · Errors, a program, what next</h3><p>Reading a compiler error, one small complete program, and the course to take next.</p><ol><li><a href="/learn/reading-errors/">25 Reading compiler errors</a></li><li><a href="/learn/a-small-program/">26 A small complete program</a></li><li><a href="/learn/where-next/">27 Where to go next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/logic-depth/index.html b/apps/website/public/learn/logic-depth/index.html
index 0f8feffcd2..81bd8f11d2 100644
--- a/apps/website/public/learn/logic-depth/index.html
+++ b/apps/website/public/learn/logic-depth/index.html
@@ -448,5 +448,6 @@ <h1>Logic depth</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The chip</h3><p>What an FPGA is, which chip we use, and how its pins meet the board.</p><ol><li><a href="/learn/fpga-is-a-grid/">01 An FPGA is a grid you configure</a></li><li><a href="/learn/which-chip/">02 Which chip is this?</a></li><li><a href="/learn/pins-and-package/">03 Pins, balls and the package</a></li></ol></li><li><h3>Module 2 · Numbers in hardware</h3><p>Bits, trits and number formats, and what arithmetic costs in logic.</p><ol><li><a href="/learn/bits-and-trits/">04 Bits, trits and GF16</a></li><li><a href="/learn/what-arithmetic-costs/">05 What arithmetic costs</a></li><li><a href="/learn/how-much-fits/">06 How much fits on the chip</a></li></ol></li><li><h3>Module 3 · Your t27 program</h3><p>Write a spec, test it, and see why compiling is not the same as being right.</p><ol><li><a href="/learn/hello-t27/">07 Hello, t27</a></li><li><a href="/learn/tests-are-the-spec/">08 Tests are part of the spec</a></li><li><a href="/learn/compiles-is-not-correct/">09 Compiles is not the same as correct</a></li></ol></li><li><h3>Module 4 · Inside t27c</h3><p>How the compiler reads a spec and what it writes, including the native t27b.</p><ol><li><a href="/learn/compiler-stages/">10 The stages of t27c</a></li><li><a href="/learn/t27c-on-a-real-machine/">11 t27c on a real machine</a></li><li><a href="/learn/seven-backends-and-t27b/">12 Seven backends, and t27b</a></li></ol></li><li><h3>Module 5 · From spec to hardware</h3><p>The Verilog t27c writes, the cells it becomes, and what one LUT does.</p><ol><li><a href="/learn/spec-to-verilog/">13 From a spec to Verilog</a></li><li><a href="/learn/count-the-cells/">14 Count the cells</a></li><li><a href="/learn/one-lut-up-close/">15 One LUT, up close</a></li></ol></li><li><h3>Module 6 · Reading synthesis</h3><p>What yosys reports about your design, and which warnings matter.</p><ol><li><a href="/learn/synthesis-warnings/">16 Synthesis warnings that matter</a></li><li><a href="/learn/state-machines/">17 State machines</a></li><li><a href="/learn/where-the-luts-go/">18 Where the LUTs go</a></li></ol></li><li><h3>Module 7 · Place, route, timing</h3><p>Where the cells land on the die, and whether the clock is met.</p><ol><li><a href="/learn/placement/">19 Placement: where the cells land</a></li><li class="on" aria-current="page"><a href="/learn/logic-depth/">20 Logic depth</a></li><li><a href="/learn/timing-slack/">21 Timing and slack</a></li></ol></li><li><h3>Module 8 · The bitstream</h3><p>How a routed design becomes the bits the chip loads, with no Vivado.</p><ol><li><a href="/learn/fasm-to-frames/">22 From FASM to frames</a></li><li><a href="/learn/a-build-receipt/">23 A build, itemised</a></li><li><a href="/learn/read-a-bitstream/">24 Read a bitstream</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>Ask the chip who it is, load the bits, and check them.</p><ol><li><a href="/learn/jtag-asks-the-chip/">25 JTAG asks the chip</a></li><li><a href="/learn/same-bits-on-the-board/">26 The same bits, on the board</a></li><li><a href="/learn/capstone/">27 Capstone: out-guess yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/long-multiplication/index.html b/apps/website/public/learn/long-multiplication/index.html
index f83f7b653d..59a56ec07b 100644
--- a/apps/website/public/learn/long-multiplication/index.html
+++ b/apps/website/public/learn/long-multiplication/index.html
@@ -769,5 +769,6 @@ <h1>Long multiplication</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Three values</h3><p>Why three, how balanced ternary writes every number without a sign, and what flipping and cutting trits do.</p><ol><li><a href="/learn/why-three/">01 Why three</a></li><li><a href="/learn/balanced-ternary/">02 Balanced ternary</a></li><li><a href="/learn/flip-and-cut/">03 Flip to negate, cut to round</a></li></ol></li><li><h3>Module 2 · Logic with unknown</h3><p>Kleene's three-valued gates, two gates binary has no twin for, and addition as a pair of tables.</p><ol><li><a href="/learn/kleene-logic/">04 Three-valued logic</a></li><li><a href="/learn/consensus-and-any/">05 Consensus and any</a></li><li><a href="/learn/sum-and-carry/">06 Sum and carry</a></li></ol></li><li><h3>Module 3 · Adding trits</h3><p>The half adder, the full adder and a carry that ripples left, one place at a time.</p><ol><li><a href="/learn/half-adder/">07 The half adder</a></li><li><a href="/learn/full-adder/">08 The full adder</a></li><li><a href="/learn/ripple-carry/">09 Ripple carry</a></li></ol></li><li><h3>Module 4 · Multiplying without a multiplier</h3><p>Copy, drop or flip: a product by one trit, long multiplication, and a MAC that only adds.</p><ol><li><a href="/learn/times-a-trit/">10 Times one trit</a></li><li class="on" aria-current="page"><a href="/learn/long-multiplication/">11 Long multiplication</a></li><li><a href="/learn/ternary-mac/">12 A ternary MAC</a></li></ol></li><li><h3>Module 5 · Trits in binary memory</h3><p>Two bits per trit, five trits per byte, and the bits a 27-trit word needs.</p><ol><li><a href="/learn/two-bits-per-trit/">13 Two bits per trit</a></li><li><a href="/learn/five-trits-a-byte/">14 Five trits in a byte</a></li><li><a href="/learn/a-word-of-27/">15 A word of 27 trits</a></li></ol></li><li><h3>Module 6 · The TRI-27 instruction word</h3><p>The 32-bit word the Trinity emulator decodes: its fields, its 47 opcodes and its 15-bit immediate.</p><ol><li><a href="/learn/the-instruction-word/">16 The instruction word</a></li><li><a href="/learn/the-opcode-map/">17 The opcode map</a></li><li><a href="/learn/the-immediate/">18 The immediate</a></li></ol></li><li><h3>Module 7 · A ternary machine</h3><p>An ALU built from this course's adders, a three-way jump, and a program you can step.</p><ol><li><a href="/learn/a-ternary-alu/">19 A ternary ALU</a></li><li><a href="/learn/three-way-jump/">20 A three-way jump</a></li><li><a href="/learn/run-a-program/">21 Run a program</a></li></ol></li><li><h3>Module 8 · Three answers</h3><p>Compare at the highest differing trit, find a number in thirds, sort with three-way compares.</p><ol><li><a href="/learn/compare-at-the-top/">22 Compare at the top</a></li><li><a href="/learn/questions-in-thirds/">23 Questions in thirds</a></li><li><a href="/learn/sort-by-three/">24 Sort with three-way compares</a></li></ol></li><li><h3>Module 9 · Ternary neurons</h3><p>Weights of -1, 0, +1: one neuron, a detector and a small layer, with no multiplier anywhere.</p><ol><li><a href="/learn/one-ternary-neuron/">25 One ternary neuron</a></li><li><a href="/learn/a-bar-detector/">26 A bar detector</a></li><li><a href="/learn/a-tiny-layer/">27 A tiny layer</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/loops/index.html b/apps/website/public/learn/loops/index.html
index 1f389bb4ad..666f6cff2c 100644
--- a/apps/website/public/learn/loops/index.html
+++ b/apps/website/public/learn/loops/index.html
@@ -97,5 +97,6 @@ <h1>Loops: while and for</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · A spec file</h3><p>What a .t27 file holds: a module line, prose lines and comments.</p><ol><li><a href="/learn/what-a-spec-is/">01 What a spec is</a></li><li><a href="/learn/the-module-line/">02 The module line</a></li><li><a href="/learn/comments-and-prose/">03 Comments and prose lines</a></li></ol></li><li><h3>Module 2 · Values and types</h3><p>Constants, the width of a whole number, and true, false and text.</p><ol><li><a href="/learn/constants/">04 Constants</a></li><li><a href="/learn/whole-numbers/">05 Whole numbers and their widths</a></li><li><a href="/learn/true-false-and-text/">06 True, false and text</a></li></ol></li><li><h3>Module 3 · Arrays, trits and expressions</h3><p>Lists of values, the three values of a trit, and what operators compute.</p><ol><li><a href="/learn/arrays/">07 Arrays</a></li><li><a href="/learn/trits/">08 Trits: minus one, zero, one</a></li><li><a href="/learn/expressions/">09 Expressions</a></li></ol></li><li><h3>Module 4 · Tests and functions</h3><p>A test block, several small tests in one spec, and a function with typed inputs.</p><ol><li><a href="/learn/a-test-block/">10 A test block</a></li><li><a href="/learn/many-tests/">11 Many tests in one spec</a></li><li><a href="/learn/functions/">12 Functions</a></li></ol></li><li><h3>Module 5 · Inside a function</h3><p>Names with let and var, choices with if and switch, and loops.</p><ol><li><a href="/learn/local-names/">13 Local names: let and var</a></li><li><a href="/learn/if-and-else/">14 Choices: if, else and switch</a></li><li class="on" aria-current="page"><a href="/learn/loops/">15 Loops: while and for</a></li></ol></li><li><h3>Module 6 · Visibility and your own types</h3><p>What pub shares with other modules, and the structs and enums you declare.</p><ol><li><a href="/learn/pub-or-private/">16 pub or private</a></li><li><a href="/learn/structs/">17 Structs</a></li><li><a href="/learn/enums/">18 Enums</a></li></ol></li><li><h3>Module 7 · Modules, rules and gen-ts</h3><p>Reading another module, stating a rule that always holds, and a spec turned into TypeScript.</p><ol><li><a href="/learn/use-other-modules/">19 Using other modules</a></li><li><a href="/learn/invariants/">20 Invariants and benches</a></li><li><a href="/learn/gen-ts/">21 gen-ts: a spec becomes TypeScript</a></li></ol></li><li><h3>Module 8 · The compiler and tri</h3><p>The backends t27c writes, the stages it reads a spec in, and the tri command.</p><ol><li><a href="/learn/seven-backends/">22 Seven backends</a></li><li><a href="/learn/stages-of-t27c/">23 The stages of t27c</a></li><li><a href="/learn/tri-commands/">24 tri: the command line</a></li></ol></li><li><h3>Module 9 · Errors, a program, what next</h3><p>Reading a compiler error, one small complete program, and the course to take next.</p><ol><li><a href="/learn/reading-errors/">25 Reading compiler errors</a></li><li><a href="/learn/a-small-program/">26 A small complete program</a></li><li><a href="/learn/where-next/">27 Where to go next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/lowering-the-rate/index.html b/apps/website/public/learn/lowering-the-rate/index.html
index f9cd03d662..35b2688c99 100644
--- a/apps/website/public/learn/lowering-the-rate/index.html
+++ b/apps/website/public/learn/lowering-the-rate/index.html
@@ -1122,5 +1122,6 @@ <h1>Lowering the rate</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Adding numbers</h3><p>Why addition is slow: the carry that walks up the word, the dedicated chain in every slice, prefix networks and carry-save trees.</p><ol><li><a href="/learn/the-carry-chain/">01 The carry chain</a></li><li><a href="/learn/prefix-adders/">02 Looking ahead: prefix adders</a></li><li><a href="/learn/carry-save/">03 Three numbers at once</a></li></ol></li><li><h3>Module 2 · Multiplying</h3><p>A product is a sum of shifted copies; Booth recoding halves them, and the DSP48E1 slice does the rest in one block.</p><ol><li><a href="/learn/shift-and-add/">04 Shift and add</a></li><li><a href="/learn/booths-trick/">05 Booth's trick</a></li><li><a href="/learn/the-dsp48e1-slice/">06 The DSP48E1 slice</a></li></ol></li><li><h3>Module 3 · Fixed point</h3><p>Where the binary point sits, what rounding does to a value and to its average, and what each bit of a quantizer buys.</p><ol><li><a href="/learn/q-formats/">07 Q formats</a></li><li><a href="/learn/rounding-and-saturation/">08 Rounding and saturation</a></li><li><a href="/learn/quantization-noise/">09 Quantization noise</a></li></ol></li><li><h3>Module 4 · Functions in hardware</h3><p>Sine, cosine, angle and length from shifts and adds, and when a table is the better answer.</p><ol><li><a href="/learn/rotating-by-shifts/">10 Rotating by shifts</a></li><li><a href="/learn/angle-and-magnitude/">11 Angle and magnitude</a></li><li><a href="/learn/tables-and-interpolation/">12 Tables and interpolation</a></li></ol></li><li><h3>Module 5 · Signals and sampling</h3><p>What sampling does to a frequency, how an oscillator is built from an adder, and what a DFT bin measures.</p><ol><li><a href="/learn/sampling-and-aliasing/">13 Sampling and aliasing</a></li><li><a href="/learn/the-nco/">14 A numerically controlled oscillator</a></li><li><a href="/learn/bins-of-a-spectrum/">15 Bins of a spectrum</a></li></ol></li><li><h3>Module 6 · FIR filters</h3><p>The moving average, a windowed-sinc design with integer taps, and folding the taps onto DSP slices.</p><ol><li><a href="/learn/the-moving-average/">16 The moving average</a></li><li><a href="/learn/designing-taps/">17 Designing taps</a></li><li><a href="/learn/folding-taps-onto-slices/">18 Folding taps onto slices</a></li></ol></li><li><h3>Module 7 · Multirate</h3><p>Lowering the sample rate without folding noise in: decimation, the CIC filter and the polyphase form.</p><ol><li class="on" aria-current="page"><a href="/learn/lowering-the-rate/">19 Lowering the rate</a></li><li><a href="/learn/the-cic-filter/">20 The CIC filter</a></li><li><a href="/learn/polyphase/">21 Polyphase</a></li></ol></li><li><h3>Module 8 · The FFT</h3><p>N log N instead of N^2: butterflies, the bit-reversed input order and the bits each stage adds.</p><ol><li><a href="/learn/the-butterfly/">22 The butterfly</a></li><li><a href="/learn/twiddles-and-bit-reversal/">23 Twiddles and bit reversal</a></li><li><a href="/learn/growth-through-the-stages/">24 Growth through the stages</a></li></ol></li><li><h3>Module 9 · On the bench</h3><p>From the arithmetic to the board: a correlator from the modem spec, a budget of slices and timing, and the capstone filter.</p><ol><li><a href="/learn/a-correlator-finds-the-frame/">25 A correlator finds the frame</a></li><li><a href="/learn/does-it-fit-and-close-timing/">26 Does it fit and close timing?</a></li><li><a href="/learn/capstone-a-filter-on-the-board/">27 Capstone: a filter on the board</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/lucas-numbers-stay-whole/index.html b/apps/website/public/learn/lucas-numbers-stay-whole/index.html
index 676f930a6d..c6cf5ccf62 100644
--- a/apps/website/public/learn/lucas-numbers-stay-whole/index.html
+++ b/apps/website/public/learn/lucas-numbers-stay-whole/index.html
@@ -207,5 +207,6 @@ <h1>Lucas numbers stay whole</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The rule and its numbers</h3><p>One rule splits every width, the ratio it aims at, and the Lucas numbers behind the 3.</p><ol><li><a href="/learn/a-float-cut-by-phi/">01 A float cut by phi</a></li><li><a href="/learn/phi-as-a-ratio/">02 Phi as a ratio</a></li><li class="on" aria-current="page"><a href="/learn/lucas-numbers-stay-whole/">03 Lucas numbers stay whole</a></li></ol></li><li><h3>Module 2 · Why phi, why three</h3><p>Why the split is phi, why base three, and how a spec checks GF16 keeps phi.</p><ol><li><a href="/learn/why-the-split-is-phi/">04 Why the split is phi</a></li><li><a href="/learn/why-base-three/">05 Why base three</a></li><li><a href="/learn/phi-in-sixteen-bits/">06 Phi in sixteen bits</a></li></ol></li><li><h3>Module 3 · The small rungs: GF4 to GF8</h3><p>GF4, GF6 and GF8, the fewest bits, where rounding to whole bits costs the most.</p><ol><li><a href="/learn/gf4-four-bits/">07 GF4: four bits</a></li><li><a href="/learn/gf6-six-bits/">08 GF6: six bits</a></li><li><a href="/learn/gf8-one-byte/">09 GF8: one byte</a></li></ol></li><li><h3>Module 4 · Ten to fourteen bits</h3><p>GF10, GF12 and GF14, and how the distance from 1 / phi moves as the word grows.</p><ol><li><a href="/learn/gf10-ten-bits/">10 GF10: ten bits</a></li><li><a href="/learn/gf12-twelve-bits/">11 GF12: twelve bits</a></li><li><a href="/learn/gf14-fourteen-bits/">12 GF14: fourteen bits</a></li></ol></li><li><h3>Module 5 · GF16 at work</h3><p>The primary 16-bit format, a two-term dot product in GF-T16, then GF20 and GF24.</p><ol><li><a href="/learn/gf16-the-primary/">13 GF16: the primary format</a></li><li><a href="/learn/two-products-one-sum/">14 Two products, one sum</a></li><li><a href="/learn/gf24-twenty-four-bits/">15 GF20 and GF24</a></li></ol></li><li><h3>Module 6 · GF32 to GF64</h3><p>GF32 beside IEEE single, GF48 with no IEEE twin, GF64 beside IEEE double.</p><ol><li><a href="/learn/gf32-a-single/">16 GF32: a single</a></li><li><a href="/learn/gf48-forty-eight-bits/">17 GF48: forty-eight bits</a></li><li><a href="/learn/gf64-a-double/">18 GF64: a double</a></li></ol></li><li><h3>Module 7 · GF96 to GF256</h3><p>GF96, GF128 and GF256, where the specs hold the layout with invariants.</p><ol><li><a href="/learn/gf96-ninety-six-bits/">19 GF96: ninety-six bits</a></li><li><a href="/learn/gf128-a-quad/">20 GF128: a quad</a></li><li><a href="/learn/gf256-two-five-six/">21 GF256: two hundred fifty-six bits</a></li></ol></li><li><h3>Module 8 · The widest rungs, then trits</h3><p>GF512 and GF1024, the two widest rungs, then GF-T8, where the exponent moves to trits.</p><ol><li><a href="/learn/gf512-five-twelve/">22 GF512: five hundred twelve bits</a></li><li><a href="/learn/gf1024-one-kilobit/">23 GF1024: one kilobit</a></li><li><a href="/learn/gft8-exponent-in-trits/">24 GF-T8: an exponent in trits</a></li></ol></li><li><h3>Module 9 · More trits, then the decode</h3><p>GF-T16 and GF-T32, then why fixed fields decode in parallel and a posit does not.</p><ol><li><a href="/learn/gft16-six-trits/">25 GF-T16: six trits</a></li><li><a href="/learn/gft32-twelve-trits/">26 GF-T32: twelve trits</a></li><li><a href="/learn/decode-in-one-step/">27 Decode in one step</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/many-tests/index.html b/apps/website/public/learn/many-tests/index.html
index 2ab09b8540..8017a69212 100644
--- a/apps/website/public/learn/many-tests/index.html
+++ b/apps/website/public/learn/many-tests/index.html
@@ -93,5 +93,6 @@ <h1>Many tests in one spec</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · A spec file</h3><p>What a .t27 file holds: a module line, prose lines and comments.</p><ol><li><a href="/learn/what-a-spec-is/">01 What a spec is</a></li><li><a href="/learn/the-module-line/">02 The module line</a></li><li><a href="/learn/comments-and-prose/">03 Comments and prose lines</a></li></ol></li><li><h3>Module 2 · Values and types</h3><p>Constants, the width of a whole number, and true, false and text.</p><ol><li><a href="/learn/constants/">04 Constants</a></li><li><a href="/learn/whole-numbers/">05 Whole numbers and their widths</a></li><li><a href="/learn/true-false-and-text/">06 True, false and text</a></li></ol></li><li><h3>Module 3 · Arrays, trits and expressions</h3><p>Lists of values, the three values of a trit, and what operators compute.</p><ol><li><a href="/learn/arrays/">07 Arrays</a></li><li><a href="/learn/trits/">08 Trits: minus one, zero, one</a></li><li><a href="/learn/expressions/">09 Expressions</a></li></ol></li><li><h3>Module 4 · Tests and functions</h3><p>A test block, several small tests in one spec, and a function with typed inputs.</p><ol><li><a href="/learn/a-test-block/">10 A test block</a></li><li class="on" aria-current="page"><a href="/learn/many-tests/">11 Many tests in one spec</a></li><li><a href="/learn/functions/">12 Functions</a></li></ol></li><li><h3>Module 5 · Inside a function</h3><p>Names with let and var, choices with if and switch, and loops.</p><ol><li><a href="/learn/local-names/">13 Local names: let and var</a></li><li><a href="/learn/if-and-else/">14 Choices: if, else and switch</a></li><li><a href="/learn/loops/">15 Loops: while and for</a></li></ol></li><li><h3>Module 6 · Visibility and your own types</h3><p>What pub shares with other modules, and the structs and enums you declare.</p><ol><li><a href="/learn/pub-or-private/">16 pub or private</a></li><li><a href="/learn/structs/">17 Structs</a></li><li><a href="/learn/enums/">18 Enums</a></li></ol></li><li><h3>Module 7 · Modules, rules and gen-ts</h3><p>Reading another module, stating a rule that always holds, and a spec turned into TypeScript.</p><ol><li><a href="/learn/use-other-modules/">19 Using other modules</a></li><li><a href="/learn/invariants/">20 Invariants and benches</a></li><li><a href="/learn/gen-ts/">21 gen-ts: a spec becomes TypeScript</a></li></ol></li><li><h3>Module 8 · The compiler and tri</h3><p>The backends t27c writes, the stages it reads a spec in, and the tri command.</p><ol><li><a href="/learn/seven-backends/">22 Seven backends</a></li><li><a href="/learn/stages-of-t27c/">23 The stages of t27c</a></li><li><a href="/learn/tri-commands/">24 tri: the command line</a></li></ol></li><li><h3>Module 9 · Errors, a program, what next</h3><p>Reading a compiler error, one small complete program, and the course to take next.</p><ol><li><a href="/learn/reading-errors/">25 Reading compiler errors</a></li><li><a href="/learn/a-small-program/">26 A small complete program</a></li><li><a href="/learn/where-next/">27 Where to go next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/master-and-servant/index.html b/apps/website/public/learn/master-and-servant/index.html
index da8b4440e1..601949b68c 100644
--- a/apps/website/public/learn/master-and-servant/index.html
+++ b/apps/website/public/learn/master-and-servant/index.html
@@ -417,5 +417,6 @@ <h1>Master and servant</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a bus is</h3><p>Why a bus exists at all: a conversation on wires, framed and addressed, and who is allowed to talk.</p><ol><li><a href="/learn/wires-that-talk/">01 Wires that talk</a></li><li><a href="/learn/framing-a-conversation/">02 Framing a conversation</a></li><li class="on" aria-current="page"><a href="/learn/master-and-servant/">03 Master and servant</a></li></ol></li><li><h3>Module 2 · UART</h3><p>The two-wire bus with no clock: the frame, the divisor that sets the rate, and the status a driver polls.</p><ol><li><a href="/learn/the-frame/">04 The frame: start, data, stop</a></li><li><a href="/learn/baud-and-the-divisor/">05 Baud and the divisor</a></li><li><a href="/learn/status-and-fifos/">06 Status and FIFOs</a></li></ol></li><li><h3>Module 3 · SPI</h3><p>The clocked conversation: four modes, a prescaler ladder, and a chip select per servant.</p><ol><li><a href="/learn/modes-cpol-cpha/">07 Modes: CPOL and CPHA</a></li><li><a href="/learn/the-prescaler-ladder/">08 The prescaler ladder</a></li><li><a href="/learn/chip-select-and-widths/">09 Chip select and widths</a></li></ol></li><li><h3>Module 4 · APB</h3><p>The register bus: PSEL and PENABLE, strobes and wait states, and how many address bits a peripheral count costs.</p><ol><li><a href="/learn/psel-and-penable/">10 PSEL and PENABLE</a></li><li><a href="/learn/strobes-and-waits/">11 Strobes and waits</a></li><li><a href="/learn/address-decode/">12 Address decode</a></li></ol></li><li><h3>Module 5 · AXI4</h3><p>The five channels: address, data and response in both directions, lite or full, bursts and IDs.</p><ol><li><a href="/learn/five-channels/">13 Five channels</a></li><li><a href="/learn/lite-or-full/">14 Lite or full</a></li><li><a href="/learn/bursts-and-ids/">15 Bursts and IDs</a></li></ol></li><li><h3>Module 6 · Memory</h3><p>What sits on the far side of every bus: memory maps, port kinds, and latency that a wait state must cover.</p><ol><li><a href="/learn/memory-maps/">16 Memory maps</a></li><li><a href="/learn/ports-and-kinds/">17 Ports and kinds</a></li><li><a href="/learn/latency/">18 Latency</a></li></ol></li><li><h3>Module 7 · Bridges</h3><p>Why designs grow more than one bus, and the packet bridge that moves work between them.</p><ol><li><a href="/learn/why-bridges/">19 Why bridges</a></li><li><a href="/learn/the-packet-bridge/">20 The packet bridge</a></li><li><a href="/learn/everything-in-one-bench/">21 Everything in one bench</a></li></ol></li><li><h3>Module 8 · Ethernet</h3><p>Frames, the frame check sequence, RGMII timing, and the pre-registered steps of a real bring-up.</p><ol><li><a href="/learn/frames-and-the-fcs/">22 Frames and the FCS</a></li><li><a href="/learn/rgmii-timing/">23 RGMII timing</a></li><li><a href="/learn/bring-up-on-the-bench/">24 Bring-up on the bench</a></li></ol></li><li><h3>Module 9 · The bench</h3><p>The discipline that guards real hardware: who holds the IO, taking and giving the lock, and what runs next.</p><ol><li><a href="/learn/who-holds-the-io/">25 Who holds the IO</a></li><li><a href="/learn/claim-and-release/">26 Claim and release</a></li><li><a href="/learn/capstone-what-runs-next/">27 Capstone: what runs next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/memory-maps/index.html b/apps/website/public/learn/memory-maps/index.html
index 5e6a6b5d06..4c35ec19c9 100644
--- a/apps/website/public/learn/memory-maps/index.html
+++ b/apps/website/public/learn/memory-maps/index.html
@@ -424,5 +424,6 @@ <h1>Memory maps</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a bus is</h3><p>Why a bus exists at all: a conversation on wires, framed and addressed, and who is allowed to talk.</p><ol><li><a href="/learn/wires-that-talk/">01 Wires that talk</a></li><li><a href="/learn/framing-a-conversation/">02 Framing a conversation</a></li><li><a href="/learn/master-and-servant/">03 Master and servant</a></li></ol></li><li><h3>Module 2 · UART</h3><p>The two-wire bus with no clock: the frame, the divisor that sets the rate, and the status a driver polls.</p><ol><li><a href="/learn/the-frame/">04 The frame: start, data, stop</a></li><li><a href="/learn/baud-and-the-divisor/">05 Baud and the divisor</a></li><li><a href="/learn/status-and-fifos/">06 Status and FIFOs</a></li></ol></li><li><h3>Module 3 · SPI</h3><p>The clocked conversation: four modes, a prescaler ladder, and a chip select per servant.</p><ol><li><a href="/learn/modes-cpol-cpha/">07 Modes: CPOL and CPHA</a></li><li><a href="/learn/the-prescaler-ladder/">08 The prescaler ladder</a></li><li><a href="/learn/chip-select-and-widths/">09 Chip select and widths</a></li></ol></li><li><h3>Module 4 · APB</h3><p>The register bus: PSEL and PENABLE, strobes and wait states, and how many address bits a peripheral count costs.</p><ol><li><a href="/learn/psel-and-penable/">10 PSEL and PENABLE</a></li><li><a href="/learn/strobes-and-waits/">11 Strobes and waits</a></li><li><a href="/learn/address-decode/">12 Address decode</a></li></ol></li><li><h3>Module 5 · AXI4</h3><p>The five channels: address, data and response in both directions, lite or full, bursts and IDs.</p><ol><li><a href="/learn/five-channels/">13 Five channels</a></li><li><a href="/learn/lite-or-full/">14 Lite or full</a></li><li><a href="/learn/bursts-and-ids/">15 Bursts and IDs</a></li></ol></li><li><h3>Module 6 · Memory</h3><p>What sits on the far side of every bus: memory maps, port kinds, and latency that a wait state must cover.</p><ol><li class="on" aria-current="page"><a href="/learn/memory-maps/">16 Memory maps</a></li><li><a href="/learn/ports-and-kinds/">17 Ports and kinds</a></li><li><a href="/learn/latency/">18 Latency</a></li></ol></li><li><h3>Module 7 · Bridges</h3><p>Why designs grow more than one bus, and the packet bridge that moves work between them.</p><ol><li><a href="/learn/why-bridges/">19 Why bridges</a></li><li><a href="/learn/the-packet-bridge/">20 The packet bridge</a></li><li><a href="/learn/everything-in-one-bench/">21 Everything in one bench</a></li></ol></li><li><h3>Module 8 · Ethernet</h3><p>Frames, the frame check sequence, RGMII timing, and the pre-registered steps of a real bring-up.</p><ol><li><a href="/learn/frames-and-the-fcs/">22 Frames and the FCS</a></li><li><a href="/learn/rgmii-timing/">23 RGMII timing</a></li><li><a href="/learn/bring-up-on-the-bench/">24 Bring-up on the bench</a></li></ol></li><li><h3>Module 9 · The bench</h3><p>The discipline that guards real hardware: who holds the IO, taking and giving the lock, and what runs next.</p><ol><li><a href="/learn/who-holds-the-io/">25 Who holds the IO</a></li><li><a href="/learn/claim-and-release/">26 Claim and release</a></li><li><a href="/learn/capstone-what-runs-next/">27 Capstone: what runs next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/modes-cpol-cpha/index.html b/apps/website/public/learn/modes-cpol-cpha/index.html
index da5574e7ba..9e2e6f4d9f 100644
--- a/apps/website/public/learn/modes-cpol-cpha/index.html
+++ b/apps/website/public/learn/modes-cpol-cpha/index.html
@@ -476,5 +476,6 @@ <h1>Modes: CPOL and CPHA</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a bus is</h3><p>Why a bus exists at all: a conversation on wires, framed and addressed, and who is allowed to talk.</p><ol><li><a href="/learn/wires-that-talk/">01 Wires that talk</a></li><li><a href="/learn/framing-a-conversation/">02 Framing a conversation</a></li><li><a href="/learn/master-and-servant/">03 Master and servant</a></li></ol></li><li><h3>Module 2 · UART</h3><p>The two-wire bus with no clock: the frame, the divisor that sets the rate, and the status a driver polls.</p><ol><li><a href="/learn/the-frame/">04 The frame: start, data, stop</a></li><li><a href="/learn/baud-and-the-divisor/">05 Baud and the divisor</a></li><li><a href="/learn/status-and-fifos/">06 Status and FIFOs</a></li></ol></li><li><h3>Module 3 · SPI</h3><p>The clocked conversation: four modes, a prescaler ladder, and a chip select per servant.</p><ol><li class="on" aria-current="page"><a href="/learn/modes-cpol-cpha/">07 Modes: CPOL and CPHA</a></li><li><a href="/learn/the-prescaler-ladder/">08 The prescaler ladder</a></li><li><a href="/learn/chip-select-and-widths/">09 Chip select and widths</a></li></ol></li><li><h3>Module 4 · APB</h3><p>The register bus: PSEL and PENABLE, strobes and wait states, and how many address bits a peripheral count costs.</p><ol><li><a href="/learn/psel-and-penable/">10 PSEL and PENABLE</a></li><li><a href="/learn/strobes-and-waits/">11 Strobes and waits</a></li><li><a href="/learn/address-decode/">12 Address decode</a></li></ol></li><li><h3>Module 5 · AXI4</h3><p>The five channels: address, data and response in both directions, lite or full, bursts and IDs.</p><ol><li><a href="/learn/five-channels/">13 Five channels</a></li><li><a href="/learn/lite-or-full/">14 Lite or full</a></li><li><a href="/learn/bursts-and-ids/">15 Bursts and IDs</a></li></ol></li><li><h3>Module 6 · Memory</h3><p>What sits on the far side of every bus: memory maps, port kinds, and latency that a wait state must cover.</p><ol><li><a href="/learn/memory-maps/">16 Memory maps</a></li><li><a href="/learn/ports-and-kinds/">17 Ports and kinds</a></li><li><a href="/learn/latency/">18 Latency</a></li></ol></li><li><h3>Module 7 · Bridges</h3><p>Why designs grow more than one bus, and the packet bridge that moves work between them.</p><ol><li><a href="/learn/why-bridges/">19 Why bridges</a></li><li><a href="/learn/the-packet-bridge/">20 The packet bridge</a></li><li><a href="/learn/everything-in-one-bench/">21 Everything in one bench</a></li></ol></li><li><h3>Module 8 · Ethernet</h3><p>Frames, the frame check sequence, RGMII timing, and the pre-registered steps of a real bring-up.</p><ol><li><a href="/learn/frames-and-the-fcs/">22 Frames and the FCS</a></li><li><a href="/learn/rgmii-timing/">23 RGMII timing</a></li><li><a href="/learn/bring-up-on-the-bench/">24 Bring-up on the bench</a></li></ol></li><li><h3>Module 9 · The bench</h3><p>The discipline that guards real hardware: who holds the IO, taking and giving the lock, and what runs next.</p><ol><li><a href="/learn/who-holds-the-io/">25 Who holds the IO</a></li><li><a href="/learn/claim-and-release/">26 Claim and release</a></li><li><a href="/learn/capstone-what-runs-next/">27 Capstone: what runs next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/mtbf/index.html b/apps/website/public/learn/mtbf/index.html
index 8e27fbddba..f6e2f04c57 100644
--- a/apps/website/public/learn/mtbf/index.html
+++ b/apps/website/public/learn/mtbf/index.html
@@ -330,5 +330,6 @@ <h1>Mean time between failures</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a clock is</h3><p>One edge, one world: what shares a clock edge shares a world, the period and the jitter of a real edge, and where the clock enters a board.</p><ol><li><a href="/learn/one-edge-one-world/">01 One edge, one world</a></li><li><a href="/learn/period-and-jitter/">02 A period, an edge, and jitter</a></li><li><a href="/learn/where-the-clock-enters/">03 Where the clock enters</a></li></ol></li><li><h3>Module 2 · Clock trees</h3><p>Skew and insertion delay, the global buffer network, and the trap of gating a clock with logic.</p><ol><li><a href="/learn/skew-and-insertion/">04 Skew and insertion delay</a></li><li><a href="/learn/global-buffers/">05 The global buffer</a></li><li><a href="/learn/gated-clock-trap/">06 The gated-clock trap</a></li></ol></li><li><h3>Module 3 · PLL and MMCM</h3><p>Multiply and divide one clock into another, move its phase in steps of the VCO, and which clocks the analyzer treats as related.</p><ol><li><a href="/learn/multiply-and-divide/">07 Multiply and divide</a></li><li><a href="/learn/phase-and-lock/">08 Phase and lock</a></li><li><a href="/learn/related-or-not/">09 Related or not</a></li></ol></li><li><h3>Module 4 · Resets</h3><p>Assert asynchronously, release synchronously: the three reset kinds, the release pipe, and the tree a reset grows.</p><ol><li><a href="/learn/sync-or-async/">10 Sync or async</a></li><li><a href="/learn/reset-release/">11 The release pipe</a></li><li><a href="/learn/reset-fanout/">12 Reset fanout</a></li></ol></li><li><h3>Module 5 · Metastability</h3><p>The setup-hold window, the mean time between failures in integer arithmetic, and the two flops that fix it.</p><ol><li><a href="/learn/setup-hold-window/">13 The setup-hold window</a></li><li class="on" aria-current="page"><a href="/learn/mtbf/">14 Mean time between failures</a></li><li><a href="/learn/two-flop-sync/">15 The two-flop synchronizer</a></li></ol></li><li><h3>Module 6 · Crossing many bits</h3><p>Why a binary bus tears, why Gray code does not, and the handshake that moves a pulse between worlds.</p><ol><li><a href="/learn/why-binary-fails/">16 Why binary fails</a></li><li><a href="/learn/gray-code/">17 Gray code</a></li><li><a href="/learn/handshakes-and-pulses/">18 Handshakes and pulses</a></li></ol></li><li><h3>Module 7 · The asynchronous FIFO</h3><p>Pointers, flags and depth: the buffer that moves a stream between two clocks.</p><ol><li><a href="/learn/pointers-and-flags/">19 Pointers and flags</a></li><li><a href="/learn/depth-sizing/">20 Sizing the depth</a></li><li><a href="/learn/fifo-testbench/">21 The FIFO testbench</a></li></ol></li><li><h3>Module 8 · Constraints</h3><p>The lines that tell the analyzer what a clock is, which paths not to check, and what the pins must meet.</p><ol><li><a href="/learn/create-clock/">22 create_clock</a></li><li><a href="/learn/false-path-and-max-delay/">23 False paths and max delay</a></li><li><a href="/learn/io-timing/">24 IO timing</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>A CDC report, one crossing captured at the flip-flops, and the bitstream diff that closes the course.</p><ol><li><a href="/learn/cdc-report/">25 The CDC report</a></li><li><a href="/learn/capture-a-crossing/">26 Capture a crossing</a></li><li><a href="/learn/capstone-diff/">27 Capstone: change it and diff it</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/multiply-and-divide/index.html b/apps/website/public/learn/multiply-and-divide/index.html
index 4e4a7d3619..e12103ecfb 100644
--- a/apps/website/public/learn/multiply-and-divide/index.html
+++ b/apps/website/public/learn/multiply-and-divide/index.html
@@ -341,5 +341,6 @@ <h1>Multiply and divide</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a clock is</h3><p>One edge, one world: what shares a clock edge shares a world, the period and the jitter of a real edge, and where the clock enters a board.</p><ol><li><a href="/learn/one-edge-one-world/">01 One edge, one world</a></li><li><a href="/learn/period-and-jitter/">02 A period, an edge, and jitter</a></li><li><a href="/learn/where-the-clock-enters/">03 Where the clock enters</a></li></ol></li><li><h3>Module 2 · Clock trees</h3><p>Skew and insertion delay, the global buffer network, and the trap of gating a clock with logic.</p><ol><li><a href="/learn/skew-and-insertion/">04 Skew and insertion delay</a></li><li><a href="/learn/global-buffers/">05 The global buffer</a></li><li><a href="/learn/gated-clock-trap/">06 The gated-clock trap</a></li></ol></li><li><h3>Module 3 · PLL and MMCM</h3><p>Multiply and divide one clock into another, move its phase in steps of the VCO, and which clocks the analyzer treats as related.</p><ol><li class="on" aria-current="page"><a href="/learn/multiply-and-divide/">07 Multiply and divide</a></li><li><a href="/learn/phase-and-lock/">08 Phase and lock</a></li><li><a href="/learn/related-or-not/">09 Related or not</a></li></ol></li><li><h3>Module 4 · Resets</h3><p>Assert asynchronously, release synchronously: the three reset kinds, the release pipe, and the tree a reset grows.</p><ol><li><a href="/learn/sync-or-async/">10 Sync or async</a></li><li><a href="/learn/reset-release/">11 The release pipe</a></li><li><a href="/learn/reset-fanout/">12 Reset fanout</a></li></ol></li><li><h3>Module 5 · Metastability</h3><p>The setup-hold window, the mean time between failures in integer arithmetic, and the two flops that fix it.</p><ol><li><a href="/learn/setup-hold-window/">13 The setup-hold window</a></li><li><a href="/learn/mtbf/">14 Mean time between failures</a></li><li><a href="/learn/two-flop-sync/">15 The two-flop synchronizer</a></li></ol></li><li><h3>Module 6 · Crossing many bits</h3><p>Why a binary bus tears, why Gray code does not, and the handshake that moves a pulse between worlds.</p><ol><li><a href="/learn/why-binary-fails/">16 Why binary fails</a></li><li><a href="/learn/gray-code/">17 Gray code</a></li><li><a href="/learn/handshakes-and-pulses/">18 Handshakes and pulses</a></li></ol></li><li><h3>Module 7 · The asynchronous FIFO</h3><p>Pointers, flags and depth: the buffer that moves a stream between two clocks.</p><ol><li><a href="/learn/pointers-and-flags/">19 Pointers and flags</a></li><li><a href="/learn/depth-sizing/">20 Sizing the depth</a></li><li><a href="/learn/fifo-testbench/">21 The FIFO testbench</a></li></ol></li><li><h3>Module 8 · Constraints</h3><p>The lines that tell the analyzer what a clock is, which paths not to check, and what the pins must meet.</p><ol><li><a href="/learn/create-clock/">22 create_clock</a></li><li><a href="/learn/false-path-and-max-delay/">23 False paths and max delay</a></li><li><a href="/learn/io-timing/">24 IO timing</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>A CDC report, one crossing captured at the flip-flops, and the bitstream diff that closes the course.</p><ol><li><a href="/learn/cdc-report/">25 The CDC report</a></li><li><a href="/learn/capture-a-crossing/">26 Capture a crossing</a></li><li><a href="/learn/capstone-diff/">27 Capstone: change it and diff it</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/multiply-then-add/index.html b/apps/website/public/learn/multiply-then-add/index.html
index 1abb44203e..357073c965 100644
--- a/apps/website/public/learn/multiply-then-add/index.html
+++ b/apps/website/public/learn/multiply-then-add/index.html
@@ -239,5 +239,6 @@ <h1>Multiply, then add</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Lab: our own research</h3><p>A number format of our own, an honest scoreboard, and a model's tables multiplied on the board.</p><ol><li><a href="/learn/t27-number-format/">01 T27: a number format of our own</a></li><li><a href="/learn/an-honest-scoreboard/">02 An honest scoreboard</a></li><li><a href="/learn/a-model-on-the-board/">03 A model's tables on the board</a></li></ol></li><li><h3>Module 2 · AI numbers: the MX block</h3><p>How AI chips keep weights in a few bits: one shared scale per block, the scale byte itself, and what one outlier does to its neighbours.</p><ol><li><a href="/learn/one-scale-per-block/">04 One scale for a block of weights</a></li><li><a href="/learn/scale-byte-native/">05 The scale byte on a real machine</a></li><li><a href="/learn/one-outlier/">06 One outlier, many zeros</a></li></ol></li><li><h3>Module 3 · Ternary weights</h3><p>Weights that are only minus, zero or plus a scale, the five rules a ternary alphabet must pass, and a test pass that checked nothing.</p><ol><li><a href="/learn/phi-scaled-ternary/">07 Weights of minus phi, zero and plus phi</a></li><li><a href="/learn/a-pass-that-checked-nothing/">08 A pass that checked nothing</a></li><li><a href="/learn/five-rules-for-weights/">09 Five rules for a weight alphabet</a></li></ol></li><li><h3>Module 4 · The Ternary Network Float</h3><p>A rule the compiler enforces before any test runs, and a 17-bit float whose exponent is four balanced trits.</p><ol><li><a href="/learn/blocked-is-not-failed/">10 Blocked is not failed</a></li><li><a href="/learn/four-trit-exponent/">11 A float whose exponent is four trits</a></li><li><a href="/learn/negate-twice/">12 Negate twice, get the number back</a></li></ol></li><li><h3>Module 5 · Arithmetic on signed numbers</h3><p>Multiply two signed numbers, add them when their signs differ, and do both at once in a multiply-accumulate.</p><ol><li><a href="/learn/a-sign-is-one-bit/">13 A sign is one bit</a></li><li><a href="/learn/opposite-signs-subtract/">14 Opposite signs subtract</a></li><li class="on" aria-current="page"><a href="/learn/multiply-then-add/">15 Multiply, then add</a></li></ol></li><li><h3>Module 6 · Parts of a neuron</h3><p>A ReLU that bends at zero, a power of two for softmax, and an argmax that names the answer.</p><ol><li><a href="/learn/a-bend-at-zero/">16 A bend at zero</a></li><li><a href="/learn/two-to-the-x/">17 Two to the x</a></li><li><a href="/learn/pick-the-largest/">18 Pick the largest</a></li></ol></li><li><h3>Module 7 · Learning from a mistake</h3><p>A loss that prices a wrong guess in bits, one step that moves a weight against its gradient, and the hidden layer that XOR needs.</p><ol><li><a href="/learn/a-loss-in-bits/">19 A loss in bits</a></li><li><a href="/learn/one-step-downhill/">20 One step downhill</a></li><li><a href="/learn/xor-needs-a-bend/">21 XOR needs a bend</a></li></ol></li><li><h3>Module 8 · BitNet: ternary networks</h3><p>A threshold that squeezes a sum back to three values, one neuron that becomes a different function when its weights change, and a neuron that reads its inputs 27 trits at a time.</p><ol><li><a href="/learn/back-to-three-values/">22 Back to three values</a></li><li><a href="/learn/same-neuron-new-weights/">23 Same neuron, new weights</a></li><li><a href="/learn/a-neuron-in-chunks/">24 A neuron in chunks</a></li></ol></li><li><h3>Module 9 · The ternary MAC as a chip</h3><p>The 27-trit dot product as wires with no register, the same sum added into a register on every clock, and a small whole network to close the course.</p><ol><li><a href="/learn/a-dot-product-in-wires/">25 A dot product made of wires</a></li><li><a href="/learn/add-it-up-every-clock/">26 Add it up on every clock</a></li><li><a href="/learn/a-whole-network/">27 A whole network</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/negate-twice/index.html b/apps/website/public/learn/negate-twice/index.html
index 98adac95d2..ba0b1b53ab 100644
--- a/apps/website/public/learn/negate-twice/index.html
+++ b/apps/website/public/learn/negate-twice/index.html
@@ -436,5 +436,6 @@ <h1>Negate twice, get the number back</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Lab: our own research</h3><p>A number format of our own, an honest scoreboard, and a model's tables multiplied on the board.</p><ol><li><a href="/learn/t27-number-format/">01 T27: a number format of our own</a></li><li><a href="/learn/an-honest-scoreboard/">02 An honest scoreboard</a></li><li><a href="/learn/a-model-on-the-board/">03 A model's tables on the board</a></li></ol></li><li><h3>Module 2 · AI numbers: the MX block</h3><p>How AI chips keep weights in a few bits: one shared scale per block, the scale byte itself, and what one outlier does to its neighbours.</p><ol><li><a href="/learn/one-scale-per-block/">04 One scale for a block of weights</a></li><li><a href="/learn/scale-byte-native/">05 The scale byte on a real machine</a></li><li><a href="/learn/one-outlier/">06 One outlier, many zeros</a></li></ol></li><li><h3>Module 3 · Ternary weights</h3><p>Weights that are only minus, zero or plus a scale, the five rules a ternary alphabet must pass, and a test pass that checked nothing.</p><ol><li><a href="/learn/phi-scaled-ternary/">07 Weights of minus phi, zero and plus phi</a></li><li><a href="/learn/a-pass-that-checked-nothing/">08 A pass that checked nothing</a></li><li><a href="/learn/five-rules-for-weights/">09 Five rules for a weight alphabet</a></li></ol></li><li><h3>Module 4 · The Ternary Network Float</h3><p>A rule the compiler enforces before any test runs, and a 17-bit float whose exponent is four balanced trits.</p><ol><li><a href="/learn/blocked-is-not-failed/">10 Blocked is not failed</a></li><li><a href="/learn/four-trit-exponent/">11 A float whose exponent is four trits</a></li><li class="on" aria-current="page"><a href="/learn/negate-twice/">12 Negate twice, get the number back</a></li></ol></li><li><h3>Module 5 · Arithmetic on signed numbers</h3><p>Multiply two signed numbers, add them when their signs differ, and do both at once in a multiply-accumulate.</p><ol><li><a href="/learn/a-sign-is-one-bit/">13 A sign is one bit</a></li><li><a href="/learn/opposite-signs-subtract/">14 Opposite signs subtract</a></li><li><a href="/learn/multiply-then-add/">15 Multiply, then add</a></li></ol></li><li><h3>Module 6 · Parts of a neuron</h3><p>A ReLU that bends at zero, a power of two for softmax, and an argmax that names the answer.</p><ol><li><a href="/learn/a-bend-at-zero/">16 A bend at zero</a></li><li><a href="/learn/two-to-the-x/">17 Two to the x</a></li><li><a href="/learn/pick-the-largest/">18 Pick the largest</a></li></ol></li><li><h3>Module 7 · Learning from a mistake</h3><p>A loss that prices a wrong guess in bits, one step that moves a weight against its gradient, and the hidden layer that XOR needs.</p><ol><li><a href="/learn/a-loss-in-bits/">19 A loss in bits</a></li><li><a href="/learn/one-step-downhill/">20 One step downhill</a></li><li><a href="/learn/xor-needs-a-bend/">21 XOR needs a bend</a></li></ol></li><li><h3>Module 8 · BitNet: ternary networks</h3><p>A threshold that squeezes a sum back to three values, one neuron that becomes a different function when its weights change, and a neuron that reads its inputs 27 trits at a time.</p><ol><li><a href="/learn/back-to-three-values/">22 Back to three values</a></li><li><a href="/learn/same-neuron-new-weights/">23 Same neuron, new weights</a></li><li><a href="/learn/a-neuron-in-chunks/">24 A neuron in chunks</a></li></ol></li><li><h3>Module 9 · The ternary MAC as a chip</h3><p>The 27-trit dot product as wires with no register, the same sum added into a register on every clock, and a small whole network to close the course.</p><ol><li><a href="/learn/a-dot-product-in-wires/">25 A dot product made of wires</a></li><li><a href="/learn/add-it-up-every-clock/">26 Add it up on every clock</a></li><li><a href="/learn/a-whole-network/">27 A whole network</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/one-edge-one-world/index.html b/apps/website/public/learn/one-edge-one-world/index.html
index 105f75a299..e2591e3d1c 100644
--- a/apps/website/public/learn/one-edge-one-world/index.html
+++ b/apps/website/public/learn/one-edge-one-world/index.html
@@ -300,5 +300,6 @@ <h1>One edge, one world</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a clock is</h3><p>One edge, one world: what shares a clock edge shares a world, the period and the jitter of a real edge, and where the clock enters a board.</p><ol><li class="on" aria-current="page"><a href="/learn/one-edge-one-world/">01 One edge, one world</a></li><li><a href="/learn/period-and-jitter/">02 A period, an edge, and jitter</a></li><li><a href="/learn/where-the-clock-enters/">03 Where the clock enters</a></li></ol></li><li><h3>Module 2 · Clock trees</h3><p>Skew and insertion delay, the global buffer network, and the trap of gating a clock with logic.</p><ol><li><a href="/learn/skew-and-insertion/">04 Skew and insertion delay</a></li><li><a href="/learn/global-buffers/">05 The global buffer</a></li><li><a href="/learn/gated-clock-trap/">06 The gated-clock trap</a></li></ol></li><li><h3>Module 3 · PLL and MMCM</h3><p>Multiply and divide one clock into another, move its phase in steps of the VCO, and which clocks the analyzer treats as related.</p><ol><li><a href="/learn/multiply-and-divide/">07 Multiply and divide</a></li><li><a href="/learn/phase-and-lock/">08 Phase and lock</a></li><li><a href="/learn/related-or-not/">09 Related or not</a></li></ol></li><li><h3>Module 4 · Resets</h3><p>Assert asynchronously, release synchronously: the three reset kinds, the release pipe, and the tree a reset grows.</p><ol><li><a href="/learn/sync-or-async/">10 Sync or async</a></li><li><a href="/learn/reset-release/">11 The release pipe</a></li><li><a href="/learn/reset-fanout/">12 Reset fanout</a></li></ol></li><li><h3>Module 5 · Metastability</h3><p>The setup-hold window, the mean time between failures in integer arithmetic, and the two flops that fix it.</p><ol><li><a href="/learn/setup-hold-window/">13 The setup-hold window</a></li><li><a href="/learn/mtbf/">14 Mean time between failures</a></li><li><a href="/learn/two-flop-sync/">15 The two-flop synchronizer</a></li></ol></li><li><h3>Module 6 · Crossing many bits</h3><p>Why a binary bus tears, why Gray code does not, and the handshake that moves a pulse between worlds.</p><ol><li><a href="/learn/why-binary-fails/">16 Why binary fails</a></li><li><a href="/learn/gray-code/">17 Gray code</a></li><li><a href="/learn/handshakes-and-pulses/">18 Handshakes and pulses</a></li></ol></li><li><h3>Module 7 · The asynchronous FIFO</h3><p>Pointers, flags and depth: the buffer that moves a stream between two clocks.</p><ol><li><a href="/learn/pointers-and-flags/">19 Pointers and flags</a></li><li><a href="/learn/depth-sizing/">20 Sizing the depth</a></li><li><a href="/learn/fifo-testbench/">21 The FIFO testbench</a></li></ol></li><li><h3>Module 8 · Constraints</h3><p>The lines that tell the analyzer what a clock is, which paths not to check, and what the pins must meet.</p><ol><li><a href="/learn/create-clock/">22 create_clock</a></li><li><a href="/learn/false-path-and-max-delay/">23 False paths and max delay</a></li><li><a href="/learn/io-timing/">24 IO timing</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>A CDC report, one crossing captured at the flip-flops, and the bitstream diff that closes the course.</p><ol><li><a href="/learn/cdc-report/">25 The CDC report</a></li><li><a href="/learn/capture-a-crossing/">26 Capture a crossing</a></li><li><a href="/learn/capstone-diff/">27 Capstone: change it and diff it</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/one-lut-up-close/index.html b/apps/website/public/learn/one-lut-up-close/index.html
index dcd5a887e4..c9246907e5 100644
--- a/apps/website/public/learn/one-lut-up-close/index.html
+++ b/apps/website/public/learn/one-lut-up-close/index.html
@@ -148,5 +148,6 @@ <h1>One LUT, up close</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The chip</h3><p>What an FPGA is, which chip we use, and how its pins meet the board.</p><ol><li><a href="/learn/fpga-is-a-grid/">01 An FPGA is a grid you configure</a></li><li><a href="/learn/which-chip/">02 Which chip is this?</a></li><li><a href="/learn/pins-and-package/">03 Pins, balls and the package</a></li></ol></li><li><h3>Module 2 · Numbers in hardware</h3><p>Bits, trits and number formats, and what arithmetic costs in logic.</p><ol><li><a href="/learn/bits-and-trits/">04 Bits, trits and GF16</a></li><li><a href="/learn/what-arithmetic-costs/">05 What arithmetic costs</a></li><li><a href="/learn/how-much-fits/">06 How much fits on the chip</a></li></ol></li><li><h3>Module 3 · Your t27 program</h3><p>Write a spec, test it, and see why compiling is not the same as being right.</p><ol><li><a href="/learn/hello-t27/">07 Hello, t27</a></li><li><a href="/learn/tests-are-the-spec/">08 Tests are part of the spec</a></li><li><a href="/learn/compiles-is-not-correct/">09 Compiles is not the same as correct</a></li></ol></li><li><h3>Module 4 · Inside t27c</h3><p>How the compiler reads a spec and what it writes, including the native t27b.</p><ol><li><a href="/learn/compiler-stages/">10 The stages of t27c</a></li><li><a href="/learn/t27c-on-a-real-machine/">11 t27c on a real machine</a></li><li><a href="/learn/seven-backends-and-t27b/">12 Seven backends, and t27b</a></li></ol></li><li><h3>Module 5 · From spec to hardware</h3><p>The Verilog t27c writes, the cells it becomes, and what one LUT does.</p><ol><li><a href="/learn/spec-to-verilog/">13 From a spec to Verilog</a></li><li><a href="/learn/count-the-cells/">14 Count the cells</a></li><li class="on" aria-current="page"><a href="/learn/one-lut-up-close/">15 One LUT, up close</a></li></ol></li><li><h3>Module 6 · Reading synthesis</h3><p>What yosys reports about your design, and which warnings matter.</p><ol><li><a href="/learn/synthesis-warnings/">16 Synthesis warnings that matter</a></li><li><a href="/learn/state-machines/">17 State machines</a></li><li><a href="/learn/where-the-luts-go/">18 Where the LUTs go</a></li></ol></li><li><h3>Module 7 · Place, route, timing</h3><p>Where the cells land on the die, and whether the clock is met.</p><ol><li><a href="/learn/placement/">19 Placement: where the cells land</a></li><li><a href="/learn/logic-depth/">20 Logic depth</a></li><li><a href="/learn/timing-slack/">21 Timing and slack</a></li></ol></li><li><h3>Module 8 · The bitstream</h3><p>How a routed design becomes the bits the chip loads, with no Vivado.</p><ol><li><a href="/learn/fasm-to-frames/">22 From FASM to frames</a></li><li><a href="/learn/a-build-receipt/">23 A build, itemised</a></li><li><a href="/learn/read-a-bitstream/">24 Read a bitstream</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>Ask the chip who it is, load the bits, and check them.</p><ol><li><a href="/learn/jtag-asks-the-chip/">25 JTAG asks the chip</a></li><li><a href="/learn/same-bits-on-the-board/">26 The same bits, on the board</a></li><li><a href="/learn/capstone/">27 Capstone: out-guess yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/one-outlier/index.html b/apps/website/public/learn/one-outlier/index.html
index dcd1441c5d..ea500ae2ef 100644
--- a/apps/website/public/learn/one-outlier/index.html
+++ b/apps/website/public/learn/one-outlier/index.html
@@ -375,5 +375,6 @@ <h1>One outlier, many zeros</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Lab: our own research</h3><p>A number format of our own, an honest scoreboard, and a model's tables multiplied on the board.</p><ol><li><a href="/learn/t27-number-format/">01 T27: a number format of our own</a></li><li><a href="/learn/an-honest-scoreboard/">02 An honest scoreboard</a></li><li><a href="/learn/a-model-on-the-board/">03 A model's tables on the board</a></li></ol></li><li><h3>Module 2 · AI numbers: the MX block</h3><p>How AI chips keep weights in a few bits: one shared scale per block, the scale byte itself, and what one outlier does to its neighbours.</p><ol><li><a href="/learn/one-scale-per-block/">04 One scale for a block of weights</a></li><li><a href="/learn/scale-byte-native/">05 The scale byte on a real machine</a></li><li class="on" aria-current="page"><a href="/learn/one-outlier/">06 One outlier, many zeros</a></li></ol></li><li><h3>Module 3 · Ternary weights</h3><p>Weights that are only minus, zero or plus a scale, the five rules a ternary alphabet must pass, and a test pass that checked nothing.</p><ol><li><a href="/learn/phi-scaled-ternary/">07 Weights of minus phi, zero and plus phi</a></li><li><a href="/learn/a-pass-that-checked-nothing/">08 A pass that checked nothing</a></li><li><a href="/learn/five-rules-for-weights/">09 Five rules for a weight alphabet</a></li></ol></li><li><h3>Module 4 · The Ternary Network Float</h3><p>A rule the compiler enforces before any test runs, and a 17-bit float whose exponent is four balanced trits.</p><ol><li><a href="/learn/blocked-is-not-failed/">10 Blocked is not failed</a></li><li><a href="/learn/four-trit-exponent/">11 A float whose exponent is four trits</a></li><li><a href="/learn/negate-twice/">12 Negate twice, get the number back</a></li></ol></li><li><h3>Module 5 · Arithmetic on signed numbers</h3><p>Multiply two signed numbers, add them when their signs differ, and do both at once in a multiply-accumulate.</p><ol><li><a href="/learn/a-sign-is-one-bit/">13 A sign is one bit</a></li><li><a href="/learn/opposite-signs-subtract/">14 Opposite signs subtract</a></li><li><a href="/learn/multiply-then-add/">15 Multiply, then add</a></li></ol></li><li><h3>Module 6 · Parts of a neuron</h3><p>A ReLU that bends at zero, a power of two for softmax, and an argmax that names the answer.</p><ol><li><a href="/learn/a-bend-at-zero/">16 A bend at zero</a></li><li><a href="/learn/two-to-the-x/">17 Two to the x</a></li><li><a href="/learn/pick-the-largest/">18 Pick the largest</a></li></ol></li><li><h3>Module 7 · Learning from a mistake</h3><p>A loss that prices a wrong guess in bits, one step that moves a weight against its gradient, and the hidden layer that XOR needs.</p><ol><li><a href="/learn/a-loss-in-bits/">19 A loss in bits</a></li><li><a href="/learn/one-step-downhill/">20 One step downhill</a></li><li><a href="/learn/xor-needs-a-bend/">21 XOR needs a bend</a></li></ol></li><li><h3>Module 8 · BitNet: ternary networks</h3><p>A threshold that squeezes a sum back to three values, one neuron that becomes a different function when its weights change, and a neuron that reads its inputs 27 trits at a time.</p><ol><li><a href="/learn/back-to-three-values/">22 Back to three values</a></li><li><a href="/learn/same-neuron-new-weights/">23 Same neuron, new weights</a></li><li><a href="/learn/a-neuron-in-chunks/">24 A neuron in chunks</a></li></ol></li><li><h3>Module 9 · The ternary MAC as a chip</h3><p>The 27-trit dot product as wires with no register, the same sum added into a register on every clock, and a small whole network to close the course.</p><ol><li><a href="/learn/a-dot-product-in-wires/">25 A dot product made of wires</a></li><li><a href="/learn/add-it-up-every-clock/">26 Add it up on every clock</a></li><li><a href="/learn/a-whole-network/">27 A whole network</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/one-scale-per-block/index.html b/apps/website/public/learn/one-scale-per-block/index.html
index ef70f4b22d..9b8b7c6e42 100644
--- a/apps/website/public/learn/one-scale-per-block/index.html
+++ b/apps/website/public/learn/one-scale-per-block/index.html
@@ -565,5 +565,6 @@ <h1>One scale for a block of weights</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Lab: our own research</h3><p>A number format of our own, an honest scoreboard, and a model's tables multiplied on the board.</p><ol><li><a href="/learn/t27-number-format/">01 T27: a number format of our own</a></li><li><a href="/learn/an-honest-scoreboard/">02 An honest scoreboard</a></li><li><a href="/learn/a-model-on-the-board/">03 A model's tables on the board</a></li></ol></li><li><h3>Module 2 · AI numbers: the MX block</h3><p>How AI chips keep weights in a few bits: one shared scale per block, the scale byte itself, and what one outlier does to its neighbours.</p><ol><li class="on" aria-current="page"><a href="/learn/one-scale-per-block/">04 One scale for a block of weights</a></li><li><a href="/learn/scale-byte-native/">05 The scale byte on a real machine</a></li><li><a href="/learn/one-outlier/">06 One outlier, many zeros</a></li></ol></li><li><h3>Module 3 · Ternary weights</h3><p>Weights that are only minus, zero or plus a scale, the five rules a ternary alphabet must pass, and a test pass that checked nothing.</p><ol><li><a href="/learn/phi-scaled-ternary/">07 Weights of minus phi, zero and plus phi</a></li><li><a href="/learn/a-pass-that-checked-nothing/">08 A pass that checked nothing</a></li><li><a href="/learn/five-rules-for-weights/">09 Five rules for a weight alphabet</a></li></ol></li><li><h3>Module 4 · The Ternary Network Float</h3><p>A rule the compiler enforces before any test runs, and a 17-bit float whose exponent is four balanced trits.</p><ol><li><a href="/learn/blocked-is-not-failed/">10 Blocked is not failed</a></li><li><a href="/learn/four-trit-exponent/">11 A float whose exponent is four trits</a></li><li><a href="/learn/negate-twice/">12 Negate twice, get the number back</a></li></ol></li><li><h3>Module 5 · Arithmetic on signed numbers</h3><p>Multiply two signed numbers, add them when their signs differ, and do both at once in a multiply-accumulate.</p><ol><li><a href="/learn/a-sign-is-one-bit/">13 A sign is one bit</a></li><li><a href="/learn/opposite-signs-subtract/">14 Opposite signs subtract</a></li><li><a href="/learn/multiply-then-add/">15 Multiply, then add</a></li></ol></li><li><h3>Module 6 · Parts of a neuron</h3><p>A ReLU that bends at zero, a power of two for softmax, and an argmax that names the answer.</p><ol><li><a href="/learn/a-bend-at-zero/">16 A bend at zero</a></li><li><a href="/learn/two-to-the-x/">17 Two to the x</a></li><li><a href="/learn/pick-the-largest/">18 Pick the largest</a></li></ol></li><li><h3>Module 7 · Learning from a mistake</h3><p>A loss that prices a wrong guess in bits, one step that moves a weight against its gradient, and the hidden layer that XOR needs.</p><ol><li><a href="/learn/a-loss-in-bits/">19 A loss in bits</a></li><li><a href="/learn/one-step-downhill/">20 One step downhill</a></li><li><a href="/learn/xor-needs-a-bend/">21 XOR needs a bend</a></li></ol></li><li><h3>Module 8 · BitNet: ternary networks</h3><p>A threshold that squeezes a sum back to three values, one neuron that becomes a different function when its weights change, and a neuron that reads its inputs 27 trits at a time.</p><ol><li><a href="/learn/back-to-three-values/">22 Back to three values</a></li><li><a href="/learn/same-neuron-new-weights/">23 Same neuron, new weights</a></li><li><a href="/learn/a-neuron-in-chunks/">24 A neuron in chunks</a></li></ol></li><li><h3>Module 9 · The ternary MAC as a chip</h3><p>The 27-trit dot product as wires with no register, the same sum added into a register on every clock, and a small whole network to close the course.</p><ol><li><a href="/learn/a-dot-product-in-wires/">25 A dot product made of wires</a></li><li><a href="/learn/add-it-up-every-clock/">26 Add it up on every clock</a></li><li><a href="/learn/a-whole-network/">27 A whole network</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/one-step-downhill/index.html b/apps/website/public/learn/one-step-downhill/index.html
index e8f39b62c8..386f6fada6 100644
--- a/apps/website/public/learn/one-step-downhill/index.html
+++ b/apps/website/public/learn/one-step-downhill/index.html
@@ -198,5 +198,6 @@ <h1>One step downhill</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Lab: our own research</h3><p>A number format of our own, an honest scoreboard, and a model's tables multiplied on the board.</p><ol><li><a href="/learn/t27-number-format/">01 T27: a number format of our own</a></li><li><a href="/learn/an-honest-scoreboard/">02 An honest scoreboard</a></li><li><a href="/learn/a-model-on-the-board/">03 A model's tables on the board</a></li></ol></li><li><h3>Module 2 · AI numbers: the MX block</h3><p>How AI chips keep weights in a few bits: one shared scale per block, the scale byte itself, and what one outlier does to its neighbours.</p><ol><li><a href="/learn/one-scale-per-block/">04 One scale for a block of weights</a></li><li><a href="/learn/scale-byte-native/">05 The scale byte on a real machine</a></li><li><a href="/learn/one-outlier/">06 One outlier, many zeros</a></li></ol></li><li><h3>Module 3 · Ternary weights</h3><p>Weights that are only minus, zero or plus a scale, the five rules a ternary alphabet must pass, and a test pass that checked nothing.</p><ol><li><a href="/learn/phi-scaled-ternary/">07 Weights of minus phi, zero and plus phi</a></li><li><a href="/learn/a-pass-that-checked-nothing/">08 A pass that checked nothing</a></li><li><a href="/learn/five-rules-for-weights/">09 Five rules for a weight alphabet</a></li></ol></li><li><h3>Module 4 · The Ternary Network Float</h3><p>A rule the compiler enforces before any test runs, and a 17-bit float whose exponent is four balanced trits.</p><ol><li><a href="/learn/blocked-is-not-failed/">10 Blocked is not failed</a></li><li><a href="/learn/four-trit-exponent/">11 A float whose exponent is four trits</a></li><li><a href="/learn/negate-twice/">12 Negate twice, get the number back</a></li></ol></li><li><h3>Module 5 · Arithmetic on signed numbers</h3><p>Multiply two signed numbers, add them when their signs differ, and do both at once in a multiply-accumulate.</p><ol><li><a href="/learn/a-sign-is-one-bit/">13 A sign is one bit</a></li><li><a href="/learn/opposite-signs-subtract/">14 Opposite signs subtract</a></li><li><a href="/learn/multiply-then-add/">15 Multiply, then add</a></li></ol></li><li><h3>Module 6 · Parts of a neuron</h3><p>A ReLU that bends at zero, a power of two for softmax, and an argmax that names the answer.</p><ol><li><a href="/learn/a-bend-at-zero/">16 A bend at zero</a></li><li><a href="/learn/two-to-the-x/">17 Two to the x</a></li><li><a href="/learn/pick-the-largest/">18 Pick the largest</a></li></ol></li><li><h3>Module 7 · Learning from a mistake</h3><p>A loss that prices a wrong guess in bits, one step that moves a weight against its gradient, and the hidden layer that XOR needs.</p><ol><li><a href="/learn/a-loss-in-bits/">19 A loss in bits</a></li><li class="on" aria-current="page"><a href="/learn/one-step-downhill/">20 One step downhill</a></li><li><a href="/learn/xor-needs-a-bend/">21 XOR needs a bend</a></li></ol></li><li><h3>Module 8 · BitNet: ternary networks</h3><p>A threshold that squeezes a sum back to three values, one neuron that becomes a different function when its weights change, and a neuron that reads its inputs 27 trits at a time.</p><ol><li><a href="/learn/back-to-three-values/">22 Back to three values</a></li><li><a href="/learn/same-neuron-new-weights/">23 Same neuron, new weights</a></li><li><a href="/learn/a-neuron-in-chunks/">24 A neuron in chunks</a></li></ol></li><li><h3>Module 9 · The ternary MAC as a chip</h3><p>The 27-trit dot product as wires with no register, the same sum added into a register on every clock, and a small whole network to close the course.</p><ol><li><a href="/learn/a-dot-product-in-wires/">25 A dot product made of wires</a></li><li><a href="/learn/add-it-up-every-clock/">26 Add it up on every clock</a></li><li><a href="/learn/a-whole-network/">27 A whole network</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/one-ternary-neuron/index.html b/apps/website/public/learn/one-ternary-neuron/index.html
index f490f87f77..15310ba67b 100644
--- a/apps/website/public/learn/one-ternary-neuron/index.html
+++ b/apps/website/public/learn/one-ternary-neuron/index.html
@@ -769,5 +769,6 @@ <h1>One ternary neuron</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Three values</h3><p>Why three, how balanced ternary writes every number without a sign, and what flipping and cutting trits do.</p><ol><li><a href="/learn/why-three/">01 Why three</a></li><li><a href="/learn/balanced-ternary/">02 Balanced ternary</a></li><li><a href="/learn/flip-and-cut/">03 Flip to negate, cut to round</a></li></ol></li><li><h3>Module 2 · Logic with unknown</h3><p>Kleene's three-valued gates, two gates binary has no twin for, and addition as a pair of tables.</p><ol><li><a href="/learn/kleene-logic/">04 Three-valued logic</a></li><li><a href="/learn/consensus-and-any/">05 Consensus and any</a></li><li><a href="/learn/sum-and-carry/">06 Sum and carry</a></li></ol></li><li><h3>Module 3 · Adding trits</h3><p>The half adder, the full adder and a carry that ripples left, one place at a time.</p><ol><li><a href="/learn/half-adder/">07 The half adder</a></li><li><a href="/learn/full-adder/">08 The full adder</a></li><li><a href="/learn/ripple-carry/">09 Ripple carry</a></li></ol></li><li><h3>Module 4 · Multiplying without a multiplier</h3><p>Copy, drop or flip: a product by one trit, long multiplication, and a MAC that only adds.</p><ol><li><a href="/learn/times-a-trit/">10 Times one trit</a></li><li><a href="/learn/long-multiplication/">11 Long multiplication</a></li><li><a href="/learn/ternary-mac/">12 A ternary MAC</a></li></ol></li><li><h3>Module 5 · Trits in binary memory</h3><p>Two bits per trit, five trits per byte, and the bits a 27-trit word needs.</p><ol><li><a href="/learn/two-bits-per-trit/">13 Two bits per trit</a></li><li><a href="/learn/five-trits-a-byte/">14 Five trits in a byte</a></li><li><a href="/learn/a-word-of-27/">15 A word of 27 trits</a></li></ol></li><li><h3>Module 6 · The TRI-27 instruction word</h3><p>The 32-bit word the Trinity emulator decodes: its fields, its 47 opcodes and its 15-bit immediate.</p><ol><li><a href="/learn/the-instruction-word/">16 The instruction word</a></li><li><a href="/learn/the-opcode-map/">17 The opcode map</a></li><li><a href="/learn/the-immediate/">18 The immediate</a></li></ol></li><li><h3>Module 7 · A ternary machine</h3><p>An ALU built from this course's adders, a three-way jump, and a program you can step.</p><ol><li><a href="/learn/a-ternary-alu/">19 A ternary ALU</a></li><li><a href="/learn/three-way-jump/">20 A three-way jump</a></li><li><a href="/learn/run-a-program/">21 Run a program</a></li></ol></li><li><h3>Module 8 · Three answers</h3><p>Compare at the highest differing trit, find a number in thirds, sort with three-way compares.</p><ol><li><a href="/learn/compare-at-the-top/">22 Compare at the top</a></li><li><a href="/learn/questions-in-thirds/">23 Questions in thirds</a></li><li><a href="/learn/sort-by-three/">24 Sort with three-way compares</a></li></ol></li><li><h3>Module 9 · Ternary neurons</h3><p>Weights of -1, 0, +1: one neuron, a detector and a small layer, with no multiplier anywhere.</p><ol><li class="on" aria-current="page"><a href="/learn/one-ternary-neuron/">25 One ternary neuron</a></li><li><a href="/learn/a-bar-detector/">26 A bar detector</a></li><li><a href="/learn/a-tiny-layer/">27 A tiny layer</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/opposite-signs-subtract/index.html b/apps/website/public/learn/opposite-signs-subtract/index.html
index b05f6f8e01..9ab4240d9b 100644
--- a/apps/website/public/learn/opposite-signs-subtract/index.html
+++ b/apps/website/public/learn/opposite-signs-subtract/index.html
@@ -191,5 +191,6 @@ <h1>Opposite signs subtract</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Lab: our own research</h3><p>A number format of our own, an honest scoreboard, and a model's tables multiplied on the board.</p><ol><li><a href="/learn/t27-number-format/">01 T27: a number format of our own</a></li><li><a href="/learn/an-honest-scoreboard/">02 An honest scoreboard</a></li><li><a href="/learn/a-model-on-the-board/">03 A model's tables on the board</a></li></ol></li><li><h3>Module 2 · AI numbers: the MX block</h3><p>How AI chips keep weights in a few bits: one shared scale per block, the scale byte itself, and what one outlier does to its neighbours.</p><ol><li><a href="/learn/one-scale-per-block/">04 One scale for a block of weights</a></li><li><a href="/learn/scale-byte-native/">05 The scale byte on a real machine</a></li><li><a href="/learn/one-outlier/">06 One outlier, many zeros</a></li></ol></li><li><h3>Module 3 · Ternary weights</h3><p>Weights that are only minus, zero or plus a scale, the five rules a ternary alphabet must pass, and a test pass that checked nothing.</p><ol><li><a href="/learn/phi-scaled-ternary/">07 Weights of minus phi, zero and plus phi</a></li><li><a href="/learn/a-pass-that-checked-nothing/">08 A pass that checked nothing</a></li><li><a href="/learn/five-rules-for-weights/">09 Five rules for a weight alphabet</a></li></ol></li><li><h3>Module 4 · The Ternary Network Float</h3><p>A rule the compiler enforces before any test runs, and a 17-bit float whose exponent is four balanced trits.</p><ol><li><a href="/learn/blocked-is-not-failed/">10 Blocked is not failed</a></li><li><a href="/learn/four-trit-exponent/">11 A float whose exponent is four trits</a></li><li><a href="/learn/negate-twice/">12 Negate twice, get the number back</a></li></ol></li><li><h3>Module 5 · Arithmetic on signed numbers</h3><p>Multiply two signed numbers, add them when their signs differ, and do both at once in a multiply-accumulate.</p><ol><li><a href="/learn/a-sign-is-one-bit/">13 A sign is one bit</a></li><li class="on" aria-current="page"><a href="/learn/opposite-signs-subtract/">14 Opposite signs subtract</a></li><li><a href="/learn/multiply-then-add/">15 Multiply, then add</a></li></ol></li><li><h3>Module 6 · Parts of a neuron</h3><p>A ReLU that bends at zero, a power of two for softmax, and an argmax that names the answer.</p><ol><li><a href="/learn/a-bend-at-zero/">16 A bend at zero</a></li><li><a href="/learn/two-to-the-x/">17 Two to the x</a></li><li><a href="/learn/pick-the-largest/">18 Pick the largest</a></li></ol></li><li><h3>Module 7 · Learning from a mistake</h3><p>A loss that prices a wrong guess in bits, one step that moves a weight against its gradient, and the hidden layer that XOR needs.</p><ol><li><a href="/learn/a-loss-in-bits/">19 A loss in bits</a></li><li><a href="/learn/one-step-downhill/">20 One step downhill</a></li><li><a href="/learn/xor-needs-a-bend/">21 XOR needs a bend</a></li></ol></li><li><h3>Module 8 · BitNet: ternary networks</h3><p>A threshold that squeezes a sum back to three values, one neuron that becomes a different function when its weights change, and a neuron that reads its inputs 27 trits at a time.</p><ol><li><a href="/learn/back-to-three-values/">22 Back to three values</a></li><li><a href="/learn/same-neuron-new-weights/">23 Same neuron, new weights</a></li><li><a href="/learn/a-neuron-in-chunks/">24 A neuron in chunks</a></li></ol></li><li><h3>Module 9 · The ternary MAC as a chip</h3><p>The 27-trit dot product as wires with no register, the same sum added into a register on every clock, and a small whole network to close the course.</p><ol><li><a href="/learn/a-dot-product-in-wires/">25 A dot product made of wires</a></li><li><a href="/learn/add-it-up-every-clock/">26 Add it up on every clock</a></li><li><a href="/learn/a-whole-network/">27 A whole network</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/period-and-jitter/index.html b/apps/website/public/learn/period-and-jitter/index.html
index 1dc5003d2a..9708d2bd67 100644
--- a/apps/website/public/learn/period-and-jitter/index.html
+++ b/apps/website/public/learn/period-and-jitter/index.html
@@ -107,5 +107,6 @@ <h1>A period, an edge, and jitter</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a clock is</h3><p>One edge, one world: what shares a clock edge shares a world, the period and the jitter of a real edge, and where the clock enters a board.</p><ol><li><a href="/learn/one-edge-one-world/">01 One edge, one world</a></li><li class="on" aria-current="page"><a href="/learn/period-and-jitter/">02 A period, an edge, and jitter</a></li><li><a href="/learn/where-the-clock-enters/">03 Where the clock enters</a></li></ol></li><li><h3>Module 2 · Clock trees</h3><p>Skew and insertion delay, the global buffer network, and the trap of gating a clock with logic.</p><ol><li><a href="/learn/skew-and-insertion/">04 Skew and insertion delay</a></li><li><a href="/learn/global-buffers/">05 The global buffer</a></li><li><a href="/learn/gated-clock-trap/">06 The gated-clock trap</a></li></ol></li><li><h3>Module 3 · PLL and MMCM</h3><p>Multiply and divide one clock into another, move its phase in steps of the VCO, and which clocks the analyzer treats as related.</p><ol><li><a href="/learn/multiply-and-divide/">07 Multiply and divide</a></li><li><a href="/learn/phase-and-lock/">08 Phase and lock</a></li><li><a href="/learn/related-or-not/">09 Related or not</a></li></ol></li><li><h3>Module 4 · Resets</h3><p>Assert asynchronously, release synchronously: the three reset kinds, the release pipe, and the tree a reset grows.</p><ol><li><a href="/learn/sync-or-async/">10 Sync or async</a></li><li><a href="/learn/reset-release/">11 The release pipe</a></li><li><a href="/learn/reset-fanout/">12 Reset fanout</a></li></ol></li><li><h3>Module 5 · Metastability</h3><p>The setup-hold window, the mean time between failures in integer arithmetic, and the two flops that fix it.</p><ol><li><a href="/learn/setup-hold-window/">13 The setup-hold window</a></li><li><a href="/learn/mtbf/">14 Mean time between failures</a></li><li><a href="/learn/two-flop-sync/">15 The two-flop synchronizer</a></li></ol></li><li><h3>Module 6 · Crossing many bits</h3><p>Why a binary bus tears, why Gray code does not, and the handshake that moves a pulse between worlds.</p><ol><li><a href="/learn/why-binary-fails/">16 Why binary fails</a></li><li><a href="/learn/gray-code/">17 Gray code</a></li><li><a href="/learn/handshakes-and-pulses/">18 Handshakes and pulses</a></li></ol></li><li><h3>Module 7 · The asynchronous FIFO</h3><p>Pointers, flags and depth: the buffer that moves a stream between two clocks.</p><ol><li><a href="/learn/pointers-and-flags/">19 Pointers and flags</a></li><li><a href="/learn/depth-sizing/">20 Sizing the depth</a></li><li><a href="/learn/fifo-testbench/">21 The FIFO testbench</a></li></ol></li><li><h3>Module 8 · Constraints</h3><p>The lines that tell the analyzer what a clock is, which paths not to check, and what the pins must meet.</p><ol><li><a href="/learn/create-clock/">22 create_clock</a></li><li><a href="/learn/false-path-and-max-delay/">23 False paths and max delay</a></li><li><a href="/learn/io-timing/">24 IO timing</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>A CDC report, one crossing captured at the flip-flops, and the bitstream diff that closes the course.</p><ol><li><a href="/learn/cdc-report/">25 The CDC report</a></li><li><a href="/learn/capture-a-crossing/">26 Capture a crossing</a></li><li><a href="/learn/capstone-diff/">27 Capstone: change it and diff it</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/phase-and-lock/index.html b/apps/website/public/learn/phase-and-lock/index.html
index 795bb4103c..f8ba7dcd2f 100644
--- a/apps/website/public/learn/phase-and-lock/index.html
+++ b/apps/website/public/learn/phase-and-lock/index.html
@@ -341,5 +341,6 @@ <h1>Phase and lock</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a clock is</h3><p>One edge, one world: what shares a clock edge shares a world, the period and the jitter of a real edge, and where the clock enters a board.</p><ol><li><a href="/learn/one-edge-one-world/">01 One edge, one world</a></li><li><a href="/learn/period-and-jitter/">02 A period, an edge, and jitter</a></li><li><a href="/learn/where-the-clock-enters/">03 Where the clock enters</a></li></ol></li><li><h3>Module 2 · Clock trees</h3><p>Skew and insertion delay, the global buffer network, and the trap of gating a clock with logic.</p><ol><li><a href="/learn/skew-and-insertion/">04 Skew and insertion delay</a></li><li><a href="/learn/global-buffers/">05 The global buffer</a></li><li><a href="/learn/gated-clock-trap/">06 The gated-clock trap</a></li></ol></li><li><h3>Module 3 · PLL and MMCM</h3><p>Multiply and divide one clock into another, move its phase in steps of the VCO, and which clocks the analyzer treats as related.</p><ol><li><a href="/learn/multiply-and-divide/">07 Multiply and divide</a></li><li class="on" aria-current="page"><a href="/learn/phase-and-lock/">08 Phase and lock</a></li><li><a href="/learn/related-or-not/">09 Related or not</a></li></ol></li><li><h3>Module 4 · Resets</h3><p>Assert asynchronously, release synchronously: the three reset kinds, the release pipe, and the tree a reset grows.</p><ol><li><a href="/learn/sync-or-async/">10 Sync or async</a></li><li><a href="/learn/reset-release/">11 The release pipe</a></li><li><a href="/learn/reset-fanout/">12 Reset fanout</a></li></ol></li><li><h3>Module 5 · Metastability</h3><p>The setup-hold window, the mean time between failures in integer arithmetic, and the two flops that fix it.</p><ol><li><a href="/learn/setup-hold-window/">13 The setup-hold window</a></li><li><a href="/learn/mtbf/">14 Mean time between failures</a></li><li><a href="/learn/two-flop-sync/">15 The two-flop synchronizer</a></li></ol></li><li><h3>Module 6 · Crossing many bits</h3><p>Why a binary bus tears, why Gray code does not, and the handshake that moves a pulse between worlds.</p><ol><li><a href="/learn/why-binary-fails/">16 Why binary fails</a></li><li><a href="/learn/gray-code/">17 Gray code</a></li><li><a href="/learn/handshakes-and-pulses/">18 Handshakes and pulses</a></li></ol></li><li><h3>Module 7 · The asynchronous FIFO</h3><p>Pointers, flags and depth: the buffer that moves a stream between two clocks.</p><ol><li><a href="/learn/pointers-and-flags/">19 Pointers and flags</a></li><li><a href="/learn/depth-sizing/">20 Sizing the depth</a></li><li><a href="/learn/fifo-testbench/">21 The FIFO testbench</a></li></ol></li><li><h3>Module 8 · Constraints</h3><p>The lines that tell the analyzer what a clock is, which paths not to check, and what the pins must meet.</p><ol><li><a href="/learn/create-clock/">22 create_clock</a></li><li><a href="/learn/false-path-and-max-delay/">23 False paths and max delay</a></li><li><a href="/learn/io-timing/">24 IO timing</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>A CDC report, one crossing captured at the flip-flops, and the bitstream diff that closes the course.</p><ol><li><a href="/learn/cdc-report/">25 The CDC report</a></li><li><a href="/learn/capture-a-crossing/">26 Capture a crossing</a></li><li><a href="/learn/capstone-diff/">27 Capstone: change it and diff it</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/phi-as-a-ratio/index.html b/apps/website/public/learn/phi-as-a-ratio/index.html
index 47c4688e79..ce29f78fde 100644
--- a/apps/website/public/learn/phi-as-a-ratio/index.html
+++ b/apps/website/public/learn/phi-as-a-ratio/index.html
@@ -828,5 +828,6 @@ <h1>Phi as a ratio</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The rule and its numbers</h3><p>One rule splits every width, the ratio it aims at, and the Lucas numbers behind the 3.</p><ol><li><a href="/learn/a-float-cut-by-phi/">01 A float cut by phi</a></li><li class="on" aria-current="page"><a href="/learn/phi-as-a-ratio/">02 Phi as a ratio</a></li><li><a href="/learn/lucas-numbers-stay-whole/">03 Lucas numbers stay whole</a></li></ol></li><li><h3>Module 2 · Why phi, why three</h3><p>Why the split is phi, why base three, and how a spec checks GF16 keeps phi.</p><ol><li><a href="/learn/why-the-split-is-phi/">04 Why the split is phi</a></li><li><a href="/learn/why-base-three/">05 Why base three</a></li><li><a href="/learn/phi-in-sixteen-bits/">06 Phi in sixteen bits</a></li></ol></li><li><h3>Module 3 · The small rungs: GF4 to GF8</h3><p>GF4, GF6 and GF8, the fewest bits, where rounding to whole bits costs the most.</p><ol><li><a href="/learn/gf4-four-bits/">07 GF4: four bits</a></li><li><a href="/learn/gf6-six-bits/">08 GF6: six bits</a></li><li><a href="/learn/gf8-one-byte/">09 GF8: one byte</a></li></ol></li><li><h3>Module 4 · Ten to fourteen bits</h3><p>GF10, GF12 and GF14, and how the distance from 1 / phi moves as the word grows.</p><ol><li><a href="/learn/gf10-ten-bits/">10 GF10: ten bits</a></li><li><a href="/learn/gf12-twelve-bits/">11 GF12: twelve bits</a></li><li><a href="/learn/gf14-fourteen-bits/">12 GF14: fourteen bits</a></li></ol></li><li><h3>Module 5 · GF16 at work</h3><p>The primary 16-bit format, a two-term dot product in GF-T16, then GF20 and GF24.</p><ol><li><a href="/learn/gf16-the-primary/">13 GF16: the primary format</a></li><li><a href="/learn/two-products-one-sum/">14 Two products, one sum</a></li><li><a href="/learn/gf24-twenty-four-bits/">15 GF20 and GF24</a></li></ol></li><li><h3>Module 6 · GF32 to GF64</h3><p>GF32 beside IEEE single, GF48 with no IEEE twin, GF64 beside IEEE double.</p><ol><li><a href="/learn/gf32-a-single/">16 GF32: a single</a></li><li><a href="/learn/gf48-forty-eight-bits/">17 GF48: forty-eight bits</a></li><li><a href="/learn/gf64-a-double/">18 GF64: a double</a></li></ol></li><li><h3>Module 7 · GF96 to GF256</h3><p>GF96, GF128 and GF256, where the specs hold the layout with invariants.</p><ol><li><a href="/learn/gf96-ninety-six-bits/">19 GF96: ninety-six bits</a></li><li><a href="/learn/gf128-a-quad/">20 GF128: a quad</a></li><li><a href="/learn/gf256-two-five-six/">21 GF256: two hundred fifty-six bits</a></li></ol></li><li><h3>Module 8 · The widest rungs, then trits</h3><p>GF512 and GF1024, the two widest rungs, then GF-T8, where the exponent moves to trits.</p><ol><li><a href="/learn/gf512-five-twelve/">22 GF512: five hundred twelve bits</a></li><li><a href="/learn/gf1024-one-kilobit/">23 GF1024: one kilobit</a></li><li><a href="/learn/gft8-exponent-in-trits/">24 GF-T8: an exponent in trits</a></li></ol></li><li><h3>Module 9 · More trits, then the decode</h3><p>GF-T16 and GF-T32, then why fixed fields decode in parallel and a posit does not.</p><ol><li><a href="/learn/gft16-six-trits/">25 GF-T16: six trits</a></li><li><a href="/learn/gft32-twelve-trits/">26 GF-T32: twelve trits</a></li><li><a href="/learn/decode-in-one-step/">27 Decode in one step</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/phi-in-sixteen-bits/index.html b/apps/website/public/learn/phi-in-sixteen-bits/index.html
index 707587addc..758f9f1744 100644
--- a/apps/website/public/learn/phi-in-sixteen-bits/index.html
+++ b/apps/website/public/learn/phi-in-sixteen-bits/index.html
@@ -177,5 +177,6 @@ <h1>Phi in sixteen bits</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The rule and its numbers</h3><p>One rule splits every width, the ratio it aims at, and the Lucas numbers behind the 3.</p><ol><li><a href="/learn/a-float-cut-by-phi/">01 A float cut by phi</a></li><li><a href="/learn/phi-as-a-ratio/">02 Phi as a ratio</a></li><li><a href="/learn/lucas-numbers-stay-whole/">03 Lucas numbers stay whole</a></li></ol></li><li><h3>Module 2 · Why phi, why three</h3><p>Why the split is phi, why base three, and how a spec checks GF16 keeps phi.</p><ol><li><a href="/learn/why-the-split-is-phi/">04 Why the split is phi</a></li><li><a href="/learn/why-base-three/">05 Why base three</a></li><li class="on" aria-current="page"><a href="/learn/phi-in-sixteen-bits/">06 Phi in sixteen bits</a></li></ol></li><li><h3>Module 3 · The small rungs: GF4 to GF8</h3><p>GF4, GF6 and GF8, the fewest bits, where rounding to whole bits costs the most.</p><ol><li><a href="/learn/gf4-four-bits/">07 GF4: four bits</a></li><li><a href="/learn/gf6-six-bits/">08 GF6: six bits</a></li><li><a href="/learn/gf8-one-byte/">09 GF8: one byte</a></li></ol></li><li><h3>Module 4 · Ten to fourteen bits</h3><p>GF10, GF12 and GF14, and how the distance from 1 / phi moves as the word grows.</p><ol><li><a href="/learn/gf10-ten-bits/">10 GF10: ten bits</a></li><li><a href="/learn/gf12-twelve-bits/">11 GF12: twelve bits</a></li><li><a href="/learn/gf14-fourteen-bits/">12 GF14: fourteen bits</a></li></ol></li><li><h3>Module 5 · GF16 at work</h3><p>The primary 16-bit format, a two-term dot product in GF-T16, then GF20 and GF24.</p><ol><li><a href="/learn/gf16-the-primary/">13 GF16: the primary format</a></li><li><a href="/learn/two-products-one-sum/">14 Two products, one sum</a></li><li><a href="/learn/gf24-twenty-four-bits/">15 GF20 and GF24</a></li></ol></li><li><h3>Module 6 · GF32 to GF64</h3><p>GF32 beside IEEE single, GF48 with no IEEE twin, GF64 beside IEEE double.</p><ol><li><a href="/learn/gf32-a-single/">16 GF32: a single</a></li><li><a href="/learn/gf48-forty-eight-bits/">17 GF48: forty-eight bits</a></li><li><a href="/learn/gf64-a-double/">18 GF64: a double</a></li></ol></li><li><h3>Module 7 · GF96 to GF256</h3><p>GF96, GF128 and GF256, where the specs hold the layout with invariants.</p><ol><li><a href="/learn/gf96-ninety-six-bits/">19 GF96: ninety-six bits</a></li><li><a href="/learn/gf128-a-quad/">20 GF128: a quad</a></li><li><a href="/learn/gf256-two-five-six/">21 GF256: two hundred fifty-six bits</a></li></ol></li><li><h3>Module 8 · The widest rungs, then trits</h3><p>GF512 and GF1024, the two widest rungs, then GF-T8, where the exponent moves to trits.</p><ol><li><a href="/learn/gf512-five-twelve/">22 GF512: five hundred twelve bits</a></li><li><a href="/learn/gf1024-one-kilobit/">23 GF1024: one kilobit</a></li><li><a href="/learn/gft8-exponent-in-trits/">24 GF-T8: an exponent in trits</a></li></ol></li><li><h3>Module 9 · More trits, then the decode</h3><p>GF-T16 and GF-T32, then why fixed fields decode in parallel and a posit does not.</p><ol><li><a href="/learn/gft16-six-trits/">25 GF-T16: six trits</a></li><li><a href="/learn/gft32-twelve-trits/">26 GF-T32: twelve trits</a></li><li><a href="/learn/decode-in-one-step/">27 Decode in one step</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/phi-scaled-ternary/index.html b/apps/website/public/learn/phi-scaled-ternary/index.html
index 73f32aa17b..ebcee147a7 100644
--- a/apps/website/public/learn/phi-scaled-ternary/index.html
+++ b/apps/website/public/learn/phi-scaled-ternary/index.html
@@ -125,5 +125,6 @@ <h1>Weights of minus phi, zero and plus phi</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Lab: our own research</h3><p>A number format of our own, an honest scoreboard, and a model's tables multiplied on the board.</p><ol><li><a href="/learn/t27-number-format/">01 T27: a number format of our own</a></li><li><a href="/learn/an-honest-scoreboard/">02 An honest scoreboard</a></li><li><a href="/learn/a-model-on-the-board/">03 A model's tables on the board</a></li></ol></li><li><h3>Module 2 · AI numbers: the MX block</h3><p>How AI chips keep weights in a few bits: one shared scale per block, the scale byte itself, and what one outlier does to its neighbours.</p><ol><li><a href="/learn/one-scale-per-block/">04 One scale for a block of weights</a></li><li><a href="/learn/scale-byte-native/">05 The scale byte on a real machine</a></li><li><a href="/learn/one-outlier/">06 One outlier, many zeros</a></li></ol></li><li><h3>Module 3 · Ternary weights</h3><p>Weights that are only minus, zero or plus a scale, the five rules a ternary alphabet must pass, and a test pass that checked nothing.</p><ol><li class="on" aria-current="page"><a href="/learn/phi-scaled-ternary/">07 Weights of minus phi, zero and plus phi</a></li><li><a href="/learn/a-pass-that-checked-nothing/">08 A pass that checked nothing</a></li><li><a href="/learn/five-rules-for-weights/">09 Five rules for a weight alphabet</a></li></ol></li><li><h3>Module 4 · The Ternary Network Float</h3><p>A rule the compiler enforces before any test runs, and a 17-bit float whose exponent is four balanced trits.</p><ol><li><a href="/learn/blocked-is-not-failed/">10 Blocked is not failed</a></li><li><a href="/learn/four-trit-exponent/">11 A float whose exponent is four trits</a></li><li><a href="/learn/negate-twice/">12 Negate twice, get the number back</a></li></ol></li><li><h3>Module 5 · Arithmetic on signed numbers</h3><p>Multiply two signed numbers, add them when their signs differ, and do both at once in a multiply-accumulate.</p><ol><li><a href="/learn/a-sign-is-one-bit/">13 A sign is one bit</a></li><li><a href="/learn/opposite-signs-subtract/">14 Opposite signs subtract</a></li><li><a href="/learn/multiply-then-add/">15 Multiply, then add</a></li></ol></li><li><h3>Module 6 · Parts of a neuron</h3><p>A ReLU that bends at zero, a power of two for softmax, and an argmax that names the answer.</p><ol><li><a href="/learn/a-bend-at-zero/">16 A bend at zero</a></li><li><a href="/learn/two-to-the-x/">17 Two to the x</a></li><li><a href="/learn/pick-the-largest/">18 Pick the largest</a></li></ol></li><li><h3>Module 7 · Learning from a mistake</h3><p>A loss that prices a wrong guess in bits, one step that moves a weight against its gradient, and the hidden layer that XOR needs.</p><ol><li><a href="/learn/a-loss-in-bits/">19 A loss in bits</a></li><li><a href="/learn/one-step-downhill/">20 One step downhill</a></li><li><a href="/learn/xor-needs-a-bend/">21 XOR needs a bend</a></li></ol></li><li><h3>Module 8 · BitNet: ternary networks</h3><p>A threshold that squeezes a sum back to three values, one neuron that becomes a different function when its weights change, and a neuron that reads its inputs 27 trits at a time.</p><ol><li><a href="/learn/back-to-three-values/">22 Back to three values</a></li><li><a href="/learn/same-neuron-new-weights/">23 Same neuron, new weights</a></li><li><a href="/learn/a-neuron-in-chunks/">24 A neuron in chunks</a></li></ol></li><li><h3>Module 9 · The ternary MAC as a chip</h3><p>The 27-trit dot product as wires with no register, the same sum added into a register on every clock, and a small whole network to close the course.</p><ol><li><a href="/learn/a-dot-product-in-wires/">25 A dot product made of wires</a></li><li><a href="/learn/add-it-up-every-clock/">26 Add it up on every clock</a></li><li><a href="/learn/a-whole-network/">27 A whole network</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/pick-the-largest/index.html b/apps/website/public/learn/pick-the-largest/index.html
index baaa5505cd..85ec182be2 100644
--- a/apps/website/public/learn/pick-the-largest/index.html
+++ b/apps/website/public/learn/pick-the-largest/index.html
@@ -121,5 +121,6 @@ <h1>Pick the largest</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Lab: our own research</h3><p>A number format of our own, an honest scoreboard, and a model's tables multiplied on the board.</p><ol><li><a href="/learn/t27-number-format/">01 T27: a number format of our own</a></li><li><a href="/learn/an-honest-scoreboard/">02 An honest scoreboard</a></li><li><a href="/learn/a-model-on-the-board/">03 A model's tables on the board</a></li></ol></li><li><h3>Module 2 · AI numbers: the MX block</h3><p>How AI chips keep weights in a few bits: one shared scale per block, the scale byte itself, and what one outlier does to its neighbours.</p><ol><li><a href="/learn/one-scale-per-block/">04 One scale for a block of weights</a></li><li><a href="/learn/scale-byte-native/">05 The scale byte on a real machine</a></li><li><a href="/learn/one-outlier/">06 One outlier, many zeros</a></li></ol></li><li><h3>Module 3 · Ternary weights</h3><p>Weights that are only minus, zero or plus a scale, the five rules a ternary alphabet must pass, and a test pass that checked nothing.</p><ol><li><a href="/learn/phi-scaled-ternary/">07 Weights of minus phi, zero and plus phi</a></li><li><a href="/learn/a-pass-that-checked-nothing/">08 A pass that checked nothing</a></li><li><a href="/learn/five-rules-for-weights/">09 Five rules for a weight alphabet</a></li></ol></li><li><h3>Module 4 · The Ternary Network Float</h3><p>A rule the compiler enforces before any test runs, and a 17-bit float whose exponent is four balanced trits.</p><ol><li><a href="/learn/blocked-is-not-failed/">10 Blocked is not failed</a></li><li><a href="/learn/four-trit-exponent/">11 A float whose exponent is four trits</a></li><li><a href="/learn/negate-twice/">12 Negate twice, get the number back</a></li></ol></li><li><h3>Module 5 · Arithmetic on signed numbers</h3><p>Multiply two signed numbers, add them when their signs differ, and do both at once in a multiply-accumulate.</p><ol><li><a href="/learn/a-sign-is-one-bit/">13 A sign is one bit</a></li><li><a href="/learn/opposite-signs-subtract/">14 Opposite signs subtract</a></li><li><a href="/learn/multiply-then-add/">15 Multiply, then add</a></li></ol></li><li><h3>Module 6 · Parts of a neuron</h3><p>A ReLU that bends at zero, a power of two for softmax, and an argmax that names the answer.</p><ol><li><a href="/learn/a-bend-at-zero/">16 A bend at zero</a></li><li><a href="/learn/two-to-the-x/">17 Two to the x</a></li><li class="on" aria-current="page"><a href="/learn/pick-the-largest/">18 Pick the largest</a></li></ol></li><li><h3>Module 7 · Learning from a mistake</h3><p>A loss that prices a wrong guess in bits, one step that moves a weight against its gradient, and the hidden layer that XOR needs.</p><ol><li><a href="/learn/a-loss-in-bits/">19 A loss in bits</a></li><li><a href="/learn/one-step-downhill/">20 One step downhill</a></li><li><a href="/learn/xor-needs-a-bend/">21 XOR needs a bend</a></li></ol></li><li><h3>Module 8 · BitNet: ternary networks</h3><p>A threshold that squeezes a sum back to three values, one neuron that becomes a different function when its weights change, and a neuron that reads its inputs 27 trits at a time.</p><ol><li><a href="/learn/back-to-three-values/">22 Back to three values</a></li><li><a href="/learn/same-neuron-new-weights/">23 Same neuron, new weights</a></li><li><a href="/learn/a-neuron-in-chunks/">24 A neuron in chunks</a></li></ol></li><li><h3>Module 9 · The ternary MAC as a chip</h3><p>The 27-trit dot product as wires with no register, the same sum added into a register on every clock, and a small whole network to close the course.</p><ol><li><a href="/learn/a-dot-product-in-wires/">25 A dot product made of wires</a></li><li><a href="/learn/add-it-up-every-clock/">26 Add it up on every clock</a></li><li><a href="/learn/a-whole-network/">27 A whole network</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/pins-and-package/index.html b/apps/website/public/learn/pins-and-package/index.html
index 5eb6cc6ca0..9e43949ddd 100644
--- a/apps/website/public/learn/pins-and-package/index.html
+++ b/apps/website/public/learn/pins-and-package/index.html
@@ -462,5 +462,6 @@ <h1>Pins, balls and the package</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The chip</h3><p>What an FPGA is, which chip we use, and how its pins meet the board.</p><ol><li><a href="/learn/fpga-is-a-grid/">01 An FPGA is a grid you configure</a></li><li><a href="/learn/which-chip/">02 Which chip is this?</a></li><li class="on" aria-current="page"><a href="/learn/pins-and-package/">03 Pins, balls and the package</a></li></ol></li><li><h3>Module 2 · Numbers in hardware</h3><p>Bits, trits and number formats, and what arithmetic costs in logic.</p><ol><li><a href="/learn/bits-and-trits/">04 Bits, trits and GF16</a></li><li><a href="/learn/what-arithmetic-costs/">05 What arithmetic costs</a></li><li><a href="/learn/how-much-fits/">06 How much fits on the chip</a></li></ol></li><li><h3>Module 3 · Your t27 program</h3><p>Write a spec, test it, and see why compiling is not the same as being right.</p><ol><li><a href="/learn/hello-t27/">07 Hello, t27</a></li><li><a href="/learn/tests-are-the-spec/">08 Tests are part of the spec</a></li><li><a href="/learn/compiles-is-not-correct/">09 Compiles is not the same as correct</a></li></ol></li><li><h3>Module 4 · Inside t27c</h3><p>How the compiler reads a spec and what it writes, including the native t27b.</p><ol><li><a href="/learn/compiler-stages/">10 The stages of t27c</a></li><li><a href="/learn/t27c-on-a-real-machine/">11 t27c on a real machine</a></li><li><a href="/learn/seven-backends-and-t27b/">12 Seven backends, and t27b</a></li></ol></li><li><h3>Module 5 · From spec to hardware</h3><p>The Verilog t27c writes, the cells it becomes, and what one LUT does.</p><ol><li><a href="/learn/spec-to-verilog/">13 From a spec to Verilog</a></li><li><a href="/learn/count-the-cells/">14 Count the cells</a></li><li><a href="/learn/one-lut-up-close/">15 One LUT, up close</a></li></ol></li><li><h3>Module 6 · Reading synthesis</h3><p>What yosys reports about your design, and which warnings matter.</p><ol><li><a href="/learn/synthesis-warnings/">16 Synthesis warnings that matter</a></li><li><a href="/learn/state-machines/">17 State machines</a></li><li><a href="/learn/where-the-luts-go/">18 Where the LUTs go</a></li></ol></li><li><h3>Module 7 · Place, route, timing</h3><p>Where the cells land on the die, and whether the clock is met.</p><ol><li><a href="/learn/placement/">19 Placement: where the cells land</a></li><li><a href="/learn/logic-depth/">20 Logic depth</a></li><li><a href="/learn/timing-slack/">21 Timing and slack</a></li></ol></li><li><h3>Module 8 · The bitstream</h3><p>How a routed design becomes the bits the chip loads, with no Vivado.</p><ol><li><a href="/learn/fasm-to-frames/">22 From FASM to frames</a></li><li><a href="/learn/a-build-receipt/">23 A build, itemised</a></li><li><a href="/learn/read-a-bitstream/">24 Read a bitstream</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>Ask the chip who it is, load the bits, and check them.</p><ol><li><a href="/learn/jtag-asks-the-chip/">25 JTAG asks the chip</a></li><li><a href="/learn/same-bits-on-the-board/">26 The same bits, on the board</a></li><li><a href="/learn/capstone/">27 Capstone: out-guess yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/placement/index.html b/apps/website/public/learn/placement/index.html
index a77008af9b..d6d4640283 100644
--- a/apps/website/public/learn/placement/index.html
+++ b/apps/website/public/learn/placement/index.html
@@ -312,5 +312,6 @@ <h1>Placement: where the cells land</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The chip</h3><p>What an FPGA is, which chip we use, and how its pins meet the board.</p><ol><li><a href="/learn/fpga-is-a-grid/">01 An FPGA is a grid you configure</a></li><li><a href="/learn/which-chip/">02 Which chip is this?</a></li><li><a href="/learn/pins-and-package/">03 Pins, balls and the package</a></li></ol></li><li><h3>Module 2 · Numbers in hardware</h3><p>Bits, trits and number formats, and what arithmetic costs in logic.</p><ol><li><a href="/learn/bits-and-trits/">04 Bits, trits and GF16</a></li><li><a href="/learn/what-arithmetic-costs/">05 What arithmetic costs</a></li><li><a href="/learn/how-much-fits/">06 How much fits on the chip</a></li></ol></li><li><h3>Module 3 · Your t27 program</h3><p>Write a spec, test it, and see why compiling is not the same as being right.</p><ol><li><a href="/learn/hello-t27/">07 Hello, t27</a></li><li><a href="/learn/tests-are-the-spec/">08 Tests are part of the spec</a></li><li><a href="/learn/compiles-is-not-correct/">09 Compiles is not the same as correct</a></li></ol></li><li><h3>Module 4 · Inside t27c</h3><p>How the compiler reads a spec and what it writes, including the native t27b.</p><ol><li><a href="/learn/compiler-stages/">10 The stages of t27c</a></li><li><a href="/learn/t27c-on-a-real-machine/">11 t27c on a real machine</a></li><li><a href="/learn/seven-backends-and-t27b/">12 Seven backends, and t27b</a></li></ol></li><li><h3>Module 5 · From spec to hardware</h3><p>The Verilog t27c writes, the cells it becomes, and what one LUT does.</p><ol><li><a href="/learn/spec-to-verilog/">13 From a spec to Verilog</a></li><li><a href="/learn/count-the-cells/">14 Count the cells</a></li><li><a href="/learn/one-lut-up-close/">15 One LUT, up close</a></li></ol></li><li><h3>Module 6 · Reading synthesis</h3><p>What yosys reports about your design, and which warnings matter.</p><ol><li><a href="/learn/synthesis-warnings/">16 Synthesis warnings that matter</a></li><li><a href="/learn/state-machines/">17 State machines</a></li><li><a href="/learn/where-the-luts-go/">18 Where the LUTs go</a></li></ol></li><li><h3>Module 7 · Place, route, timing</h3><p>Where the cells land on the die, and whether the clock is met.</p><ol><li class="on" aria-current="page"><a href="/learn/placement/">19 Placement: where the cells land</a></li><li><a href="/learn/logic-depth/">20 Logic depth</a></li><li><a href="/learn/timing-slack/">21 Timing and slack</a></li></ol></li><li><h3>Module 8 · The bitstream</h3><p>How a routed design becomes the bits the chip loads, with no Vivado.</p><ol><li><a href="/learn/fasm-to-frames/">22 From FASM to frames</a></li><li><a href="/learn/a-build-receipt/">23 A build, itemised</a></li><li><a href="/learn/read-a-bitstream/">24 Read a bitstream</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>Ask the chip who it is, load the bits, and check them.</p><ol><li><a href="/learn/jtag-asks-the-chip/">25 JTAG asks the chip</a></li><li><a href="/learn/same-bits-on-the-board/">26 The same bits, on the board</a></li><li><a href="/learn/capstone/">27 Capstone: out-guess yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/plan-before-code/index.html b/apps/website/public/learn/plan-before-code/index.html
index 981a0f4999..a6270fef4a 100644
--- a/apps/website/public/learn/plan-before-code/index.html
+++ b/apps/website/public/learn/plan-before-code/index.html
@@ -193,5 +193,6 @@ <h1>Plan before code</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Why verify</h3><p>Designs that compile and are wrong, the model that decides, and the plan written before the code.</p><ol><li><a href="/learn/bugs-that-compile/">01 Bugs that compile</a></li><li><a href="/learn/golden-model/">02 The golden model</a></li><li class="on" aria-current="page"><a href="/learn/plan-before-code/">03 Plan before code</a></li></ol></li><li><h3>Module 2 · Testbenches</h3><p>Stimulus, checks and a verdict, written as one spec beside the design it judges.</p><ol><li><a href="/learn/anatomy-of-a-tb/">04 Anatomy of a testbench</a></li><li><a href="/learn/self-checking/">05 Self-checking, not staring</a></li><li><a href="/learn/directed-vs-random/">06 Directed versus random</a></li></ol></li><li><h3>Module 3 · Waveforms</h3><p>A trace of every signal, read the way a hardware engineer reads it, and two runs compared.</p><ol><li><a href="/learn/vcd-format/">07 The VCD format</a></li><li><a href="/learn/debug-with-waves/">08 Debug with waves</a></li><li><a href="/learn/compare-traces/">09 Compare two traces</a></li></ol></li><li><h3>Module 4 · Conformance vectors</h3><p>Cases with the answer written beside them, kept where the compiler can reach them.</p><ol><li><a href="/learn/what-a-vector-is/">10 What a vector is</a></li><li><a href="/learn/golden-ruler-vectors/">11 Golden ruler vectors</a></li><li><a href="/learn/end-to-end-scenarios/">12 End-to-end scenarios</a></li></ol></li><li><h3>Module 5 · Cosimulation</h3><p>Spec, simulator and board agreeing on the bench Artix-7 XC7A200T, and what to do when they do not.</p><ol><li><a href="/learn/three-way-agreement/">13 Three-way agreement</a></li><li><a href="/learn/simulator-internals/">14 Inside the simulator</a></li><li><a href="/learn/when-they-disagree/">15 When they disagree</a></li></ol></li><li><h3>Module 6 · Coverage</h3><p>What the tests touched: lines, toggles, states, and what that number hides.</p><ol><li><a href="/learn/line-and-toggle/">16 Line and toggle</a></li><li><a href="/learn/fsm-coverage/">17 FSM coverage</a></li><li><a href="/learn/coverage-lies/">18 Coverage lies</a></li></ol></li><li><h3>Module 7 · Formal</h3><p>Assertions that hold every cycle, bounded search for a counterexample, and why a proof needs induction.</p><ol><li><a href="/learn/assertions/">19 Assertions</a></li><li><a href="/learn/bounded-model-checking/">20 Bounded model checking</a></li><li><a href="/learn/induction/">21 Induction</a></li></ol></li><li><h3>Module 8 · Mutation</h3><p>Break the design on purpose and count what the tests catch.</p><ol><li><a href="/learn/kill-the-mutant/">22 Kill the mutant</a></li><li><a href="/learn/surviving-mutants/">23 Surviving mutants</a></li><li><a href="/learn/test-strength/">24 Test strength</a></li></ol></li><li><h3>Module 9 · Sign-off</h3><p>One command, every receipt, a clean verdict you can show.</p><ol><li><a href="/learn/build-verify/">25 Build and verify</a></li><li><a href="/learn/board-selftest/">26 The board self-tests</a></li><li><a href="/learn/verify-capstone/">27 Capstone: verify a module whole</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/pointers-and-flags/index.html b/apps/website/public/learn/pointers-and-flags/index.html
index e4683a2857..ecbcd88219 100644
--- a/apps/website/public/learn/pointers-and-flags/index.html
+++ b/apps/website/public/learn/pointers-and-flags/index.html
@@ -434,5 +434,6 @@ <h1>Pointers and flags</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a clock is</h3><p>One edge, one world: what shares a clock edge shares a world, the period and the jitter of a real edge, and where the clock enters a board.</p><ol><li><a href="/learn/one-edge-one-world/">01 One edge, one world</a></li><li><a href="/learn/period-and-jitter/">02 A period, an edge, and jitter</a></li><li><a href="/learn/where-the-clock-enters/">03 Where the clock enters</a></li></ol></li><li><h3>Module 2 · Clock trees</h3><p>Skew and insertion delay, the global buffer network, and the trap of gating a clock with logic.</p><ol><li><a href="/learn/skew-and-insertion/">04 Skew and insertion delay</a></li><li><a href="/learn/global-buffers/">05 The global buffer</a></li><li><a href="/learn/gated-clock-trap/">06 The gated-clock trap</a></li></ol></li><li><h3>Module 3 · PLL and MMCM</h3><p>Multiply and divide one clock into another, move its phase in steps of the VCO, and which clocks the analyzer treats as related.</p><ol><li><a href="/learn/multiply-and-divide/">07 Multiply and divide</a></li><li><a href="/learn/phase-and-lock/">08 Phase and lock</a></li><li><a href="/learn/related-or-not/">09 Related or not</a></li></ol></li><li><h3>Module 4 · Resets</h3><p>Assert asynchronously, release synchronously: the three reset kinds, the release pipe, and the tree a reset grows.</p><ol><li><a href="/learn/sync-or-async/">10 Sync or async</a></li><li><a href="/learn/reset-release/">11 The release pipe</a></li><li><a href="/learn/reset-fanout/">12 Reset fanout</a></li></ol></li><li><h3>Module 5 · Metastability</h3><p>The setup-hold window, the mean time between failures in integer arithmetic, and the two flops that fix it.</p><ol><li><a href="/learn/setup-hold-window/">13 The setup-hold window</a></li><li><a href="/learn/mtbf/">14 Mean time between failures</a></li><li><a href="/learn/two-flop-sync/">15 The two-flop synchronizer</a></li></ol></li><li><h3>Module 6 · Crossing many bits</h3><p>Why a binary bus tears, why Gray code does not, and the handshake that moves a pulse between worlds.</p><ol><li><a href="/learn/why-binary-fails/">16 Why binary fails</a></li><li><a href="/learn/gray-code/">17 Gray code</a></li><li><a href="/learn/handshakes-and-pulses/">18 Handshakes and pulses</a></li></ol></li><li><h3>Module 7 · The asynchronous FIFO</h3><p>Pointers, flags and depth: the buffer that moves a stream between two clocks.</p><ol><li class="on" aria-current="page"><a href="/learn/pointers-and-flags/">19 Pointers and flags</a></li><li><a href="/learn/depth-sizing/">20 Sizing the depth</a></li><li><a href="/learn/fifo-testbench/">21 The FIFO testbench</a></li></ol></li><li><h3>Module 8 · Constraints</h3><p>The lines that tell the analyzer what a clock is, which paths not to check, and what the pins must meet.</p><ol><li><a href="/learn/create-clock/">22 create_clock</a></li><li><a href="/learn/false-path-and-max-delay/">23 False paths and max delay</a></li><li><a href="/learn/io-timing/">24 IO timing</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>A CDC report, one crossing captured at the flip-flops, and the bitstream diff that closes the course.</p><ol><li><a href="/learn/cdc-report/">25 The CDC report</a></li><li><a href="/learn/capture-a-crossing/">26 Capture a crossing</a></li><li><a href="/learn/capstone-diff/">27 Capstone: change it and diff it</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/polyphase/index.html b/apps/website/public/learn/polyphase/index.html
index d6d84caa15..2010749016 100644
--- a/apps/website/public/learn/polyphase/index.html
+++ b/apps/website/public/learn/polyphase/index.html
@@ -1122,5 +1122,6 @@ <h1>Polyphase</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Adding numbers</h3><p>Why addition is slow: the carry that walks up the word, the dedicated chain in every slice, prefix networks and carry-save trees.</p><ol><li><a href="/learn/the-carry-chain/">01 The carry chain</a></li><li><a href="/learn/prefix-adders/">02 Looking ahead: prefix adders</a></li><li><a href="/learn/carry-save/">03 Three numbers at once</a></li></ol></li><li><h3>Module 2 · Multiplying</h3><p>A product is a sum of shifted copies; Booth recoding halves them, and the DSP48E1 slice does the rest in one block.</p><ol><li><a href="/learn/shift-and-add/">04 Shift and add</a></li><li><a href="/learn/booths-trick/">05 Booth's trick</a></li><li><a href="/learn/the-dsp48e1-slice/">06 The DSP48E1 slice</a></li></ol></li><li><h3>Module 3 · Fixed point</h3><p>Where the binary point sits, what rounding does to a value and to its average, and what each bit of a quantizer buys.</p><ol><li><a href="/learn/q-formats/">07 Q formats</a></li><li><a href="/learn/rounding-and-saturation/">08 Rounding and saturation</a></li><li><a href="/learn/quantization-noise/">09 Quantization noise</a></li></ol></li><li><h3>Module 4 · Functions in hardware</h3><p>Sine, cosine, angle and length from shifts and adds, and when a table is the better answer.</p><ol><li><a href="/learn/rotating-by-shifts/">10 Rotating by shifts</a></li><li><a href="/learn/angle-and-magnitude/">11 Angle and magnitude</a></li><li><a href="/learn/tables-and-interpolation/">12 Tables and interpolation</a></li></ol></li><li><h3>Module 5 · Signals and sampling</h3><p>What sampling does to a frequency, how an oscillator is built from an adder, and what a DFT bin measures.</p><ol><li><a href="/learn/sampling-and-aliasing/">13 Sampling and aliasing</a></li><li><a href="/learn/the-nco/">14 A numerically controlled oscillator</a></li><li><a href="/learn/bins-of-a-spectrum/">15 Bins of a spectrum</a></li></ol></li><li><h3>Module 6 · FIR filters</h3><p>The moving average, a windowed-sinc design with integer taps, and folding the taps onto DSP slices.</p><ol><li><a href="/learn/the-moving-average/">16 The moving average</a></li><li><a href="/learn/designing-taps/">17 Designing taps</a></li><li><a href="/learn/folding-taps-onto-slices/">18 Folding taps onto slices</a></li></ol></li><li><h3>Module 7 · Multirate</h3><p>Lowering the sample rate without folding noise in: decimation, the CIC filter and the polyphase form.</p><ol><li><a href="/learn/lowering-the-rate/">19 Lowering the rate</a></li><li><a href="/learn/the-cic-filter/">20 The CIC filter</a></li><li class="on" aria-current="page"><a href="/learn/polyphase/">21 Polyphase</a></li></ol></li><li><h3>Module 8 · The FFT</h3><p>N log N instead of N^2: butterflies, the bit-reversed input order and the bits each stage adds.</p><ol><li><a href="/learn/the-butterfly/">22 The butterfly</a></li><li><a href="/learn/twiddles-and-bit-reversal/">23 Twiddles and bit reversal</a></li><li><a href="/learn/growth-through-the-stages/">24 Growth through the stages</a></li></ol></li><li><h3>Module 9 · On the bench</h3><p>From the arithmetic to the board: a correlator from the modem spec, a budget of slices and timing, and the capstone filter.</p><ol><li><a href="/learn/a-correlator-finds-the-frame/">25 A correlator finds the frame</a></li><li><a href="/learn/does-it-fit-and-close-timing/">26 Does it fit and close timing?</a></li><li><a href="/learn/capstone-a-filter-on-the-board/">27 Capstone: a filter on the board</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/ports-and-kinds/index.html b/apps/website/public/learn/ports-and-kinds/index.html
index 8a96120347..a1013e1f9d 100644
--- a/apps/website/public/learn/ports-and-kinds/index.html
+++ b/apps/website/public/learn/ports-and-kinds/index.html
@@ -424,5 +424,6 @@ <h1>Ports and kinds</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a bus is</h3><p>Why a bus exists at all: a conversation on wires, framed and addressed, and who is allowed to talk.</p><ol><li><a href="/learn/wires-that-talk/">01 Wires that talk</a></li><li><a href="/learn/framing-a-conversation/">02 Framing a conversation</a></li><li><a href="/learn/master-and-servant/">03 Master and servant</a></li></ol></li><li><h3>Module 2 · UART</h3><p>The two-wire bus with no clock: the frame, the divisor that sets the rate, and the status a driver polls.</p><ol><li><a href="/learn/the-frame/">04 The frame: start, data, stop</a></li><li><a href="/learn/baud-and-the-divisor/">05 Baud and the divisor</a></li><li><a href="/learn/status-and-fifos/">06 Status and FIFOs</a></li></ol></li><li><h3>Module 3 · SPI</h3><p>The clocked conversation: four modes, a prescaler ladder, and a chip select per servant.</p><ol><li><a href="/learn/modes-cpol-cpha/">07 Modes: CPOL and CPHA</a></li><li><a href="/learn/the-prescaler-ladder/">08 The prescaler ladder</a></li><li><a href="/learn/chip-select-and-widths/">09 Chip select and widths</a></li></ol></li><li><h3>Module 4 · APB</h3><p>The register bus: PSEL and PENABLE, strobes and wait states, and how many address bits a peripheral count costs.</p><ol><li><a href="/learn/psel-and-penable/">10 PSEL and PENABLE</a></li><li><a href="/learn/strobes-and-waits/">11 Strobes and waits</a></li><li><a href="/learn/address-decode/">12 Address decode</a></li></ol></li><li><h3>Module 5 · AXI4</h3><p>The five channels: address, data and response in both directions, lite or full, bursts and IDs.</p><ol><li><a href="/learn/five-channels/">13 Five channels</a></li><li><a href="/learn/lite-or-full/">14 Lite or full</a></li><li><a href="/learn/bursts-and-ids/">15 Bursts and IDs</a></li></ol></li><li><h3>Module 6 · Memory</h3><p>What sits on the far side of every bus: memory maps, port kinds, and latency that a wait state must cover.</p><ol><li><a href="/learn/memory-maps/">16 Memory maps</a></li><li class="on" aria-current="page"><a href="/learn/ports-and-kinds/">17 Ports and kinds</a></li><li><a href="/learn/latency/">18 Latency</a></li></ol></li><li><h3>Module 7 · Bridges</h3><p>Why designs grow more than one bus, and the packet bridge that moves work between them.</p><ol><li><a href="/learn/why-bridges/">19 Why bridges</a></li><li><a href="/learn/the-packet-bridge/">20 The packet bridge</a></li><li><a href="/learn/everything-in-one-bench/">21 Everything in one bench</a></li></ol></li><li><h3>Module 8 · Ethernet</h3><p>Frames, the frame check sequence, RGMII timing, and the pre-registered steps of a real bring-up.</p><ol><li><a href="/learn/frames-and-the-fcs/">22 Frames and the FCS</a></li><li><a href="/learn/rgmii-timing/">23 RGMII timing</a></li><li><a href="/learn/bring-up-on-the-bench/">24 Bring-up on the bench</a></li></ol></li><li><h3>Module 9 · The bench</h3><p>The discipline that guards real hardware: who holds the IO, taking and giving the lock, and what runs next.</p><ol><li><a href="/learn/who-holds-the-io/">25 Who holds the IO</a></li><li><a href="/learn/claim-and-release/">26 Claim and release</a></li><li><a href="/learn/capstone-what-runs-next/">27 Capstone: what runs next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/prefix-adders/index.html b/apps/website/public/learn/prefix-adders/index.html
index 545e29a407..6623cf5fef 100644
--- a/apps/website/public/learn/prefix-adders/index.html
+++ b/apps/website/public/learn/prefix-adders/index.html
@@ -554,5 +554,6 @@ <h1>Looking ahead: prefix adders</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Adding numbers</h3><p>Why addition is slow: the carry that walks up the word, the dedicated chain in every slice, prefix networks and carry-save trees.</p><ol><li><a href="/learn/the-carry-chain/">01 The carry chain</a></li><li class="on" aria-current="page"><a href="/learn/prefix-adders/">02 Looking ahead: prefix adders</a></li><li><a href="/learn/carry-save/">03 Three numbers at once</a></li></ol></li><li><h3>Module 2 · Multiplying</h3><p>A product is a sum of shifted copies; Booth recoding halves them, and the DSP48E1 slice does the rest in one block.</p><ol><li><a href="/learn/shift-and-add/">04 Shift and add</a></li><li><a href="/learn/booths-trick/">05 Booth's trick</a></li><li><a href="/learn/the-dsp48e1-slice/">06 The DSP48E1 slice</a></li></ol></li><li><h3>Module 3 · Fixed point</h3><p>Where the binary point sits, what rounding does to a value and to its average, and what each bit of a quantizer buys.</p><ol><li><a href="/learn/q-formats/">07 Q formats</a></li><li><a href="/learn/rounding-and-saturation/">08 Rounding and saturation</a></li><li><a href="/learn/quantization-noise/">09 Quantization noise</a></li></ol></li><li><h3>Module 4 · Functions in hardware</h3><p>Sine, cosine, angle and length from shifts and adds, and when a table is the better answer.</p><ol><li><a href="/learn/rotating-by-shifts/">10 Rotating by shifts</a></li><li><a href="/learn/angle-and-magnitude/">11 Angle and magnitude</a></li><li><a href="/learn/tables-and-interpolation/">12 Tables and interpolation</a></li></ol></li><li><h3>Module 5 · Signals and sampling</h3><p>What sampling does to a frequency, how an oscillator is built from an adder, and what a DFT bin measures.</p><ol><li><a href="/learn/sampling-and-aliasing/">13 Sampling and aliasing</a></li><li><a href="/learn/the-nco/">14 A numerically controlled oscillator</a></li><li><a href="/learn/bins-of-a-spectrum/">15 Bins of a spectrum</a></li></ol></li><li><h3>Module 6 · FIR filters</h3><p>The moving average, a windowed-sinc design with integer taps, and folding the taps onto DSP slices.</p><ol><li><a href="/learn/the-moving-average/">16 The moving average</a></li><li><a href="/learn/designing-taps/">17 Designing taps</a></li><li><a href="/learn/folding-taps-onto-slices/">18 Folding taps onto slices</a></li></ol></li><li><h3>Module 7 · Multirate</h3><p>Lowering the sample rate without folding noise in: decimation, the CIC filter and the polyphase form.</p><ol><li><a href="/learn/lowering-the-rate/">19 Lowering the rate</a></li><li><a href="/learn/the-cic-filter/">20 The CIC filter</a></li><li><a href="/learn/polyphase/">21 Polyphase</a></li></ol></li><li><h3>Module 8 · The FFT</h3><p>N log N instead of N^2: butterflies, the bit-reversed input order and the bits each stage adds.</p><ol><li><a href="/learn/the-butterfly/">22 The butterfly</a></li><li><a href="/learn/twiddles-and-bit-reversal/">23 Twiddles and bit reversal</a></li><li><a href="/learn/growth-through-the-stages/">24 Growth through the stages</a></li></ol></li><li><h3>Module 9 · On the bench</h3><p>From the arithmetic to the board: a correlator from the modem spec, a budget of slices and timing, and the capstone filter.</p><ol><li><a href="/learn/a-correlator-finds-the-frame/">25 A correlator finds the frame</a></li><li><a href="/learn/does-it-fit-and-close-timing/">26 Does it fit and close timing?</a></li><li><a href="/learn/capstone-a-filter-on-the-board/">27 Capstone: a filter on the board</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/psel-and-penable/index.html b/apps/website/public/learn/psel-and-penable/index.html
index 4dcc5e7710..c29e70ddf5 100644
--- a/apps/website/public/learn/psel-and-penable/index.html
+++ b/apps/website/public/learn/psel-and-penable/index.html
@@ -417,5 +417,6 @@ <h1>PSEL and PENABLE</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a bus is</h3><p>Why a bus exists at all: a conversation on wires, framed and addressed, and who is allowed to talk.</p><ol><li><a href="/learn/wires-that-talk/">01 Wires that talk</a></li><li><a href="/learn/framing-a-conversation/">02 Framing a conversation</a></li><li><a href="/learn/master-and-servant/">03 Master and servant</a></li></ol></li><li><h3>Module 2 · UART</h3><p>The two-wire bus with no clock: the frame, the divisor that sets the rate, and the status a driver polls.</p><ol><li><a href="/learn/the-frame/">04 The frame: start, data, stop</a></li><li><a href="/learn/baud-and-the-divisor/">05 Baud and the divisor</a></li><li><a href="/learn/status-and-fifos/">06 Status and FIFOs</a></li></ol></li><li><h3>Module 3 · SPI</h3><p>The clocked conversation: four modes, a prescaler ladder, and a chip select per servant.</p><ol><li><a href="/learn/modes-cpol-cpha/">07 Modes: CPOL and CPHA</a></li><li><a href="/learn/the-prescaler-ladder/">08 The prescaler ladder</a></li><li><a href="/learn/chip-select-and-widths/">09 Chip select and widths</a></li></ol></li><li><h3>Module 4 · APB</h3><p>The register bus: PSEL and PENABLE, strobes and wait states, and how many address bits a peripheral count costs.</p><ol><li class="on" aria-current="page"><a href="/learn/psel-and-penable/">10 PSEL and PENABLE</a></li><li><a href="/learn/strobes-and-waits/">11 Strobes and waits</a></li><li><a href="/learn/address-decode/">12 Address decode</a></li></ol></li><li><h3>Module 5 · AXI4</h3><p>The five channels: address, data and response in both directions, lite or full, bursts and IDs.</p><ol><li><a href="/learn/five-channels/">13 Five channels</a></li><li><a href="/learn/lite-or-full/">14 Lite or full</a></li><li><a href="/learn/bursts-and-ids/">15 Bursts and IDs</a></li></ol></li><li><h3>Module 6 · Memory</h3><p>What sits on the far side of every bus: memory maps, port kinds, and latency that a wait state must cover.</p><ol><li><a href="/learn/memory-maps/">16 Memory maps</a></li><li><a href="/learn/ports-and-kinds/">17 Ports and kinds</a></li><li><a href="/learn/latency/">18 Latency</a></li></ol></li><li><h3>Module 7 · Bridges</h3><p>Why designs grow more than one bus, and the packet bridge that moves work between them.</p><ol><li><a href="/learn/why-bridges/">19 Why bridges</a></li><li><a href="/learn/the-packet-bridge/">20 The packet bridge</a></li><li><a href="/learn/everything-in-one-bench/">21 Everything in one bench</a></li></ol></li><li><h3>Module 8 · Ethernet</h3><p>Frames, the frame check sequence, RGMII timing, and the pre-registered steps of a real bring-up.</p><ol><li><a href="/learn/frames-and-the-fcs/">22 Frames and the FCS</a></li><li><a href="/learn/rgmii-timing/">23 RGMII timing</a></li><li><a href="/learn/bring-up-on-the-bench/">24 Bring-up on the bench</a></li></ol></li><li><h3>Module 9 · The bench</h3><p>The discipline that guards real hardware: who holds the IO, taking and giving the lock, and what runs next.</p><ol><li><a href="/learn/who-holds-the-io/">25 Who holds the IO</a></li><li><a href="/learn/claim-and-release/">26 Claim and release</a></li><li><a href="/learn/capstone-what-runs-next/">27 Capstone: what runs next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/pub-or-private/index.html b/apps/website/public/learn/pub-or-private/index.html
index fae4eb335f..35fcb66b03 100644
--- a/apps/website/public/learn/pub-or-private/index.html
+++ b/apps/website/public/learn/pub-or-private/index.html
@@ -83,5 +83,6 @@ <h1>pub or private</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · A spec file</h3><p>What a .t27 file holds: a module line, prose lines and comments.</p><ol><li><a href="/learn/what-a-spec-is/">01 What a spec is</a></li><li><a href="/learn/the-module-line/">02 The module line</a></li><li><a href="/learn/comments-and-prose/">03 Comments and prose lines</a></li></ol></li><li><h3>Module 2 · Values and types</h3><p>Constants, the width of a whole number, and true, false and text.</p><ol><li><a href="/learn/constants/">04 Constants</a></li><li><a href="/learn/whole-numbers/">05 Whole numbers and their widths</a></li><li><a href="/learn/true-false-and-text/">06 True, false and text</a></li></ol></li><li><h3>Module 3 · Arrays, trits and expressions</h3><p>Lists of values, the three values of a trit, and what operators compute.</p><ol><li><a href="/learn/arrays/">07 Arrays</a></li><li><a href="/learn/trits/">08 Trits: minus one, zero, one</a></li><li><a href="/learn/expressions/">09 Expressions</a></li></ol></li><li><h3>Module 4 · Tests and functions</h3><p>A test block, several small tests in one spec, and a function with typed inputs.</p><ol><li><a href="/learn/a-test-block/">10 A test block</a></li><li><a href="/learn/many-tests/">11 Many tests in one spec</a></li><li><a href="/learn/functions/">12 Functions</a></li></ol></li><li><h3>Module 5 · Inside a function</h3><p>Names with let and var, choices with if and switch, and loops.</p><ol><li><a href="/learn/local-names/">13 Local names: let and var</a></li><li><a href="/learn/if-and-else/">14 Choices: if, else and switch</a></li><li><a href="/learn/loops/">15 Loops: while and for</a></li></ol></li><li><h3>Module 6 · Visibility and your own types</h3><p>What pub shares with other modules, and the structs and enums you declare.</p><ol><li class="on" aria-current="page"><a href="/learn/pub-or-private/">16 pub or private</a></li><li><a href="/learn/structs/">17 Structs</a></li><li><a href="/learn/enums/">18 Enums</a></li></ol></li><li><h3>Module 7 · Modules, rules and gen-ts</h3><p>Reading another module, stating a rule that always holds, and a spec turned into TypeScript.</p><ol><li><a href="/learn/use-other-modules/">19 Using other modules</a></li><li><a href="/learn/invariants/">20 Invariants and benches</a></li><li><a href="/learn/gen-ts/">21 gen-ts: a spec becomes TypeScript</a></li></ol></li><li><h3>Module 8 · The compiler and tri</h3><p>The backends t27c writes, the stages it reads a spec in, and the tri command.</p><ol><li><a href="/learn/seven-backends/">22 Seven backends</a></li><li><a href="/learn/stages-of-t27c/">23 The stages of t27c</a></li><li><a href="/learn/tri-commands/">24 tri: the command line</a></li></ol></li><li><h3>Module 9 · Errors, a program, what next</h3><p>Reading a compiler error, one small complete program, and the course to take next.</p><ol><li><a href="/learn/reading-errors/">25 Reading compiler errors</a></li><li><a href="/learn/a-small-program/">26 A small complete program</a></li><li><a href="/learn/where-next/">27 Where to go next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/q-formats/index.html b/apps/website/public/learn/q-formats/index.html
index 8fb4115a4e..705c77348b 100644
--- a/apps/website/public/learn/q-formats/index.html
+++ b/apps/website/public/learn/q-formats/index.html
@@ -346,5 +346,6 @@ <h1>Q formats</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Adding numbers</h3><p>Why addition is slow: the carry that walks up the word, the dedicated chain in every slice, prefix networks and carry-save trees.</p><ol><li><a href="/learn/the-carry-chain/">01 The carry chain</a></li><li><a href="/learn/prefix-adders/">02 Looking ahead: prefix adders</a></li><li><a href="/learn/carry-save/">03 Three numbers at once</a></li></ol></li><li><h3>Module 2 · Multiplying</h3><p>A product is a sum of shifted copies; Booth recoding halves them, and the DSP48E1 slice does the rest in one block.</p><ol><li><a href="/learn/shift-and-add/">04 Shift and add</a></li><li><a href="/learn/booths-trick/">05 Booth's trick</a></li><li><a href="/learn/the-dsp48e1-slice/">06 The DSP48E1 slice</a></li></ol></li><li><h3>Module 3 · Fixed point</h3><p>Where the binary point sits, what rounding does to a value and to its average, and what each bit of a quantizer buys.</p><ol><li class="on" aria-current="page"><a href="/learn/q-formats/">07 Q formats</a></li><li><a href="/learn/rounding-and-saturation/">08 Rounding and saturation</a></li><li><a href="/learn/quantization-noise/">09 Quantization noise</a></li></ol></li><li><h3>Module 4 · Functions in hardware</h3><p>Sine, cosine, angle and length from shifts and adds, and when a table is the better answer.</p><ol><li><a href="/learn/rotating-by-shifts/">10 Rotating by shifts</a></li><li><a href="/learn/angle-and-magnitude/">11 Angle and magnitude</a></li><li><a href="/learn/tables-and-interpolation/">12 Tables and interpolation</a></li></ol></li><li><h3>Module 5 · Signals and sampling</h3><p>What sampling does to a frequency, how an oscillator is built from an adder, and what a DFT bin measures.</p><ol><li><a href="/learn/sampling-and-aliasing/">13 Sampling and aliasing</a></li><li><a href="/learn/the-nco/">14 A numerically controlled oscillator</a></li><li><a href="/learn/bins-of-a-spectrum/">15 Bins of a spectrum</a></li></ol></li><li><h3>Module 6 · FIR filters</h3><p>The moving average, a windowed-sinc design with integer taps, and folding the taps onto DSP slices.</p><ol><li><a href="/learn/the-moving-average/">16 The moving average</a></li><li><a href="/learn/designing-taps/">17 Designing taps</a></li><li><a href="/learn/folding-taps-onto-slices/">18 Folding taps onto slices</a></li></ol></li><li><h3>Module 7 · Multirate</h3><p>Lowering the sample rate without folding noise in: decimation, the CIC filter and the polyphase form.</p><ol><li><a href="/learn/lowering-the-rate/">19 Lowering the rate</a></li><li><a href="/learn/the-cic-filter/">20 The CIC filter</a></li><li><a href="/learn/polyphase/">21 Polyphase</a></li></ol></li><li><h3>Module 8 · The FFT</h3><p>N log N instead of N^2: butterflies, the bit-reversed input order and the bits each stage adds.</p><ol><li><a href="/learn/the-butterfly/">22 The butterfly</a></li><li><a href="/learn/twiddles-and-bit-reversal/">23 Twiddles and bit reversal</a></li><li><a href="/learn/growth-through-the-stages/">24 Growth through the stages</a></li></ol></li><li><h3>Module 9 · On the bench</h3><p>From the arithmetic to the board: a correlator from the modem spec, a budget of slices and timing, and the capstone filter.</p><ol><li><a href="/learn/a-correlator-finds-the-frame/">25 A correlator finds the frame</a></li><li><a href="/learn/does-it-fit-and-close-timing/">26 Does it fit and close timing?</a></li><li><a href="/learn/capstone-a-filter-on-the-board/">27 Capstone: a filter on the board</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/quantization-noise/index.html b/apps/website/public/learn/quantization-noise/index.html
index 8d5acb3494..7bfbc972d5 100644
--- a/apps/website/public/learn/quantization-noise/index.html
+++ b/apps/website/public/learn/quantization-noise/index.html
@@ -1122,5 +1122,6 @@ <h1>Quantization noise</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Adding numbers</h3><p>Why addition is slow: the carry that walks up the word, the dedicated chain in every slice, prefix networks and carry-save trees.</p><ol><li><a href="/learn/the-carry-chain/">01 The carry chain</a></li><li><a href="/learn/prefix-adders/">02 Looking ahead: prefix adders</a></li><li><a href="/learn/carry-save/">03 Three numbers at once</a></li></ol></li><li><h3>Module 2 · Multiplying</h3><p>A product is a sum of shifted copies; Booth recoding halves them, and the DSP48E1 slice does the rest in one block.</p><ol><li><a href="/learn/shift-and-add/">04 Shift and add</a></li><li><a href="/learn/booths-trick/">05 Booth's trick</a></li><li><a href="/learn/the-dsp48e1-slice/">06 The DSP48E1 slice</a></li></ol></li><li><h3>Module 3 · Fixed point</h3><p>Where the binary point sits, what rounding does to a value and to its average, and what each bit of a quantizer buys.</p><ol><li><a href="/learn/q-formats/">07 Q formats</a></li><li><a href="/learn/rounding-and-saturation/">08 Rounding and saturation</a></li><li class="on" aria-current="page"><a href="/learn/quantization-noise/">09 Quantization noise</a></li></ol></li><li><h3>Module 4 · Functions in hardware</h3><p>Sine, cosine, angle and length from shifts and adds, and when a table is the better answer.</p><ol><li><a href="/learn/rotating-by-shifts/">10 Rotating by shifts</a></li><li><a href="/learn/angle-and-magnitude/">11 Angle and magnitude</a></li><li><a href="/learn/tables-and-interpolation/">12 Tables and interpolation</a></li></ol></li><li><h3>Module 5 · Signals and sampling</h3><p>What sampling does to a frequency, how an oscillator is built from an adder, and what a DFT bin measures.</p><ol><li><a href="/learn/sampling-and-aliasing/">13 Sampling and aliasing</a></li><li><a href="/learn/the-nco/">14 A numerically controlled oscillator</a></li><li><a href="/learn/bins-of-a-spectrum/">15 Bins of a spectrum</a></li></ol></li><li><h3>Module 6 · FIR filters</h3><p>The moving average, a windowed-sinc design with integer taps, and folding the taps onto DSP slices.</p><ol><li><a href="/learn/the-moving-average/">16 The moving average</a></li><li><a href="/learn/designing-taps/">17 Designing taps</a></li><li><a href="/learn/folding-taps-onto-slices/">18 Folding taps onto slices</a></li></ol></li><li><h3>Module 7 · Multirate</h3><p>Lowering the sample rate without folding noise in: decimation, the CIC filter and the polyphase form.</p><ol><li><a href="/learn/lowering-the-rate/">19 Lowering the rate</a></li><li><a href="/learn/the-cic-filter/">20 The CIC filter</a></li><li><a href="/learn/polyphase/">21 Polyphase</a></li></ol></li><li><h3>Module 8 · The FFT</h3><p>N log N instead of N^2: butterflies, the bit-reversed input order and the bits each stage adds.</p><ol><li><a href="/learn/the-butterfly/">22 The butterfly</a></li><li><a href="/learn/twiddles-and-bit-reversal/">23 Twiddles and bit reversal</a></li><li><a href="/learn/growth-through-the-stages/">24 Growth through the stages</a></li></ol></li><li><h3>Module 9 · On the bench</h3><p>From the arithmetic to the board: a correlator from the modem spec, a budget of slices and timing, and the capstone filter.</p><ol><li><a href="/learn/a-correlator-finds-the-frame/">25 A correlator finds the frame</a></li><li><a href="/learn/does-it-fit-and-close-timing/">26 Does it fit and close timing?</a></li><li><a href="/learn/capstone-a-filter-on-the-board/">27 Capstone: a filter on the board</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/questions-in-thirds/index.html b/apps/website/public/learn/questions-in-thirds/index.html
index 1191b51dcc..7af9bdcd56 100644
--- a/apps/website/public/learn/questions-in-thirds/index.html
+++ b/apps/website/public/learn/questions-in-thirds/index.html
@@ -769,5 +769,6 @@ <h1>Questions in thirds</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Three values</h3><p>Why three, how balanced ternary writes every number without a sign, and what flipping and cutting trits do.</p><ol><li><a href="/learn/why-three/">01 Why three</a></li><li><a href="/learn/balanced-ternary/">02 Balanced ternary</a></li><li><a href="/learn/flip-and-cut/">03 Flip to negate, cut to round</a></li></ol></li><li><h3>Module 2 · Logic with unknown</h3><p>Kleene's three-valued gates, two gates binary has no twin for, and addition as a pair of tables.</p><ol><li><a href="/learn/kleene-logic/">04 Three-valued logic</a></li><li><a href="/learn/consensus-and-any/">05 Consensus and any</a></li><li><a href="/learn/sum-and-carry/">06 Sum and carry</a></li></ol></li><li><h3>Module 3 · Adding trits</h3><p>The half adder, the full adder and a carry that ripples left, one place at a time.</p><ol><li><a href="/learn/half-adder/">07 The half adder</a></li><li><a href="/learn/full-adder/">08 The full adder</a></li><li><a href="/learn/ripple-carry/">09 Ripple carry</a></li></ol></li><li><h3>Module 4 · Multiplying without a multiplier</h3><p>Copy, drop or flip: a product by one trit, long multiplication, and a MAC that only adds.</p><ol><li><a href="/learn/times-a-trit/">10 Times one trit</a></li><li><a href="/learn/long-multiplication/">11 Long multiplication</a></li><li><a href="/learn/ternary-mac/">12 A ternary MAC</a></li></ol></li><li><h3>Module 5 · Trits in binary memory</h3><p>Two bits per trit, five trits per byte, and the bits a 27-trit word needs.</p><ol><li><a href="/learn/two-bits-per-trit/">13 Two bits per trit</a></li><li><a href="/learn/five-trits-a-byte/">14 Five trits in a byte</a></li><li><a href="/learn/a-word-of-27/">15 A word of 27 trits</a></li></ol></li><li><h3>Module 6 · The TRI-27 instruction word</h3><p>The 32-bit word the Trinity emulator decodes: its fields, its 47 opcodes and its 15-bit immediate.</p><ol><li><a href="/learn/the-instruction-word/">16 The instruction word</a></li><li><a href="/learn/the-opcode-map/">17 The opcode map</a></li><li><a href="/learn/the-immediate/">18 The immediate</a></li></ol></li><li><h3>Module 7 · A ternary machine</h3><p>An ALU built from this course's adders, a three-way jump, and a program you can step.</p><ol><li><a href="/learn/a-ternary-alu/">19 A ternary ALU</a></li><li><a href="/learn/three-way-jump/">20 A three-way jump</a></li><li><a href="/learn/run-a-program/">21 Run a program</a></li></ol></li><li><h3>Module 8 · Three answers</h3><p>Compare at the highest differing trit, find a number in thirds, sort with three-way compares.</p><ol><li><a href="/learn/compare-at-the-top/">22 Compare at the top</a></li><li class="on" aria-current="page"><a href="/learn/questions-in-thirds/">23 Questions in thirds</a></li><li><a href="/learn/sort-by-three/">24 Sort with three-way compares</a></li></ol></li><li><h3>Module 9 · Ternary neurons</h3><p>Weights of -1, 0, +1: one neuron, a detector and a small layer, with no multiplier anywhere.</p><ol><li><a href="/learn/one-ternary-neuron/">25 One ternary neuron</a></li><li><a href="/learn/a-bar-detector/">26 A bar detector</a></li><li><a href="/learn/a-tiny-layer/">27 A tiny layer</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/read-a-bitstream/index.html b/apps/website/public/learn/read-a-bitstream/index.html
index 7149b18315..94adcff60b 100644
--- a/apps/website/public/learn/read-a-bitstream/index.html
+++ b/apps/website/public/learn/read-a-bitstream/index.html
@@ -110,5 +110,6 @@ <h1>Read a bitstream</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The chip</h3><p>What an FPGA is, which chip we use, and how its pins meet the board.</p><ol><li><a href="/learn/fpga-is-a-grid/">01 An FPGA is a grid you configure</a></li><li><a href="/learn/which-chip/">02 Which chip is this?</a></li><li><a href="/learn/pins-and-package/">03 Pins, balls and the package</a></li></ol></li><li><h3>Module 2 · Numbers in hardware</h3><p>Bits, trits and number formats, and what arithmetic costs in logic.</p><ol><li><a href="/learn/bits-and-trits/">04 Bits, trits and GF16</a></li><li><a href="/learn/what-arithmetic-costs/">05 What arithmetic costs</a></li><li><a href="/learn/how-much-fits/">06 How much fits on the chip</a></li></ol></li><li><h3>Module 3 · Your t27 program</h3><p>Write a spec, test it, and see why compiling is not the same as being right.</p><ol><li><a href="/learn/hello-t27/">07 Hello, t27</a></li><li><a href="/learn/tests-are-the-spec/">08 Tests are part of the spec</a></li><li><a href="/learn/compiles-is-not-correct/">09 Compiles is not the same as correct</a></li></ol></li><li><h3>Module 4 · Inside t27c</h3><p>How the compiler reads a spec and what it writes, including the native t27b.</p><ol><li><a href="/learn/compiler-stages/">10 The stages of t27c</a></li><li><a href="/learn/t27c-on-a-real-machine/">11 t27c on a real machine</a></li><li><a href="/learn/seven-backends-and-t27b/">12 Seven backends, and t27b</a></li></ol></li><li><h3>Module 5 · From spec to hardware</h3><p>The Verilog t27c writes, the cells it becomes, and what one LUT does.</p><ol><li><a href="/learn/spec-to-verilog/">13 From a spec to Verilog</a></li><li><a href="/learn/count-the-cells/">14 Count the cells</a></li><li><a href="/learn/one-lut-up-close/">15 One LUT, up close</a></li></ol></li><li><h3>Module 6 · Reading synthesis</h3><p>What yosys reports about your design, and which warnings matter.</p><ol><li><a href="/learn/synthesis-warnings/">16 Synthesis warnings that matter</a></li><li><a href="/learn/state-machines/">17 State machines</a></li><li><a href="/learn/where-the-luts-go/">18 Where the LUTs go</a></li></ol></li><li><h3>Module 7 · Place, route, timing</h3><p>Where the cells land on the die, and whether the clock is met.</p><ol><li><a href="/learn/placement/">19 Placement: where the cells land</a></li><li><a href="/learn/logic-depth/">20 Logic depth</a></li><li><a href="/learn/timing-slack/">21 Timing and slack</a></li></ol></li><li><h3>Module 8 · The bitstream</h3><p>How a routed design becomes the bits the chip loads, with no Vivado.</p><ol><li><a href="/learn/fasm-to-frames/">22 From FASM to frames</a></li><li><a href="/learn/a-build-receipt/">23 A build, itemised</a></li><li class="on" aria-current="page"><a href="/learn/read-a-bitstream/">24 Read a bitstream</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>Ask the chip who it is, load the bits, and check them.</p><ol><li><a href="/learn/jtag-asks-the-chip/">25 JTAG asks the chip</a></li><li><a href="/learn/same-bits-on-the-board/">26 The same bits, on the board</a></li><li><a href="/learn/capstone/">27 Capstone: out-guess yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/reading-errors/index.html b/apps/website/public/learn/reading-errors/index.html
index 184f38daf3..e8a77f30a0 100644
--- a/apps/website/public/learn/reading-errors/index.html
+++ b/apps/website/public/learn/reading-errors/index.html
@@ -81,5 +81,6 @@ <h1>Reading compiler errors</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · A spec file</h3><p>What a .t27 file holds: a module line, prose lines and comments.</p><ol><li><a href="/learn/what-a-spec-is/">01 What a spec is</a></li><li><a href="/learn/the-module-line/">02 The module line</a></li><li><a href="/learn/comments-and-prose/">03 Comments and prose lines</a></li></ol></li><li><h3>Module 2 · Values and types</h3><p>Constants, the width of a whole number, and true, false and text.</p><ol><li><a href="/learn/constants/">04 Constants</a></li><li><a href="/learn/whole-numbers/">05 Whole numbers and their widths</a></li><li><a href="/learn/true-false-and-text/">06 True, false and text</a></li></ol></li><li><h3>Module 3 · Arrays, trits and expressions</h3><p>Lists of values, the three values of a trit, and what operators compute.</p><ol><li><a href="/learn/arrays/">07 Arrays</a></li><li><a href="/learn/trits/">08 Trits: minus one, zero, one</a></li><li><a href="/learn/expressions/">09 Expressions</a></li></ol></li><li><h3>Module 4 · Tests and functions</h3><p>A test block, several small tests in one spec, and a function with typed inputs.</p><ol><li><a href="/learn/a-test-block/">10 A test block</a></li><li><a href="/learn/many-tests/">11 Many tests in one spec</a></li><li><a href="/learn/functions/">12 Functions</a></li></ol></li><li><h3>Module 5 · Inside a function</h3><p>Names with let and var, choices with if and switch, and loops.</p><ol><li><a href="/learn/local-names/">13 Local names: let and var</a></li><li><a href="/learn/if-and-else/">14 Choices: if, else and switch</a></li><li><a href="/learn/loops/">15 Loops: while and for</a></li></ol></li><li><h3>Module 6 · Visibility and your own types</h3><p>What pub shares with other modules, and the structs and enums you declare.</p><ol><li><a href="/learn/pub-or-private/">16 pub or private</a></li><li><a href="/learn/structs/">17 Structs</a></li><li><a href="/learn/enums/">18 Enums</a></li></ol></li><li><h3>Module 7 · Modules, rules and gen-ts</h3><p>Reading another module, stating a rule that always holds, and a spec turned into TypeScript.</p><ol><li><a href="/learn/use-other-modules/">19 Using other modules</a></li><li><a href="/learn/invariants/">20 Invariants and benches</a></li><li><a href="/learn/gen-ts/">21 gen-ts: a spec becomes TypeScript</a></li></ol></li><li><h3>Module 8 · The compiler and tri</h3><p>The backends t27c writes, the stages it reads a spec in, and the tri command.</p><ol><li><a href="/learn/seven-backends/">22 Seven backends</a></li><li><a href="/learn/stages-of-t27c/">23 The stages of t27c</a></li><li><a href="/learn/tri-commands/">24 tri: the command line</a></li></ol></li><li><h3>Module 9 · Errors, a program, what next</h3><p>Reading a compiler error, one small complete program, and the course to take next.</p><ol><li class="on" aria-current="page"><a href="/learn/reading-errors/">25 Reading compiler errors</a></li><li><a href="/learn/a-small-program/">26 A small complete program</a></li><li><a href="/learn/where-next/">27 Where to go next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/related-or-not/index.html b/apps/website/public/learn/related-or-not/index.html
index 421d6df2c1..de4bcdc9bc 100644
--- a/apps/website/public/learn/related-or-not/index.html
+++ b/apps/website/public/learn/related-or-not/index.html
@@ -448,5 +448,6 @@ <h1>Related or not</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a clock is</h3><p>One edge, one world: what shares a clock edge shares a world, the period and the jitter of a real edge, and where the clock enters a board.</p><ol><li><a href="/learn/one-edge-one-world/">01 One edge, one world</a></li><li><a href="/learn/period-and-jitter/">02 A period, an edge, and jitter</a></li><li><a href="/learn/where-the-clock-enters/">03 Where the clock enters</a></li></ol></li><li><h3>Module 2 · Clock trees</h3><p>Skew and insertion delay, the global buffer network, and the trap of gating a clock with logic.</p><ol><li><a href="/learn/skew-and-insertion/">04 Skew and insertion delay</a></li><li><a href="/learn/global-buffers/">05 The global buffer</a></li><li><a href="/learn/gated-clock-trap/">06 The gated-clock trap</a></li></ol></li><li><h3>Module 3 · PLL and MMCM</h3><p>Multiply and divide one clock into another, move its phase in steps of the VCO, and which clocks the analyzer treats as related.</p><ol><li><a href="/learn/multiply-and-divide/">07 Multiply and divide</a></li><li><a href="/learn/phase-and-lock/">08 Phase and lock</a></li><li class="on" aria-current="page"><a href="/learn/related-or-not/">09 Related or not</a></li></ol></li><li><h3>Module 4 · Resets</h3><p>Assert asynchronously, release synchronously: the three reset kinds, the release pipe, and the tree a reset grows.</p><ol><li><a href="/learn/sync-or-async/">10 Sync or async</a></li><li><a href="/learn/reset-release/">11 The release pipe</a></li><li><a href="/learn/reset-fanout/">12 Reset fanout</a></li></ol></li><li><h3>Module 5 · Metastability</h3><p>The setup-hold window, the mean time between failures in integer arithmetic, and the two flops that fix it.</p><ol><li><a href="/learn/setup-hold-window/">13 The setup-hold window</a></li><li><a href="/learn/mtbf/">14 Mean time between failures</a></li><li><a href="/learn/two-flop-sync/">15 The two-flop synchronizer</a></li></ol></li><li><h3>Module 6 · Crossing many bits</h3><p>Why a binary bus tears, why Gray code does not, and the handshake that moves a pulse between worlds.</p><ol><li><a href="/learn/why-binary-fails/">16 Why binary fails</a></li><li><a href="/learn/gray-code/">17 Gray code</a></li><li><a href="/learn/handshakes-and-pulses/">18 Handshakes and pulses</a></li></ol></li><li><h3>Module 7 · The asynchronous FIFO</h3><p>Pointers, flags and depth: the buffer that moves a stream between two clocks.</p><ol><li><a href="/learn/pointers-and-flags/">19 Pointers and flags</a></li><li><a href="/learn/depth-sizing/">20 Sizing the depth</a></li><li><a href="/learn/fifo-testbench/">21 The FIFO testbench</a></li></ol></li><li><h3>Module 8 · Constraints</h3><p>The lines that tell the analyzer what a clock is, which paths not to check, and what the pins must meet.</p><ol><li><a href="/learn/create-clock/">22 create_clock</a></li><li><a href="/learn/false-path-and-max-delay/">23 False paths and max delay</a></li><li><a href="/learn/io-timing/">24 IO timing</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>A CDC report, one crossing captured at the flip-flops, and the bitstream diff that closes the course.</p><ol><li><a href="/learn/cdc-report/">25 The CDC report</a></li><li><a href="/learn/capture-a-crossing/">26 Capture a crossing</a></li><li><a href="/learn/capstone-diff/">27 Capstone: change it and diff it</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/reset-fanout/index.html b/apps/website/public/learn/reset-fanout/index.html
index 359f31203f..e3b6698410 100644
--- a/apps/website/public/learn/reset-fanout/index.html
+++ b/apps/website/public/learn/reset-fanout/index.html
@@ -294,5 +294,6 @@ <h1>Reset fanout</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a clock is</h3><p>One edge, one world: what shares a clock edge shares a world, the period and the jitter of a real edge, and where the clock enters a board.</p><ol><li><a href="/learn/one-edge-one-world/">01 One edge, one world</a></li><li><a href="/learn/period-and-jitter/">02 A period, an edge, and jitter</a></li><li><a href="/learn/where-the-clock-enters/">03 Where the clock enters</a></li></ol></li><li><h3>Module 2 · Clock trees</h3><p>Skew and insertion delay, the global buffer network, and the trap of gating a clock with logic.</p><ol><li><a href="/learn/skew-and-insertion/">04 Skew and insertion delay</a></li><li><a href="/learn/global-buffers/">05 The global buffer</a></li><li><a href="/learn/gated-clock-trap/">06 The gated-clock trap</a></li></ol></li><li><h3>Module 3 · PLL and MMCM</h3><p>Multiply and divide one clock into another, move its phase in steps of the VCO, and which clocks the analyzer treats as related.</p><ol><li><a href="/learn/multiply-and-divide/">07 Multiply and divide</a></li><li><a href="/learn/phase-and-lock/">08 Phase and lock</a></li><li><a href="/learn/related-or-not/">09 Related or not</a></li></ol></li><li><h3>Module 4 · Resets</h3><p>Assert asynchronously, release synchronously: the three reset kinds, the release pipe, and the tree a reset grows.</p><ol><li><a href="/learn/sync-or-async/">10 Sync or async</a></li><li><a href="/learn/reset-release/">11 The release pipe</a></li><li class="on" aria-current="page"><a href="/learn/reset-fanout/">12 Reset fanout</a></li></ol></li><li><h3>Module 5 · Metastability</h3><p>The setup-hold window, the mean time between failures in integer arithmetic, and the two flops that fix it.</p><ol><li><a href="/learn/setup-hold-window/">13 The setup-hold window</a></li><li><a href="/learn/mtbf/">14 Mean time between failures</a></li><li><a href="/learn/two-flop-sync/">15 The two-flop synchronizer</a></li></ol></li><li><h3>Module 6 · Crossing many bits</h3><p>Why a binary bus tears, why Gray code does not, and the handshake that moves a pulse between worlds.</p><ol><li><a href="/learn/why-binary-fails/">16 Why binary fails</a></li><li><a href="/learn/gray-code/">17 Gray code</a></li><li><a href="/learn/handshakes-and-pulses/">18 Handshakes and pulses</a></li></ol></li><li><h3>Module 7 · The asynchronous FIFO</h3><p>Pointers, flags and depth: the buffer that moves a stream between two clocks.</p><ol><li><a href="/learn/pointers-and-flags/">19 Pointers and flags</a></li><li><a href="/learn/depth-sizing/">20 Sizing the depth</a></li><li><a href="/learn/fifo-testbench/">21 The FIFO testbench</a></li></ol></li><li><h3>Module 8 · Constraints</h3><p>The lines that tell the analyzer what a clock is, which paths not to check, and what the pins must meet.</p><ol><li><a href="/learn/create-clock/">22 create_clock</a></li><li><a href="/learn/false-path-and-max-delay/">23 False paths and max delay</a></li><li><a href="/learn/io-timing/">24 IO timing</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>A CDC report, one crossing captured at the flip-flops, and the bitstream diff that closes the course.</p><ol><li><a href="/learn/cdc-report/">25 The CDC report</a></li><li><a href="/learn/capture-a-crossing/">26 Capture a crossing</a></li><li><a href="/learn/capstone-diff/">27 Capstone: change it and diff it</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/reset-release/index.html b/apps/website/public/learn/reset-release/index.html
index 1d8eec79a7..d9bbad0638 100644
--- a/apps/website/public/learn/reset-release/index.html
+++ b/apps/website/public/learn/reset-release/index.html
@@ -329,5 +329,6 @@ <h1>The release pipe</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a clock is</h3><p>One edge, one world: what shares a clock edge shares a world, the period and the jitter of a real edge, and where the clock enters a board.</p><ol><li><a href="/learn/one-edge-one-world/">01 One edge, one world</a></li><li><a href="/learn/period-and-jitter/">02 A period, an edge, and jitter</a></li><li><a href="/learn/where-the-clock-enters/">03 Where the clock enters</a></li></ol></li><li><h3>Module 2 · Clock trees</h3><p>Skew and insertion delay, the global buffer network, and the trap of gating a clock with logic.</p><ol><li><a href="/learn/skew-and-insertion/">04 Skew and insertion delay</a></li><li><a href="/learn/global-buffers/">05 The global buffer</a></li><li><a href="/learn/gated-clock-trap/">06 The gated-clock trap</a></li></ol></li><li><h3>Module 3 · PLL and MMCM</h3><p>Multiply and divide one clock into another, move its phase in steps of the VCO, and which clocks the analyzer treats as related.</p><ol><li><a href="/learn/multiply-and-divide/">07 Multiply and divide</a></li><li><a href="/learn/phase-and-lock/">08 Phase and lock</a></li><li><a href="/learn/related-or-not/">09 Related or not</a></li></ol></li><li><h3>Module 4 · Resets</h3><p>Assert asynchronously, release synchronously: the three reset kinds, the release pipe, and the tree a reset grows.</p><ol><li><a href="/learn/sync-or-async/">10 Sync or async</a></li><li class="on" aria-current="page"><a href="/learn/reset-release/">11 The release pipe</a></li><li><a href="/learn/reset-fanout/">12 Reset fanout</a></li></ol></li><li><h3>Module 5 · Metastability</h3><p>The setup-hold window, the mean time between failures in integer arithmetic, and the two flops that fix it.</p><ol><li><a href="/learn/setup-hold-window/">13 The setup-hold window</a></li><li><a href="/learn/mtbf/">14 Mean time between failures</a></li><li><a href="/learn/two-flop-sync/">15 The two-flop synchronizer</a></li></ol></li><li><h3>Module 6 · Crossing many bits</h3><p>Why a binary bus tears, why Gray code does not, and the handshake that moves a pulse between worlds.</p><ol><li><a href="/learn/why-binary-fails/">16 Why binary fails</a></li><li><a href="/learn/gray-code/">17 Gray code</a></li><li><a href="/learn/handshakes-and-pulses/">18 Handshakes and pulses</a></li></ol></li><li><h3>Module 7 · The asynchronous FIFO</h3><p>Pointers, flags and depth: the buffer that moves a stream between two clocks.</p><ol><li><a href="/learn/pointers-and-flags/">19 Pointers and flags</a></li><li><a href="/learn/depth-sizing/">20 Sizing the depth</a></li><li><a href="/learn/fifo-testbench/">21 The FIFO testbench</a></li></ol></li><li><h3>Module 8 · Constraints</h3><p>The lines that tell the analyzer what a clock is, which paths not to check, and what the pins must meet.</p><ol><li><a href="/learn/create-clock/">22 create_clock</a></li><li><a href="/learn/false-path-and-max-delay/">23 False paths and max delay</a></li><li><a href="/learn/io-timing/">24 IO timing</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>A CDC report, one crossing captured at the flip-flops, and the bitstream diff that closes the course.</p><ol><li><a href="/learn/cdc-report/">25 The CDC report</a></li><li><a href="/learn/capture-a-crossing/">26 Capture a crossing</a></li><li><a href="/learn/capstone-diff/">27 Capstone: change it and diff it</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/rgmii-timing/index.html b/apps/website/public/learn/rgmii-timing/index.html
index 2e539625eb..bcc681048f 100644
--- a/apps/website/public/learn/rgmii-timing/index.html
+++ b/apps/website/public/learn/rgmii-timing/index.html
@@ -186,5 +186,6 @@ <h1>RGMII timing</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a bus is</h3><p>Why a bus exists at all: a conversation on wires, framed and addressed, and who is allowed to talk.</p><ol><li><a href="/learn/wires-that-talk/">01 Wires that talk</a></li><li><a href="/learn/framing-a-conversation/">02 Framing a conversation</a></li><li><a href="/learn/master-and-servant/">03 Master and servant</a></li></ol></li><li><h3>Module 2 · UART</h3><p>The two-wire bus with no clock: the frame, the divisor that sets the rate, and the status a driver polls.</p><ol><li><a href="/learn/the-frame/">04 The frame: start, data, stop</a></li><li><a href="/learn/baud-and-the-divisor/">05 Baud and the divisor</a></li><li><a href="/learn/status-and-fifos/">06 Status and FIFOs</a></li></ol></li><li><h3>Module 3 · SPI</h3><p>The clocked conversation: four modes, a prescaler ladder, and a chip select per servant.</p><ol><li><a href="/learn/modes-cpol-cpha/">07 Modes: CPOL and CPHA</a></li><li><a href="/learn/the-prescaler-ladder/">08 The prescaler ladder</a></li><li><a href="/learn/chip-select-and-widths/">09 Chip select and widths</a></li></ol></li><li><h3>Module 4 · APB</h3><p>The register bus: PSEL and PENABLE, strobes and wait states, and how many address bits a peripheral count costs.</p><ol><li><a href="/learn/psel-and-penable/">10 PSEL and PENABLE</a></li><li><a href="/learn/strobes-and-waits/">11 Strobes and waits</a></li><li><a href="/learn/address-decode/">12 Address decode</a></li></ol></li><li><h3>Module 5 · AXI4</h3><p>The five channels: address, data and response in both directions, lite or full, bursts and IDs.</p><ol><li><a href="/learn/five-channels/">13 Five channels</a></li><li><a href="/learn/lite-or-full/">14 Lite or full</a></li><li><a href="/learn/bursts-and-ids/">15 Bursts and IDs</a></li></ol></li><li><h3>Module 6 · Memory</h3><p>What sits on the far side of every bus: memory maps, port kinds, and latency that a wait state must cover.</p><ol><li><a href="/learn/memory-maps/">16 Memory maps</a></li><li><a href="/learn/ports-and-kinds/">17 Ports and kinds</a></li><li><a href="/learn/latency/">18 Latency</a></li></ol></li><li><h3>Module 7 · Bridges</h3><p>Why designs grow more than one bus, and the packet bridge that moves work between them.</p><ol><li><a href="/learn/why-bridges/">19 Why bridges</a></li><li><a href="/learn/the-packet-bridge/">20 The packet bridge</a></li><li><a href="/learn/everything-in-one-bench/">21 Everything in one bench</a></li></ol></li><li><h3>Module 8 · Ethernet</h3><p>Frames, the frame check sequence, RGMII timing, and the pre-registered steps of a real bring-up.</p><ol><li><a href="/learn/frames-and-the-fcs/">22 Frames and the FCS</a></li><li class="on" aria-current="page"><a href="/learn/rgmii-timing/">23 RGMII timing</a></li><li><a href="/learn/bring-up-on-the-bench/">24 Bring-up on the bench</a></li></ol></li><li><h3>Module 9 · The bench</h3><p>The discipline that guards real hardware: who holds the IO, taking and giving the lock, and what runs next.</p><ol><li><a href="/learn/who-holds-the-io/">25 Who holds the IO</a></li><li><a href="/learn/claim-and-release/">26 Claim and release</a></li><li><a href="/learn/capstone-what-runs-next/">27 Capstone: what runs next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/ripple-carry/index.html b/apps/website/public/learn/ripple-carry/index.html
index 5f5d313e44..06ab7a2825 100644
--- a/apps/website/public/learn/ripple-carry/index.html
+++ b/apps/website/public/learn/ripple-carry/index.html
@@ -159,5 +159,6 @@ <h1>Ripple carry</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Three values</h3><p>Why three, how balanced ternary writes every number without a sign, and what flipping and cutting trits do.</p><ol><li><a href="/learn/why-three/">01 Why three</a></li><li><a href="/learn/balanced-ternary/">02 Balanced ternary</a></li><li><a href="/learn/flip-and-cut/">03 Flip to negate, cut to round</a></li></ol></li><li><h3>Module 2 · Logic with unknown</h3><p>Kleene's three-valued gates, two gates binary has no twin for, and addition as a pair of tables.</p><ol><li><a href="/learn/kleene-logic/">04 Three-valued logic</a></li><li><a href="/learn/consensus-and-any/">05 Consensus and any</a></li><li><a href="/learn/sum-and-carry/">06 Sum and carry</a></li></ol></li><li><h3>Module 3 · Adding trits</h3><p>The half adder, the full adder and a carry that ripples left, one place at a time.</p><ol><li><a href="/learn/half-adder/">07 The half adder</a></li><li><a href="/learn/full-adder/">08 The full adder</a></li><li class="on" aria-current="page"><a href="/learn/ripple-carry/">09 Ripple carry</a></li></ol></li><li><h3>Module 4 · Multiplying without a multiplier</h3><p>Copy, drop or flip: a product by one trit, long multiplication, and a MAC that only adds.</p><ol><li><a href="/learn/times-a-trit/">10 Times one trit</a></li><li><a href="/learn/long-multiplication/">11 Long multiplication</a></li><li><a href="/learn/ternary-mac/">12 A ternary MAC</a></li></ol></li><li><h3>Module 5 · Trits in binary memory</h3><p>Two bits per trit, five trits per byte, and the bits a 27-trit word needs.</p><ol><li><a href="/learn/two-bits-per-trit/">13 Two bits per trit</a></li><li><a href="/learn/five-trits-a-byte/">14 Five trits in a byte</a></li><li><a href="/learn/a-word-of-27/">15 A word of 27 trits</a></li></ol></li><li><h3>Module 6 · The TRI-27 instruction word</h3><p>The 32-bit word the Trinity emulator decodes: its fields, its 47 opcodes and its 15-bit immediate.</p><ol><li><a href="/learn/the-instruction-word/">16 The instruction word</a></li><li><a href="/learn/the-opcode-map/">17 The opcode map</a></li><li><a href="/learn/the-immediate/">18 The immediate</a></li></ol></li><li><h3>Module 7 · A ternary machine</h3><p>An ALU built from this course's adders, a three-way jump, and a program you can step.</p><ol><li><a href="/learn/a-ternary-alu/">19 A ternary ALU</a></li><li><a href="/learn/three-way-jump/">20 A three-way jump</a></li><li><a href="/learn/run-a-program/">21 Run a program</a></li></ol></li><li><h3>Module 8 · Three answers</h3><p>Compare at the highest differing trit, find a number in thirds, sort with three-way compares.</p><ol><li><a href="/learn/compare-at-the-top/">22 Compare at the top</a></li><li><a href="/learn/questions-in-thirds/">23 Questions in thirds</a></li><li><a href="/learn/sort-by-three/">24 Sort with three-way compares</a></li></ol></li><li><h3>Module 9 · Ternary neurons</h3><p>Weights of -1, 0, +1: one neuron, a detector and a small layer, with no multiplier anywhere.</p><ol><li><a href="/learn/one-ternary-neuron/">25 One ternary neuron</a></li><li><a href="/learn/a-bar-detector/">26 A bar detector</a></li><li><a href="/learn/a-tiny-layer/">27 A tiny layer</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/rotating-by-shifts/index.html b/apps/website/public/learn/rotating-by-shifts/index.html
index bd6405d58b..07040c248c 100644
--- a/apps/website/public/learn/rotating-by-shifts/index.html
+++ b/apps/website/public/learn/rotating-by-shifts/index.html
@@ -439,5 +439,6 @@ <h1>Rotating by shifts</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Adding numbers</h3><p>Why addition is slow: the carry that walks up the word, the dedicated chain in every slice, prefix networks and carry-save trees.</p><ol><li><a href="/learn/the-carry-chain/">01 The carry chain</a></li><li><a href="/learn/prefix-adders/">02 Looking ahead: prefix adders</a></li><li><a href="/learn/carry-save/">03 Three numbers at once</a></li></ol></li><li><h3>Module 2 · Multiplying</h3><p>A product is a sum of shifted copies; Booth recoding halves them, and the DSP48E1 slice does the rest in one block.</p><ol><li><a href="/learn/shift-and-add/">04 Shift and add</a></li><li><a href="/learn/booths-trick/">05 Booth's trick</a></li><li><a href="/learn/the-dsp48e1-slice/">06 The DSP48E1 slice</a></li></ol></li><li><h3>Module 3 · Fixed point</h3><p>Where the binary point sits, what rounding does to a value and to its average, and what each bit of a quantizer buys.</p><ol><li><a href="/learn/q-formats/">07 Q formats</a></li><li><a href="/learn/rounding-and-saturation/">08 Rounding and saturation</a></li><li><a href="/learn/quantization-noise/">09 Quantization noise</a></li></ol></li><li><h3>Module 4 · Functions in hardware</h3><p>Sine, cosine, angle and length from shifts and adds, and when a table is the better answer.</p><ol><li class="on" aria-current="page"><a href="/learn/rotating-by-shifts/">10 Rotating by shifts</a></li><li><a href="/learn/angle-and-magnitude/">11 Angle and magnitude</a></li><li><a href="/learn/tables-and-interpolation/">12 Tables and interpolation</a></li></ol></li><li><h3>Module 5 · Signals and sampling</h3><p>What sampling does to a frequency, how an oscillator is built from an adder, and what a DFT bin measures.</p><ol><li><a href="/learn/sampling-and-aliasing/">13 Sampling and aliasing</a></li><li><a href="/learn/the-nco/">14 A numerically controlled oscillator</a></li><li><a href="/learn/bins-of-a-spectrum/">15 Bins of a spectrum</a></li></ol></li><li><h3>Module 6 · FIR filters</h3><p>The moving average, a windowed-sinc design with integer taps, and folding the taps onto DSP slices.</p><ol><li><a href="/learn/the-moving-average/">16 The moving average</a></li><li><a href="/learn/designing-taps/">17 Designing taps</a></li><li><a href="/learn/folding-taps-onto-slices/">18 Folding taps onto slices</a></li></ol></li><li><h3>Module 7 · Multirate</h3><p>Lowering the sample rate without folding noise in: decimation, the CIC filter and the polyphase form.</p><ol><li><a href="/learn/lowering-the-rate/">19 Lowering the rate</a></li><li><a href="/learn/the-cic-filter/">20 The CIC filter</a></li><li><a href="/learn/polyphase/">21 Polyphase</a></li></ol></li><li><h3>Module 8 · The FFT</h3><p>N log N instead of N^2: butterflies, the bit-reversed input order and the bits each stage adds.</p><ol><li><a href="/learn/the-butterfly/">22 The butterfly</a></li><li><a href="/learn/twiddles-and-bit-reversal/">23 Twiddles and bit reversal</a></li><li><a href="/learn/growth-through-the-stages/">24 Growth through the stages</a></li></ol></li><li><h3>Module 9 · On the bench</h3><p>From the arithmetic to the board: a correlator from the modem spec, a budget of slices and timing, and the capstone filter.</p><ol><li><a href="/learn/a-correlator-finds-the-frame/">25 A correlator finds the frame</a></li><li><a href="/learn/does-it-fit-and-close-timing/">26 Does it fit and close timing?</a></li><li><a href="/learn/capstone-a-filter-on-the-board/">27 Capstone: a filter on the board</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/rounding-and-saturation/index.html b/apps/website/public/learn/rounding-and-saturation/index.html
index 161ba9389a..cde008ead4 100644
--- a/apps/website/public/learn/rounding-and-saturation/index.html
+++ b/apps/website/public/learn/rounding-and-saturation/index.html
@@ -346,5 +346,6 @@ <h1>Rounding and saturation</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Adding numbers</h3><p>Why addition is slow: the carry that walks up the word, the dedicated chain in every slice, prefix networks and carry-save trees.</p><ol><li><a href="/learn/the-carry-chain/">01 The carry chain</a></li><li><a href="/learn/prefix-adders/">02 Looking ahead: prefix adders</a></li><li><a href="/learn/carry-save/">03 Three numbers at once</a></li></ol></li><li><h3>Module 2 · Multiplying</h3><p>A product is a sum of shifted copies; Booth recoding halves them, and the DSP48E1 slice does the rest in one block.</p><ol><li><a href="/learn/shift-and-add/">04 Shift and add</a></li><li><a href="/learn/booths-trick/">05 Booth's trick</a></li><li><a href="/learn/the-dsp48e1-slice/">06 The DSP48E1 slice</a></li></ol></li><li><h3>Module 3 · Fixed point</h3><p>Where the binary point sits, what rounding does to a value and to its average, and what each bit of a quantizer buys.</p><ol><li><a href="/learn/q-formats/">07 Q formats</a></li><li class="on" aria-current="page"><a href="/learn/rounding-and-saturation/">08 Rounding and saturation</a></li><li><a href="/learn/quantization-noise/">09 Quantization noise</a></li></ol></li><li><h3>Module 4 · Functions in hardware</h3><p>Sine, cosine, angle and length from shifts and adds, and when a table is the better answer.</p><ol><li><a href="/learn/rotating-by-shifts/">10 Rotating by shifts</a></li><li><a href="/learn/angle-and-magnitude/">11 Angle and magnitude</a></li><li><a href="/learn/tables-and-interpolation/">12 Tables and interpolation</a></li></ol></li><li><h3>Module 5 · Signals and sampling</h3><p>What sampling does to a frequency, how an oscillator is built from an adder, and what a DFT bin measures.</p><ol><li><a href="/learn/sampling-and-aliasing/">13 Sampling and aliasing</a></li><li><a href="/learn/the-nco/">14 A numerically controlled oscillator</a></li><li><a href="/learn/bins-of-a-spectrum/">15 Bins of a spectrum</a></li></ol></li><li><h3>Module 6 · FIR filters</h3><p>The moving average, a windowed-sinc design with integer taps, and folding the taps onto DSP slices.</p><ol><li><a href="/learn/the-moving-average/">16 The moving average</a></li><li><a href="/learn/designing-taps/">17 Designing taps</a></li><li><a href="/learn/folding-taps-onto-slices/">18 Folding taps onto slices</a></li></ol></li><li><h3>Module 7 · Multirate</h3><p>Lowering the sample rate without folding noise in: decimation, the CIC filter and the polyphase form.</p><ol><li><a href="/learn/lowering-the-rate/">19 Lowering the rate</a></li><li><a href="/learn/the-cic-filter/">20 The CIC filter</a></li><li><a href="/learn/polyphase/">21 Polyphase</a></li></ol></li><li><h3>Module 8 · The FFT</h3><p>N log N instead of N^2: butterflies, the bit-reversed input order and the bits each stage adds.</p><ol><li><a href="/learn/the-butterfly/">22 The butterfly</a></li><li><a href="/learn/twiddles-and-bit-reversal/">23 Twiddles and bit reversal</a></li><li><a href="/learn/growth-through-the-stages/">24 Growth through the stages</a></li></ol></li><li><h3>Module 9 · On the bench</h3><p>From the arithmetic to the board: a correlator from the modem spec, a budget of slices and timing, and the capstone filter.</p><ol><li><a href="/learn/a-correlator-finds-the-frame/">25 A correlator finds the frame</a></li><li><a href="/learn/does-it-fit-and-close-timing/">26 Does it fit and close timing?</a></li><li><a href="/learn/capstone-a-filter-on-the-board/">27 Capstone: a filter on the board</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/run-a-program/index.html b/apps/website/public/learn/run-a-program/index.html
index 5c827e704f..1d42d4a3e4 100644
--- a/apps/website/public/learn/run-a-program/index.html
+++ b/apps/website/public/learn/run-a-program/index.html
@@ -278,5 +278,6 @@ <h1>Run a program</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Three values</h3><p>Why three, how balanced ternary writes every number without a sign, and what flipping and cutting trits do.</p><ol><li><a href="/learn/why-three/">01 Why three</a></li><li><a href="/learn/balanced-ternary/">02 Balanced ternary</a></li><li><a href="/learn/flip-and-cut/">03 Flip to negate, cut to round</a></li></ol></li><li><h3>Module 2 · Logic with unknown</h3><p>Kleene's three-valued gates, two gates binary has no twin for, and addition as a pair of tables.</p><ol><li><a href="/learn/kleene-logic/">04 Three-valued logic</a></li><li><a href="/learn/consensus-and-any/">05 Consensus and any</a></li><li><a href="/learn/sum-and-carry/">06 Sum and carry</a></li></ol></li><li><h3>Module 3 · Adding trits</h3><p>The half adder, the full adder and a carry that ripples left, one place at a time.</p><ol><li><a href="/learn/half-adder/">07 The half adder</a></li><li><a href="/learn/full-adder/">08 The full adder</a></li><li><a href="/learn/ripple-carry/">09 Ripple carry</a></li></ol></li><li><h3>Module 4 · Multiplying without a multiplier</h3><p>Copy, drop or flip: a product by one trit, long multiplication, and a MAC that only adds.</p><ol><li><a href="/learn/times-a-trit/">10 Times one trit</a></li><li><a href="/learn/long-multiplication/">11 Long multiplication</a></li><li><a href="/learn/ternary-mac/">12 A ternary MAC</a></li></ol></li><li><h3>Module 5 · Trits in binary memory</h3><p>Two bits per trit, five trits per byte, and the bits a 27-trit word needs.</p><ol><li><a href="/learn/two-bits-per-trit/">13 Two bits per trit</a></li><li><a href="/learn/five-trits-a-byte/">14 Five trits in a byte</a></li><li><a href="/learn/a-word-of-27/">15 A word of 27 trits</a></li></ol></li><li><h3>Module 6 · The TRI-27 instruction word</h3><p>The 32-bit word the Trinity emulator decodes: its fields, its 47 opcodes and its 15-bit immediate.</p><ol><li><a href="/learn/the-instruction-word/">16 The instruction word</a></li><li><a href="/learn/the-opcode-map/">17 The opcode map</a></li><li><a href="/learn/the-immediate/">18 The immediate</a></li></ol></li><li><h3>Module 7 · A ternary machine</h3><p>An ALU built from this course's adders, a three-way jump, and a program you can step.</p><ol><li><a href="/learn/a-ternary-alu/">19 A ternary ALU</a></li><li><a href="/learn/three-way-jump/">20 A three-way jump</a></li><li class="on" aria-current="page"><a href="/learn/run-a-program/">21 Run a program</a></li></ol></li><li><h3>Module 8 · Three answers</h3><p>Compare at the highest differing trit, find a number in thirds, sort with three-way compares.</p><ol><li><a href="/learn/compare-at-the-top/">22 Compare at the top</a></li><li><a href="/learn/questions-in-thirds/">23 Questions in thirds</a></li><li><a href="/learn/sort-by-three/">24 Sort with three-way compares</a></li></ol></li><li><h3>Module 9 · Ternary neurons</h3><p>Weights of -1, 0, +1: one neuron, a detector and a small layer, with no multiplier anywhere.</p><ol><li><a href="/learn/one-ternary-neuron/">25 One ternary neuron</a></li><li><a href="/learn/a-bar-detector/">26 A bar detector</a></li><li><a href="/learn/a-tiny-layer/">27 A tiny layer</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/same-bits-on-the-board/index.html b/apps/website/public/learn/same-bits-on-the-board/index.html
index 8c473d1b8f..6145cdac12 100644
--- a/apps/website/public/learn/same-bits-on-the-board/index.html
+++ b/apps/website/public/learn/same-bits-on-the-board/index.html
@@ -110,5 +110,6 @@ <h1>The same bits, on the board</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The chip</h3><p>What an FPGA is, which chip we use, and how its pins meet the board.</p><ol><li><a href="/learn/fpga-is-a-grid/">01 An FPGA is a grid you configure</a></li><li><a href="/learn/which-chip/">02 Which chip is this?</a></li><li><a href="/learn/pins-and-package/">03 Pins, balls and the package</a></li></ol></li><li><h3>Module 2 · Numbers in hardware</h3><p>Bits, trits and number formats, and what arithmetic costs in logic.</p><ol><li><a href="/learn/bits-and-trits/">04 Bits, trits and GF16</a></li><li><a href="/learn/what-arithmetic-costs/">05 What arithmetic costs</a></li><li><a href="/learn/how-much-fits/">06 How much fits on the chip</a></li></ol></li><li><h3>Module 3 · Your t27 program</h3><p>Write a spec, test it, and see why compiling is not the same as being right.</p><ol><li><a href="/learn/hello-t27/">07 Hello, t27</a></li><li><a href="/learn/tests-are-the-spec/">08 Tests are part of the spec</a></li><li><a href="/learn/compiles-is-not-correct/">09 Compiles is not the same as correct</a></li></ol></li><li><h3>Module 4 · Inside t27c</h3><p>How the compiler reads a spec and what it writes, including the native t27b.</p><ol><li><a href="/learn/compiler-stages/">10 The stages of t27c</a></li><li><a href="/learn/t27c-on-a-real-machine/">11 t27c on a real machine</a></li><li><a href="/learn/seven-backends-and-t27b/">12 Seven backends, and t27b</a></li></ol></li><li><h3>Module 5 · From spec to hardware</h3><p>The Verilog t27c writes, the cells it becomes, and what one LUT does.</p><ol><li><a href="/learn/spec-to-verilog/">13 From a spec to Verilog</a></li><li><a href="/learn/count-the-cells/">14 Count the cells</a></li><li><a href="/learn/one-lut-up-close/">15 One LUT, up close</a></li></ol></li><li><h3>Module 6 · Reading synthesis</h3><p>What yosys reports about your design, and which warnings matter.</p><ol><li><a href="/learn/synthesis-warnings/">16 Synthesis warnings that matter</a></li><li><a href="/learn/state-machines/">17 State machines</a></li><li><a href="/learn/where-the-luts-go/">18 Where the LUTs go</a></li></ol></li><li><h3>Module 7 · Place, route, timing</h3><p>Where the cells land on the die, and whether the clock is met.</p><ol><li><a href="/learn/placement/">19 Placement: where the cells land</a></li><li><a href="/learn/logic-depth/">20 Logic depth</a></li><li><a href="/learn/timing-slack/">21 Timing and slack</a></li></ol></li><li><h3>Module 8 · The bitstream</h3><p>How a routed design becomes the bits the chip loads, with no Vivado.</p><ol><li><a href="/learn/fasm-to-frames/">22 From FASM to frames</a></li><li><a href="/learn/a-build-receipt/">23 A build, itemised</a></li><li><a href="/learn/read-a-bitstream/">24 Read a bitstream</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>Ask the chip who it is, load the bits, and check them.</p><ol><li><a href="/learn/jtag-asks-the-chip/">25 JTAG asks the chip</a></li><li class="on" aria-current="page"><a href="/learn/same-bits-on-the-board/">26 The same bits, on the board</a></li><li><a href="/learn/capstone/">27 Capstone: out-guess yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/same-neuron-new-weights/index.html b/apps/website/public/learn/same-neuron-new-weights/index.html
index 93ebe6448b..09adcb10ad 100644
--- a/apps/website/public/learn/same-neuron-new-weights/index.html
+++ b/apps/website/public/learn/same-neuron-new-weights/index.html
@@ -135,5 +135,6 @@ <h1>Same neuron, new weights</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Lab: our own research</h3><p>A number format of our own, an honest scoreboard, and a model's tables multiplied on the board.</p><ol><li><a href="/learn/t27-number-format/">01 T27: a number format of our own</a></li><li><a href="/learn/an-honest-scoreboard/">02 An honest scoreboard</a></li><li><a href="/learn/a-model-on-the-board/">03 A model's tables on the board</a></li></ol></li><li><h3>Module 2 · AI numbers: the MX block</h3><p>How AI chips keep weights in a few bits: one shared scale per block, the scale byte itself, and what one outlier does to its neighbours.</p><ol><li><a href="/learn/one-scale-per-block/">04 One scale for a block of weights</a></li><li><a href="/learn/scale-byte-native/">05 The scale byte on a real machine</a></li><li><a href="/learn/one-outlier/">06 One outlier, many zeros</a></li></ol></li><li><h3>Module 3 · Ternary weights</h3><p>Weights that are only minus, zero or plus a scale, the five rules a ternary alphabet must pass, and a test pass that checked nothing.</p><ol><li><a href="/learn/phi-scaled-ternary/">07 Weights of minus phi, zero and plus phi</a></li><li><a href="/learn/a-pass-that-checked-nothing/">08 A pass that checked nothing</a></li><li><a href="/learn/five-rules-for-weights/">09 Five rules for a weight alphabet</a></li></ol></li><li><h3>Module 4 · The Ternary Network Float</h3><p>A rule the compiler enforces before any test runs, and a 17-bit float whose exponent is four balanced trits.</p><ol><li><a href="/learn/blocked-is-not-failed/">10 Blocked is not failed</a></li><li><a href="/learn/four-trit-exponent/">11 A float whose exponent is four trits</a></li><li><a href="/learn/negate-twice/">12 Negate twice, get the number back</a></li></ol></li><li><h3>Module 5 · Arithmetic on signed numbers</h3><p>Multiply two signed numbers, add them when their signs differ, and do both at once in a multiply-accumulate.</p><ol><li><a href="/learn/a-sign-is-one-bit/">13 A sign is one bit</a></li><li><a href="/learn/opposite-signs-subtract/">14 Opposite signs subtract</a></li><li><a href="/learn/multiply-then-add/">15 Multiply, then add</a></li></ol></li><li><h3>Module 6 · Parts of a neuron</h3><p>A ReLU that bends at zero, a power of two for softmax, and an argmax that names the answer.</p><ol><li><a href="/learn/a-bend-at-zero/">16 A bend at zero</a></li><li><a href="/learn/two-to-the-x/">17 Two to the x</a></li><li><a href="/learn/pick-the-largest/">18 Pick the largest</a></li></ol></li><li><h3>Module 7 · Learning from a mistake</h3><p>A loss that prices a wrong guess in bits, one step that moves a weight against its gradient, and the hidden layer that XOR needs.</p><ol><li><a href="/learn/a-loss-in-bits/">19 A loss in bits</a></li><li><a href="/learn/one-step-downhill/">20 One step downhill</a></li><li><a href="/learn/xor-needs-a-bend/">21 XOR needs a bend</a></li></ol></li><li><h3>Module 8 · BitNet: ternary networks</h3><p>A threshold that squeezes a sum back to three values, one neuron that becomes a different function when its weights change, and a neuron that reads its inputs 27 trits at a time.</p><ol><li><a href="/learn/back-to-three-values/">22 Back to three values</a></li><li class="on" aria-current="page"><a href="/learn/same-neuron-new-weights/">23 Same neuron, new weights</a></li><li><a href="/learn/a-neuron-in-chunks/">24 A neuron in chunks</a></li></ol></li><li><h3>Module 9 · The ternary MAC as a chip</h3><p>The 27-trit dot product as wires with no register, the same sum added into a register on every clock, and a small whole network to close the course.</p><ol><li><a href="/learn/a-dot-product-in-wires/">25 A dot product made of wires</a></li><li><a href="/learn/add-it-up-every-clock/">26 Add it up on every clock</a></li><li><a href="/learn/a-whole-network/">27 A whole network</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/sampling-and-aliasing/index.html b/apps/website/public/learn/sampling-and-aliasing/index.html
index e859e12ce4..af750d2ea3 100644
--- a/apps/website/public/learn/sampling-and-aliasing/index.html
+++ b/apps/website/public/learn/sampling-and-aliasing/index.html
@@ -1122,5 +1122,6 @@ <h1>Sampling and aliasing</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Adding numbers</h3><p>Why addition is slow: the carry that walks up the word, the dedicated chain in every slice, prefix networks and carry-save trees.</p><ol><li><a href="/learn/the-carry-chain/">01 The carry chain</a></li><li><a href="/learn/prefix-adders/">02 Looking ahead: prefix adders</a></li><li><a href="/learn/carry-save/">03 Three numbers at once</a></li></ol></li><li><h3>Module 2 · Multiplying</h3><p>A product is a sum of shifted copies; Booth recoding halves them, and the DSP48E1 slice does the rest in one block.</p><ol><li><a href="/learn/shift-and-add/">04 Shift and add</a></li><li><a href="/learn/booths-trick/">05 Booth's trick</a></li><li><a href="/learn/the-dsp48e1-slice/">06 The DSP48E1 slice</a></li></ol></li><li><h3>Module 3 · Fixed point</h3><p>Where the binary point sits, what rounding does to a value and to its average, and what each bit of a quantizer buys.</p><ol><li><a href="/learn/q-formats/">07 Q formats</a></li><li><a href="/learn/rounding-and-saturation/">08 Rounding and saturation</a></li><li><a href="/learn/quantization-noise/">09 Quantization noise</a></li></ol></li><li><h3>Module 4 · Functions in hardware</h3><p>Sine, cosine, angle and length from shifts and adds, and when a table is the better answer.</p><ol><li><a href="/learn/rotating-by-shifts/">10 Rotating by shifts</a></li><li><a href="/learn/angle-and-magnitude/">11 Angle and magnitude</a></li><li><a href="/learn/tables-and-interpolation/">12 Tables and interpolation</a></li></ol></li><li><h3>Module 5 · Signals and sampling</h3><p>What sampling does to a frequency, how an oscillator is built from an adder, and what a DFT bin measures.</p><ol><li class="on" aria-current="page"><a href="/learn/sampling-and-aliasing/">13 Sampling and aliasing</a></li><li><a href="/learn/the-nco/">14 A numerically controlled oscillator</a></li><li><a href="/learn/bins-of-a-spectrum/">15 Bins of a spectrum</a></li></ol></li><li><h3>Module 6 · FIR filters</h3><p>The moving average, a windowed-sinc design with integer taps, and folding the taps onto DSP slices.</p><ol><li><a href="/learn/the-moving-average/">16 The moving average</a></li><li><a href="/learn/designing-taps/">17 Designing taps</a></li><li><a href="/learn/folding-taps-onto-slices/">18 Folding taps onto slices</a></li></ol></li><li><h3>Module 7 · Multirate</h3><p>Lowering the sample rate without folding noise in: decimation, the CIC filter and the polyphase form.</p><ol><li><a href="/learn/lowering-the-rate/">19 Lowering the rate</a></li><li><a href="/learn/the-cic-filter/">20 The CIC filter</a></li><li><a href="/learn/polyphase/">21 Polyphase</a></li></ol></li><li><h3>Module 8 · The FFT</h3><p>N log N instead of N^2: butterflies, the bit-reversed input order and the bits each stage adds.</p><ol><li><a href="/learn/the-butterfly/">22 The butterfly</a></li><li><a href="/learn/twiddles-and-bit-reversal/">23 Twiddles and bit reversal</a></li><li><a href="/learn/growth-through-the-stages/">24 Growth through the stages</a></li></ol></li><li><h3>Module 9 · On the bench</h3><p>From the arithmetic to the board: a correlator from the modem spec, a budget of slices and timing, and the capstone filter.</p><ol><li><a href="/learn/a-correlator-finds-the-frame/">25 A correlator finds the frame</a></li><li><a href="/learn/does-it-fit-and-close-timing/">26 Does it fit and close timing?</a></li><li><a href="/learn/capstone-a-filter-on-the-board/">27 Capstone: a filter on the board</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/scale-byte-native/index.html b/apps/website/public/learn/scale-byte-native/index.html
index 72bff30f77..e206cb9612 100644
--- a/apps/website/public/learn/scale-byte-native/index.html
+++ b/apps/website/public/learn/scale-byte-native/index.html
@@ -336,5 +336,6 @@ <h1>The scale byte on a real machine</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Lab: our own research</h3><p>A number format of our own, an honest scoreboard, and a model's tables multiplied on the board.</p><ol><li><a href="/learn/t27-number-format/">01 T27: a number format of our own</a></li><li><a href="/learn/an-honest-scoreboard/">02 An honest scoreboard</a></li><li><a href="/learn/a-model-on-the-board/">03 A model's tables on the board</a></li></ol></li><li><h3>Module 2 · AI numbers: the MX block</h3><p>How AI chips keep weights in a few bits: one shared scale per block, the scale byte itself, and what one outlier does to its neighbours.</p><ol><li><a href="/learn/one-scale-per-block/">04 One scale for a block of weights</a></li><li class="on" aria-current="page"><a href="/learn/scale-byte-native/">05 The scale byte on a real machine</a></li><li><a href="/learn/one-outlier/">06 One outlier, many zeros</a></li></ol></li><li><h3>Module 3 · Ternary weights</h3><p>Weights that are only minus, zero or plus a scale, the five rules a ternary alphabet must pass, and a test pass that checked nothing.</p><ol><li><a href="/learn/phi-scaled-ternary/">07 Weights of minus phi, zero and plus phi</a></li><li><a href="/learn/a-pass-that-checked-nothing/">08 A pass that checked nothing</a></li><li><a href="/learn/five-rules-for-weights/">09 Five rules for a weight alphabet</a></li></ol></li><li><h3>Module 4 · The Ternary Network Float</h3><p>A rule the compiler enforces before any test runs, and a 17-bit float whose exponent is four balanced trits.</p><ol><li><a href="/learn/blocked-is-not-failed/">10 Blocked is not failed</a></li><li><a href="/learn/four-trit-exponent/">11 A float whose exponent is four trits</a></li><li><a href="/learn/negate-twice/">12 Negate twice, get the number back</a></li></ol></li><li><h3>Module 5 · Arithmetic on signed numbers</h3><p>Multiply two signed numbers, add them when their signs differ, and do both at once in a multiply-accumulate.</p><ol><li><a href="/learn/a-sign-is-one-bit/">13 A sign is one bit</a></li><li><a href="/learn/opposite-signs-subtract/">14 Opposite signs subtract</a></li><li><a href="/learn/multiply-then-add/">15 Multiply, then add</a></li></ol></li><li><h3>Module 6 · Parts of a neuron</h3><p>A ReLU that bends at zero, a power of two for softmax, and an argmax that names the answer.</p><ol><li><a href="/learn/a-bend-at-zero/">16 A bend at zero</a></li><li><a href="/learn/two-to-the-x/">17 Two to the x</a></li><li><a href="/learn/pick-the-largest/">18 Pick the largest</a></li></ol></li><li><h3>Module 7 · Learning from a mistake</h3><p>A loss that prices a wrong guess in bits, one step that moves a weight against its gradient, and the hidden layer that XOR needs.</p><ol><li><a href="/learn/a-loss-in-bits/">19 A loss in bits</a></li><li><a href="/learn/one-step-downhill/">20 One step downhill</a></li><li><a href="/learn/xor-needs-a-bend/">21 XOR needs a bend</a></li></ol></li><li><h3>Module 8 · BitNet: ternary networks</h3><p>A threshold that squeezes a sum back to three values, one neuron that becomes a different function when its weights change, and a neuron that reads its inputs 27 trits at a time.</p><ol><li><a href="/learn/back-to-three-values/">22 Back to three values</a></li><li><a href="/learn/same-neuron-new-weights/">23 Same neuron, new weights</a></li><li><a href="/learn/a-neuron-in-chunks/">24 A neuron in chunks</a></li></ol></li><li><h3>Module 9 · The ternary MAC as a chip</h3><p>The 27-trit dot product as wires with no register, the same sum added into a register on every clock, and a small whole network to close the course.</p><ol><li><a href="/learn/a-dot-product-in-wires/">25 A dot product made of wires</a></li><li><a href="/learn/add-it-up-every-clock/">26 Add it up on every clock</a></li><li><a href="/learn/a-whole-network/">27 A whole network</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/self-checking/index.html b/apps/website/public/learn/self-checking/index.html
index 1f966f15b8..bde21194c9 100644
--- a/apps/website/public/learn/self-checking/index.html
+++ b/apps/website/public/learn/self-checking/index.html
@@ -209,5 +209,6 @@ <h1>Self-checking, not staring</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Why verify</h3><p>Designs that compile and are wrong, the model that decides, and the plan written before the code.</p><ol><li><a href="/learn/bugs-that-compile/">01 Bugs that compile</a></li><li><a href="/learn/golden-model/">02 The golden model</a></li><li><a href="/learn/plan-before-code/">03 Plan before code</a></li></ol></li><li><h3>Module 2 · Testbenches</h3><p>Stimulus, checks and a verdict, written as one spec beside the design it judges.</p><ol><li><a href="/learn/anatomy-of-a-tb/">04 Anatomy of a testbench</a></li><li class="on" aria-current="page"><a href="/learn/self-checking/">05 Self-checking, not staring</a></li><li><a href="/learn/directed-vs-random/">06 Directed versus random</a></li></ol></li><li><h3>Module 3 · Waveforms</h3><p>A trace of every signal, read the way a hardware engineer reads it, and two runs compared.</p><ol><li><a href="/learn/vcd-format/">07 The VCD format</a></li><li><a href="/learn/debug-with-waves/">08 Debug with waves</a></li><li><a href="/learn/compare-traces/">09 Compare two traces</a></li></ol></li><li><h3>Module 4 · Conformance vectors</h3><p>Cases with the answer written beside them, kept where the compiler can reach them.</p><ol><li><a href="/learn/what-a-vector-is/">10 What a vector is</a></li><li><a href="/learn/golden-ruler-vectors/">11 Golden ruler vectors</a></li><li><a href="/learn/end-to-end-scenarios/">12 End-to-end scenarios</a></li></ol></li><li><h3>Module 5 · Cosimulation</h3><p>Spec, simulator and board agreeing on the bench Artix-7 XC7A200T, and what to do when they do not.</p><ol><li><a href="/learn/three-way-agreement/">13 Three-way agreement</a></li><li><a href="/learn/simulator-internals/">14 Inside the simulator</a></li><li><a href="/learn/when-they-disagree/">15 When they disagree</a></li></ol></li><li><h3>Module 6 · Coverage</h3><p>What the tests touched: lines, toggles, states, and what that number hides.</p><ol><li><a href="/learn/line-and-toggle/">16 Line and toggle</a></li><li><a href="/learn/fsm-coverage/">17 FSM coverage</a></li><li><a href="/learn/coverage-lies/">18 Coverage lies</a></li></ol></li><li><h3>Module 7 · Formal</h3><p>Assertions that hold every cycle, bounded search for a counterexample, and why a proof needs induction.</p><ol><li><a href="/learn/assertions/">19 Assertions</a></li><li><a href="/learn/bounded-model-checking/">20 Bounded model checking</a></li><li><a href="/learn/induction/">21 Induction</a></li></ol></li><li><h3>Module 8 · Mutation</h3><p>Break the design on purpose and count what the tests catch.</p><ol><li><a href="/learn/kill-the-mutant/">22 Kill the mutant</a></li><li><a href="/learn/surviving-mutants/">23 Surviving mutants</a></li><li><a href="/learn/test-strength/">24 Test strength</a></li></ol></li><li><h3>Module 9 · Sign-off</h3><p>One command, every receipt, a clean verdict you can show.</p><ol><li><a href="/learn/build-verify/">25 Build and verify</a></li><li><a href="/learn/board-selftest/">26 The board self-tests</a></li><li><a href="/learn/verify-capstone/">27 Capstone: verify a module whole</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/setup-hold-window/index.html b/apps/website/public/learn/setup-hold-window/index.html
index f08c3180cf..4ccbf90202 100644
--- a/apps/website/public/learn/setup-hold-window/index.html
+++ b/apps/website/public/learn/setup-hold-window/index.html
@@ -300,5 +300,6 @@ <h1>The setup-hold window</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a clock is</h3><p>One edge, one world: what shares a clock edge shares a world, the period and the jitter of a real edge, and where the clock enters a board.</p><ol><li><a href="/learn/one-edge-one-world/">01 One edge, one world</a></li><li><a href="/learn/period-and-jitter/">02 A period, an edge, and jitter</a></li><li><a href="/learn/where-the-clock-enters/">03 Where the clock enters</a></li></ol></li><li><h3>Module 2 · Clock trees</h3><p>Skew and insertion delay, the global buffer network, and the trap of gating a clock with logic.</p><ol><li><a href="/learn/skew-and-insertion/">04 Skew and insertion delay</a></li><li><a href="/learn/global-buffers/">05 The global buffer</a></li><li><a href="/learn/gated-clock-trap/">06 The gated-clock trap</a></li></ol></li><li><h3>Module 3 · PLL and MMCM</h3><p>Multiply and divide one clock into another, move its phase in steps of the VCO, and which clocks the analyzer treats as related.</p><ol><li><a href="/learn/multiply-and-divide/">07 Multiply and divide</a></li><li><a href="/learn/phase-and-lock/">08 Phase and lock</a></li><li><a href="/learn/related-or-not/">09 Related or not</a></li></ol></li><li><h3>Module 4 · Resets</h3><p>Assert asynchronously, release synchronously: the three reset kinds, the release pipe, and the tree a reset grows.</p><ol><li><a href="/learn/sync-or-async/">10 Sync or async</a></li><li><a href="/learn/reset-release/">11 The release pipe</a></li><li><a href="/learn/reset-fanout/">12 Reset fanout</a></li></ol></li><li><h3>Module 5 · Metastability</h3><p>The setup-hold window, the mean time between failures in integer arithmetic, and the two flops that fix it.</p><ol><li class="on" aria-current="page"><a href="/learn/setup-hold-window/">13 The setup-hold window</a></li><li><a href="/learn/mtbf/">14 Mean time between failures</a></li><li><a href="/learn/two-flop-sync/">15 The two-flop synchronizer</a></li></ol></li><li><h3>Module 6 · Crossing many bits</h3><p>Why a binary bus tears, why Gray code does not, and the handshake that moves a pulse between worlds.</p><ol><li><a href="/learn/why-binary-fails/">16 Why binary fails</a></li><li><a href="/learn/gray-code/">17 Gray code</a></li><li><a href="/learn/handshakes-and-pulses/">18 Handshakes and pulses</a></li></ol></li><li><h3>Module 7 · The asynchronous FIFO</h3><p>Pointers, flags and depth: the buffer that moves a stream between two clocks.</p><ol><li><a href="/learn/pointers-and-flags/">19 Pointers and flags</a></li><li><a href="/learn/depth-sizing/">20 Sizing the depth</a></li><li><a href="/learn/fifo-testbench/">21 The FIFO testbench</a></li></ol></li><li><h3>Module 8 · Constraints</h3><p>The lines that tell the analyzer what a clock is, which paths not to check, and what the pins must meet.</p><ol><li><a href="/learn/create-clock/">22 create_clock</a></li><li><a href="/learn/false-path-and-max-delay/">23 False paths and max delay</a></li><li><a href="/learn/io-timing/">24 IO timing</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>A CDC report, one crossing captured at the flip-flops, and the bitstream diff that closes the course.</p><ol><li><a href="/learn/cdc-report/">25 The CDC report</a></li><li><a href="/learn/capture-a-crossing/">26 Capture a crossing</a></li><li><a href="/learn/capstone-diff/">27 Capstone: change it and diff it</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/seven-backends-and-t27b/index.html b/apps/website/public/learn/seven-backends-and-t27b/index.html
index ac209c160f..76614bf735 100644
--- a/apps/website/public/learn/seven-backends-and-t27b/index.html
+++ b/apps/website/public/learn/seven-backends-and-t27b/index.html
@@ -159,5 +159,6 @@ <h1>Seven backends, and t27b</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The chip</h3><p>What an FPGA is, which chip we use, and how its pins meet the board.</p><ol><li><a href="/learn/fpga-is-a-grid/">01 An FPGA is a grid you configure</a></li><li><a href="/learn/which-chip/">02 Which chip is this?</a></li><li><a href="/learn/pins-and-package/">03 Pins, balls and the package</a></li></ol></li><li><h3>Module 2 · Numbers in hardware</h3><p>Bits, trits and number formats, and what arithmetic costs in logic.</p><ol><li><a href="/learn/bits-and-trits/">04 Bits, trits and GF16</a></li><li><a href="/learn/what-arithmetic-costs/">05 What arithmetic costs</a></li><li><a href="/learn/how-much-fits/">06 How much fits on the chip</a></li></ol></li><li><h3>Module 3 · Your t27 program</h3><p>Write a spec, test it, and see why compiling is not the same as being right.</p><ol><li><a href="/learn/hello-t27/">07 Hello, t27</a></li><li><a href="/learn/tests-are-the-spec/">08 Tests are part of the spec</a></li><li><a href="/learn/compiles-is-not-correct/">09 Compiles is not the same as correct</a></li></ol></li><li><h3>Module 4 · Inside t27c</h3><p>How the compiler reads a spec and what it writes, including the native t27b.</p><ol><li><a href="/learn/compiler-stages/">10 The stages of t27c</a></li><li><a href="/learn/t27c-on-a-real-machine/">11 t27c on a real machine</a></li><li class="on" aria-current="page"><a href="/learn/seven-backends-and-t27b/">12 Seven backends, and t27b</a></li></ol></li><li><h3>Module 5 · From spec to hardware</h3><p>The Verilog t27c writes, the cells it becomes, and what one LUT does.</p><ol><li><a href="/learn/spec-to-verilog/">13 From a spec to Verilog</a></li><li><a href="/learn/count-the-cells/">14 Count the cells</a></li><li><a href="/learn/one-lut-up-close/">15 One LUT, up close</a></li></ol></li><li><h3>Module 6 · Reading synthesis</h3><p>What yosys reports about your design, and which warnings matter.</p><ol><li><a href="/learn/synthesis-warnings/">16 Synthesis warnings that matter</a></li><li><a href="/learn/state-machines/">17 State machines</a></li><li><a href="/learn/where-the-luts-go/">18 Where the LUTs go</a></li></ol></li><li><h3>Module 7 · Place, route, timing</h3><p>Where the cells land on the die, and whether the clock is met.</p><ol><li><a href="/learn/placement/">19 Placement: where the cells land</a></li><li><a href="/learn/logic-depth/">20 Logic depth</a></li><li><a href="/learn/timing-slack/">21 Timing and slack</a></li></ol></li><li><h3>Module 8 · The bitstream</h3><p>How a routed design becomes the bits the chip loads, with no Vivado.</p><ol><li><a href="/learn/fasm-to-frames/">22 From FASM to frames</a></li><li><a href="/learn/a-build-receipt/">23 A build, itemised</a></li><li><a href="/learn/read-a-bitstream/">24 Read a bitstream</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>Ask the chip who it is, load the bits, and check them.</p><ol><li><a href="/learn/jtag-asks-the-chip/">25 JTAG asks the chip</a></li><li><a href="/learn/same-bits-on-the-board/">26 The same bits, on the board</a></li><li><a href="/learn/capstone/">27 Capstone: out-guess yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/seven-backends/index.html b/apps/website/public/learn/seven-backends/index.html
index 744ecbcbde..1166effae0 100644
--- a/apps/website/public/learn/seven-backends/index.html
+++ b/apps/website/public/learn/seven-backends/index.html
@@ -85,5 +85,6 @@ <h1>Seven backends</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · A spec file</h3><p>What a .t27 file holds: a module line, prose lines and comments.</p><ol><li><a href="/learn/what-a-spec-is/">01 What a spec is</a></li><li><a href="/learn/the-module-line/">02 The module line</a></li><li><a href="/learn/comments-and-prose/">03 Comments and prose lines</a></li></ol></li><li><h3>Module 2 · Values and types</h3><p>Constants, the width of a whole number, and true, false and text.</p><ol><li><a href="/learn/constants/">04 Constants</a></li><li><a href="/learn/whole-numbers/">05 Whole numbers and their widths</a></li><li><a href="/learn/true-false-and-text/">06 True, false and text</a></li></ol></li><li><h3>Module 3 · Arrays, trits and expressions</h3><p>Lists of values, the three values of a trit, and what operators compute.</p><ol><li><a href="/learn/arrays/">07 Arrays</a></li><li><a href="/learn/trits/">08 Trits: minus one, zero, one</a></li><li><a href="/learn/expressions/">09 Expressions</a></li></ol></li><li><h3>Module 4 · Tests and functions</h3><p>A test block, several small tests in one spec, and a function with typed inputs.</p><ol><li><a href="/learn/a-test-block/">10 A test block</a></li><li><a href="/learn/many-tests/">11 Many tests in one spec</a></li><li><a href="/learn/functions/">12 Functions</a></li></ol></li><li><h3>Module 5 · Inside a function</h3><p>Names with let and var, choices with if and switch, and loops.</p><ol><li><a href="/learn/local-names/">13 Local names: let and var</a></li><li><a href="/learn/if-and-else/">14 Choices: if, else and switch</a></li><li><a href="/learn/loops/">15 Loops: while and for</a></li></ol></li><li><h3>Module 6 · Visibility and your own types</h3><p>What pub shares with other modules, and the structs and enums you declare.</p><ol><li><a href="/learn/pub-or-private/">16 pub or private</a></li><li><a href="/learn/structs/">17 Structs</a></li><li><a href="/learn/enums/">18 Enums</a></li></ol></li><li><h3>Module 7 · Modules, rules and gen-ts</h3><p>Reading another module, stating a rule that always holds, and a spec turned into TypeScript.</p><ol><li><a href="/learn/use-other-modules/">19 Using other modules</a></li><li><a href="/learn/invariants/">20 Invariants and benches</a></li><li><a href="/learn/gen-ts/">21 gen-ts: a spec becomes TypeScript</a></li></ol></li><li><h3>Module 8 · The compiler and tri</h3><p>The backends t27c writes, the stages it reads a spec in, and the tri command.</p><ol><li class="on" aria-current="page"><a href="/learn/seven-backends/">22 Seven backends</a></li><li><a href="/learn/stages-of-t27c/">23 The stages of t27c</a></li><li><a href="/learn/tri-commands/">24 tri: the command line</a></li></ol></li><li><h3>Module 9 · Errors, a program, what next</h3><p>Reading a compiler error, one small complete program, and the course to take next.</p><ol><li><a href="/learn/reading-errors/">25 Reading compiler errors</a></li><li><a href="/learn/a-small-program/">26 A small complete program</a></li><li><a href="/learn/where-next/">27 Where to go next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/shift-and-add/index.html b/apps/website/public/learn/shift-and-add/index.html
index 5bf7f23b80..1b4d10f6fb 100644
--- a/apps/website/public/learn/shift-and-add/index.html
+++ b/apps/website/public/learn/shift-and-add/index.html
@@ -372,5 +372,6 @@ <h1>Shift and add</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Adding numbers</h3><p>Why addition is slow: the carry that walks up the word, the dedicated chain in every slice, prefix networks and carry-save trees.</p><ol><li><a href="/learn/the-carry-chain/">01 The carry chain</a></li><li><a href="/learn/prefix-adders/">02 Looking ahead: prefix adders</a></li><li><a href="/learn/carry-save/">03 Three numbers at once</a></li></ol></li><li><h3>Module 2 · Multiplying</h3><p>A product is a sum of shifted copies; Booth recoding halves them, and the DSP48E1 slice does the rest in one block.</p><ol><li class="on" aria-current="page"><a href="/learn/shift-and-add/">04 Shift and add</a></li><li><a href="/learn/booths-trick/">05 Booth's trick</a></li><li><a href="/learn/the-dsp48e1-slice/">06 The DSP48E1 slice</a></li></ol></li><li><h3>Module 3 · Fixed point</h3><p>Where the binary point sits, what rounding does to a value and to its average, and what each bit of a quantizer buys.</p><ol><li><a href="/learn/q-formats/">07 Q formats</a></li><li><a href="/learn/rounding-and-saturation/">08 Rounding and saturation</a></li><li><a href="/learn/quantization-noise/">09 Quantization noise</a></li></ol></li><li><h3>Module 4 · Functions in hardware</h3><p>Sine, cosine, angle and length from shifts and adds, and when a table is the better answer.</p><ol><li><a href="/learn/rotating-by-shifts/">10 Rotating by shifts</a></li><li><a href="/learn/angle-and-magnitude/">11 Angle and magnitude</a></li><li><a href="/learn/tables-and-interpolation/">12 Tables and interpolation</a></li></ol></li><li><h3>Module 5 · Signals and sampling</h3><p>What sampling does to a frequency, how an oscillator is built from an adder, and what a DFT bin measures.</p><ol><li><a href="/learn/sampling-and-aliasing/">13 Sampling and aliasing</a></li><li><a href="/learn/the-nco/">14 A numerically controlled oscillator</a></li><li><a href="/learn/bins-of-a-spectrum/">15 Bins of a spectrum</a></li></ol></li><li><h3>Module 6 · FIR filters</h3><p>The moving average, a windowed-sinc design with integer taps, and folding the taps onto DSP slices.</p><ol><li><a href="/learn/the-moving-average/">16 The moving average</a></li><li><a href="/learn/designing-taps/">17 Designing taps</a></li><li><a href="/learn/folding-taps-onto-slices/">18 Folding taps onto slices</a></li></ol></li><li><h3>Module 7 · Multirate</h3><p>Lowering the sample rate without folding noise in: decimation, the CIC filter and the polyphase form.</p><ol><li><a href="/learn/lowering-the-rate/">19 Lowering the rate</a></li><li><a href="/learn/the-cic-filter/">20 The CIC filter</a></li><li><a href="/learn/polyphase/">21 Polyphase</a></li></ol></li><li><h3>Module 8 · The FFT</h3><p>N log N instead of N^2: butterflies, the bit-reversed input order and the bits each stage adds.</p><ol><li><a href="/learn/the-butterfly/">22 The butterfly</a></li><li><a href="/learn/twiddles-and-bit-reversal/">23 Twiddles and bit reversal</a></li><li><a href="/learn/growth-through-the-stages/">24 Growth through the stages</a></li></ol></li><li><h3>Module 9 · On the bench</h3><p>From the arithmetic to the board: a correlator from the modem spec, a budget of slices and timing, and the capstone filter.</p><ol><li><a href="/learn/a-correlator-finds-the-frame/">25 A correlator finds the frame</a></li><li><a href="/learn/does-it-fit-and-close-timing/">26 Does it fit and close timing?</a></li><li><a href="/learn/capstone-a-filter-on-the-board/">27 Capstone: a filter on the board</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/simulator-internals/index.html b/apps/website/public/learn/simulator-internals/index.html
index 3bc3de156a..ee26a2af53 100644
--- a/apps/website/public/learn/simulator-internals/index.html
+++ b/apps/website/public/learn/simulator-internals/index.html
@@ -384,5 +384,6 @@ <h1>Inside the simulator</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Why verify</h3><p>Designs that compile and are wrong, the model that decides, and the plan written before the code.</p><ol><li><a href="/learn/bugs-that-compile/">01 Bugs that compile</a></li><li><a href="/learn/golden-model/">02 The golden model</a></li><li><a href="/learn/plan-before-code/">03 Plan before code</a></li></ol></li><li><h3>Module 2 · Testbenches</h3><p>Stimulus, checks and a verdict, written as one spec beside the design it judges.</p><ol><li><a href="/learn/anatomy-of-a-tb/">04 Anatomy of a testbench</a></li><li><a href="/learn/self-checking/">05 Self-checking, not staring</a></li><li><a href="/learn/directed-vs-random/">06 Directed versus random</a></li></ol></li><li><h3>Module 3 · Waveforms</h3><p>A trace of every signal, read the way a hardware engineer reads it, and two runs compared.</p><ol><li><a href="/learn/vcd-format/">07 The VCD format</a></li><li><a href="/learn/debug-with-waves/">08 Debug with waves</a></li><li><a href="/learn/compare-traces/">09 Compare two traces</a></li></ol></li><li><h3>Module 4 · Conformance vectors</h3><p>Cases with the answer written beside them, kept where the compiler can reach them.</p><ol><li><a href="/learn/what-a-vector-is/">10 What a vector is</a></li><li><a href="/learn/golden-ruler-vectors/">11 Golden ruler vectors</a></li><li><a href="/learn/end-to-end-scenarios/">12 End-to-end scenarios</a></li></ol></li><li><h3>Module 5 · Cosimulation</h3><p>Spec, simulator and board agreeing on the bench Artix-7 XC7A200T, and what to do when they do not.</p><ol><li><a href="/learn/three-way-agreement/">13 Three-way agreement</a></li><li class="on" aria-current="page"><a href="/learn/simulator-internals/">14 Inside the simulator</a></li><li><a href="/learn/when-they-disagree/">15 When they disagree</a></li></ol></li><li><h3>Module 6 · Coverage</h3><p>What the tests touched: lines, toggles, states, and what that number hides.</p><ol><li><a href="/learn/line-and-toggle/">16 Line and toggle</a></li><li><a href="/learn/fsm-coverage/">17 FSM coverage</a></li><li><a href="/learn/coverage-lies/">18 Coverage lies</a></li></ol></li><li><h3>Module 7 · Formal</h3><p>Assertions that hold every cycle, bounded search for a counterexample, and why a proof needs induction.</p><ol><li><a href="/learn/assertions/">19 Assertions</a></li><li><a href="/learn/bounded-model-checking/">20 Bounded model checking</a></li><li><a href="/learn/induction/">21 Induction</a></li></ol></li><li><h3>Module 8 · Mutation</h3><p>Break the design on purpose and count what the tests catch.</p><ol><li><a href="/learn/kill-the-mutant/">22 Kill the mutant</a></li><li><a href="/learn/surviving-mutants/">23 Surviving mutants</a></li><li><a href="/learn/test-strength/">24 Test strength</a></li></ol></li><li><h3>Module 9 · Sign-off</h3><p>One command, every receipt, a clean verdict you can show.</p><ol><li><a href="/learn/build-verify/">25 Build and verify</a></li><li><a href="/learn/board-selftest/">26 The board self-tests</a></li><li><a href="/learn/verify-capstone/">27 Capstone: verify a module whole</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/skew-and-insertion/index.html b/apps/website/public/learn/skew-and-insertion/index.html
index 336daf2f69..828a75d076 100644
--- a/apps/website/public/learn/skew-and-insertion/index.html
+++ b/apps/website/public/learn/skew-and-insertion/index.html
@@ -294,5 +294,6 @@ <h1>Skew and insertion delay</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a clock is</h3><p>One edge, one world: what shares a clock edge shares a world, the period and the jitter of a real edge, and where the clock enters a board.</p><ol><li><a href="/learn/one-edge-one-world/">01 One edge, one world</a></li><li><a href="/learn/period-and-jitter/">02 A period, an edge, and jitter</a></li><li><a href="/learn/where-the-clock-enters/">03 Where the clock enters</a></li></ol></li><li><h3>Module 2 · Clock trees</h3><p>Skew and insertion delay, the global buffer network, and the trap of gating a clock with logic.</p><ol><li class="on" aria-current="page"><a href="/learn/skew-and-insertion/">04 Skew and insertion delay</a></li><li><a href="/learn/global-buffers/">05 The global buffer</a></li><li><a href="/learn/gated-clock-trap/">06 The gated-clock trap</a></li></ol></li><li><h3>Module 3 · PLL and MMCM</h3><p>Multiply and divide one clock into another, move its phase in steps of the VCO, and which clocks the analyzer treats as related.</p><ol><li><a href="/learn/multiply-and-divide/">07 Multiply and divide</a></li><li><a href="/learn/phase-and-lock/">08 Phase and lock</a></li><li><a href="/learn/related-or-not/">09 Related or not</a></li></ol></li><li><h3>Module 4 · Resets</h3><p>Assert asynchronously, release synchronously: the three reset kinds, the release pipe, and the tree a reset grows.</p><ol><li><a href="/learn/sync-or-async/">10 Sync or async</a></li><li><a href="/learn/reset-release/">11 The release pipe</a></li><li><a href="/learn/reset-fanout/">12 Reset fanout</a></li></ol></li><li><h3>Module 5 · Metastability</h3><p>The setup-hold window, the mean time between failures in integer arithmetic, and the two flops that fix it.</p><ol><li><a href="/learn/setup-hold-window/">13 The setup-hold window</a></li><li><a href="/learn/mtbf/">14 Mean time between failures</a></li><li><a href="/learn/two-flop-sync/">15 The two-flop synchronizer</a></li></ol></li><li><h3>Module 6 · Crossing many bits</h3><p>Why a binary bus tears, why Gray code does not, and the handshake that moves a pulse between worlds.</p><ol><li><a href="/learn/why-binary-fails/">16 Why binary fails</a></li><li><a href="/learn/gray-code/">17 Gray code</a></li><li><a href="/learn/handshakes-and-pulses/">18 Handshakes and pulses</a></li></ol></li><li><h3>Module 7 · The asynchronous FIFO</h3><p>Pointers, flags and depth: the buffer that moves a stream between two clocks.</p><ol><li><a href="/learn/pointers-and-flags/">19 Pointers and flags</a></li><li><a href="/learn/depth-sizing/">20 Sizing the depth</a></li><li><a href="/learn/fifo-testbench/">21 The FIFO testbench</a></li></ol></li><li><h3>Module 8 · Constraints</h3><p>The lines that tell the analyzer what a clock is, which paths not to check, and what the pins must meet.</p><ol><li><a href="/learn/create-clock/">22 create_clock</a></li><li><a href="/learn/false-path-and-max-delay/">23 False paths and max delay</a></li><li><a href="/learn/io-timing/">24 IO timing</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>A CDC report, one crossing captured at the flip-flops, and the bitstream diff that closes the course.</p><ol><li><a href="/learn/cdc-report/">25 The CDC report</a></li><li><a href="/learn/capture-a-crossing/">26 Capture a crossing</a></li><li><a href="/learn/capstone-diff/">27 Capstone: change it and diff it</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/sort-by-three/index.html b/apps/website/public/learn/sort-by-three/index.html
index 6851d9d97a..cdc45d9089 100644
--- a/apps/website/public/learn/sort-by-three/index.html
+++ b/apps/website/public/learn/sort-by-three/index.html
@@ -769,5 +769,6 @@ <h1>Sort with three-way compares</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Three values</h3><p>Why three, how balanced ternary writes every number without a sign, and what flipping and cutting trits do.</p><ol><li><a href="/learn/why-three/">01 Why three</a></li><li><a href="/learn/balanced-ternary/">02 Balanced ternary</a></li><li><a href="/learn/flip-and-cut/">03 Flip to negate, cut to round</a></li></ol></li><li><h3>Module 2 · Logic with unknown</h3><p>Kleene's three-valued gates, two gates binary has no twin for, and addition as a pair of tables.</p><ol><li><a href="/learn/kleene-logic/">04 Three-valued logic</a></li><li><a href="/learn/consensus-and-any/">05 Consensus and any</a></li><li><a href="/learn/sum-and-carry/">06 Sum and carry</a></li></ol></li><li><h3>Module 3 · Adding trits</h3><p>The half adder, the full adder and a carry that ripples left, one place at a time.</p><ol><li><a href="/learn/half-adder/">07 The half adder</a></li><li><a href="/learn/full-adder/">08 The full adder</a></li><li><a href="/learn/ripple-carry/">09 Ripple carry</a></li></ol></li><li><h3>Module 4 · Multiplying without a multiplier</h3><p>Copy, drop or flip: a product by one trit, long multiplication, and a MAC that only adds.</p><ol><li><a href="/learn/times-a-trit/">10 Times one trit</a></li><li><a href="/learn/long-multiplication/">11 Long multiplication</a></li><li><a href="/learn/ternary-mac/">12 A ternary MAC</a></li></ol></li><li><h3>Module 5 · Trits in binary memory</h3><p>Two bits per trit, five trits per byte, and the bits a 27-trit word needs.</p><ol><li><a href="/learn/two-bits-per-trit/">13 Two bits per trit</a></li><li><a href="/learn/five-trits-a-byte/">14 Five trits in a byte</a></li><li><a href="/learn/a-word-of-27/">15 A word of 27 trits</a></li></ol></li><li><h3>Module 6 · The TRI-27 instruction word</h3><p>The 32-bit word the Trinity emulator decodes: its fields, its 47 opcodes and its 15-bit immediate.</p><ol><li><a href="/learn/the-instruction-word/">16 The instruction word</a></li><li><a href="/learn/the-opcode-map/">17 The opcode map</a></li><li><a href="/learn/the-immediate/">18 The immediate</a></li></ol></li><li><h3>Module 7 · A ternary machine</h3><p>An ALU built from this course's adders, a three-way jump, and a program you can step.</p><ol><li><a href="/learn/a-ternary-alu/">19 A ternary ALU</a></li><li><a href="/learn/three-way-jump/">20 A three-way jump</a></li><li><a href="/learn/run-a-program/">21 Run a program</a></li></ol></li><li><h3>Module 8 · Three answers</h3><p>Compare at the highest differing trit, find a number in thirds, sort with three-way compares.</p><ol><li><a href="/learn/compare-at-the-top/">22 Compare at the top</a></li><li><a href="/learn/questions-in-thirds/">23 Questions in thirds</a></li><li class="on" aria-current="page"><a href="/learn/sort-by-three/">24 Sort with three-way compares</a></li></ol></li><li><h3>Module 9 · Ternary neurons</h3><p>Weights of -1, 0, +1: one neuron, a detector and a small layer, with no multiplier anywhere.</p><ol><li><a href="/learn/one-ternary-neuron/">25 One ternary neuron</a></li><li><a href="/learn/a-bar-detector/">26 A bar detector</a></li><li><a href="/learn/a-tiny-layer/">27 A tiny layer</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/spec-to-verilog/index.html b/apps/website/public/learn/spec-to-verilog/index.html
index 0a050dfada..7613ed9417 100644
--- a/apps/website/public/learn/spec-to-verilog/index.html
+++ b/apps/website/public/learn/spec-to-verilog/index.html
@@ -107,5 +107,6 @@ <h1>From a spec to Verilog</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The chip</h3><p>What an FPGA is, which chip we use, and how its pins meet the board.</p><ol><li><a href="/learn/fpga-is-a-grid/">01 An FPGA is a grid you configure</a></li><li><a href="/learn/which-chip/">02 Which chip is this?</a></li><li><a href="/learn/pins-and-package/">03 Pins, balls and the package</a></li></ol></li><li><h3>Module 2 · Numbers in hardware</h3><p>Bits, trits and number formats, and what arithmetic costs in logic.</p><ol><li><a href="/learn/bits-and-trits/">04 Bits, trits and GF16</a></li><li><a href="/learn/what-arithmetic-costs/">05 What arithmetic costs</a></li><li><a href="/learn/how-much-fits/">06 How much fits on the chip</a></li></ol></li><li><h3>Module 3 · Your t27 program</h3><p>Write a spec, test it, and see why compiling is not the same as being right.</p><ol><li><a href="/learn/hello-t27/">07 Hello, t27</a></li><li><a href="/learn/tests-are-the-spec/">08 Tests are part of the spec</a></li><li><a href="/learn/compiles-is-not-correct/">09 Compiles is not the same as correct</a></li></ol></li><li><h3>Module 4 · Inside t27c</h3><p>How the compiler reads a spec and what it writes, including the native t27b.</p><ol><li><a href="/learn/compiler-stages/">10 The stages of t27c</a></li><li><a href="/learn/t27c-on-a-real-machine/">11 t27c on a real machine</a></li><li><a href="/learn/seven-backends-and-t27b/">12 Seven backends, and t27b</a></li></ol></li><li><h3>Module 5 · From spec to hardware</h3><p>The Verilog t27c writes, the cells it becomes, and what one LUT does.</p><ol><li class="on" aria-current="page"><a href="/learn/spec-to-verilog/">13 From a spec to Verilog</a></li><li><a href="/learn/count-the-cells/">14 Count the cells</a></li><li><a href="/learn/one-lut-up-close/">15 One LUT, up close</a></li></ol></li><li><h3>Module 6 · Reading synthesis</h3><p>What yosys reports about your design, and which warnings matter.</p><ol><li><a href="/learn/synthesis-warnings/">16 Synthesis warnings that matter</a></li><li><a href="/learn/state-machines/">17 State machines</a></li><li><a href="/learn/where-the-luts-go/">18 Where the LUTs go</a></li></ol></li><li><h3>Module 7 · Place, route, timing</h3><p>Where the cells land on the die, and whether the clock is met.</p><ol><li><a href="/learn/placement/">19 Placement: where the cells land</a></li><li><a href="/learn/logic-depth/">20 Logic depth</a></li><li><a href="/learn/timing-slack/">21 Timing and slack</a></li></ol></li><li><h3>Module 8 · The bitstream</h3><p>How a routed design becomes the bits the chip loads, with no Vivado.</p><ol><li><a href="/learn/fasm-to-frames/">22 From FASM to frames</a></li><li><a href="/learn/a-build-receipt/">23 A build, itemised</a></li><li><a href="/learn/read-a-bitstream/">24 Read a bitstream</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>Ask the chip who it is, load the bits, and check them.</p><ol><li><a href="/learn/jtag-asks-the-chip/">25 JTAG asks the chip</a></li><li><a href="/learn/same-bits-on-the-board/">26 The same bits, on the board</a></li><li><a href="/learn/capstone/">27 Capstone: out-guess yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/stages-of-t27c/index.html b/apps/website/public/learn/stages-of-t27c/index.html
index 9a9295fb52..50129deb31 100644
--- a/apps/website/public/learn/stages-of-t27c/index.html
+++ b/apps/website/public/learn/stages-of-t27c/index.html
@@ -85,5 +85,6 @@ <h1>The stages of t27c</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · A spec file</h3><p>What a .t27 file holds: a module line, prose lines and comments.</p><ol><li><a href="/learn/what-a-spec-is/">01 What a spec is</a></li><li><a href="/learn/the-module-line/">02 The module line</a></li><li><a href="/learn/comments-and-prose/">03 Comments and prose lines</a></li></ol></li><li><h3>Module 2 · Values and types</h3><p>Constants, the width of a whole number, and true, false and text.</p><ol><li><a href="/learn/constants/">04 Constants</a></li><li><a href="/learn/whole-numbers/">05 Whole numbers and their widths</a></li><li><a href="/learn/true-false-and-text/">06 True, false and text</a></li></ol></li><li><h3>Module 3 · Arrays, trits and expressions</h3><p>Lists of values, the three values of a trit, and what operators compute.</p><ol><li><a href="/learn/arrays/">07 Arrays</a></li><li><a href="/learn/trits/">08 Trits: minus one, zero, one</a></li><li><a href="/learn/expressions/">09 Expressions</a></li></ol></li><li><h3>Module 4 · Tests and functions</h3><p>A test block, several small tests in one spec, and a function with typed inputs.</p><ol><li><a href="/learn/a-test-block/">10 A test block</a></li><li><a href="/learn/many-tests/">11 Many tests in one spec</a></li><li><a href="/learn/functions/">12 Functions</a></li></ol></li><li><h3>Module 5 · Inside a function</h3><p>Names with let and var, choices with if and switch, and loops.</p><ol><li><a href="/learn/local-names/">13 Local names: let and var</a></li><li><a href="/learn/if-and-else/">14 Choices: if, else and switch</a></li><li><a href="/learn/loops/">15 Loops: while and for</a></li></ol></li><li><h3>Module 6 · Visibility and your own types</h3><p>What pub shares with other modules, and the structs and enums you declare.</p><ol><li><a href="/learn/pub-or-private/">16 pub or private</a></li><li><a href="/learn/structs/">17 Structs</a></li><li><a href="/learn/enums/">18 Enums</a></li></ol></li><li><h3>Module 7 · Modules, rules and gen-ts</h3><p>Reading another module, stating a rule that always holds, and a spec turned into TypeScript.</p><ol><li><a href="/learn/use-other-modules/">19 Using other modules</a></li><li><a href="/learn/invariants/">20 Invariants and benches</a></li><li><a href="/learn/gen-ts/">21 gen-ts: a spec becomes TypeScript</a></li></ol></li><li><h3>Module 8 · The compiler and tri</h3><p>The backends t27c writes, the stages it reads a spec in, and the tri command.</p><ol><li><a href="/learn/seven-backends/">22 Seven backends</a></li><li class="on" aria-current="page"><a href="/learn/stages-of-t27c/">23 The stages of t27c</a></li><li><a href="/learn/tri-commands/">24 tri: the command line</a></li></ol></li><li><h3>Module 9 · Errors, a program, what next</h3><p>Reading a compiler error, one small complete program, and the course to take next.</p><ol><li><a href="/learn/reading-errors/">25 Reading compiler errors</a></li><li><a href="/learn/a-small-program/">26 A small complete program</a></li><li><a href="/learn/where-next/">27 Where to go next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/state-machines/index.html b/apps/website/public/learn/state-machines/index.html
index fb4ff0952a..be0d989ceb 100644
--- a/apps/website/public/learn/state-machines/index.html
+++ b/apps/website/public/learn/state-machines/index.html
@@ -333,5 +333,6 @@ <h1>State machines</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The chip</h3><p>What an FPGA is, which chip we use, and how its pins meet the board.</p><ol><li><a href="/learn/fpga-is-a-grid/">01 An FPGA is a grid you configure</a></li><li><a href="/learn/which-chip/">02 Which chip is this?</a></li><li><a href="/learn/pins-and-package/">03 Pins, balls and the package</a></li></ol></li><li><h3>Module 2 · Numbers in hardware</h3><p>Bits, trits and number formats, and what arithmetic costs in logic.</p><ol><li><a href="/learn/bits-and-trits/">04 Bits, trits and GF16</a></li><li><a href="/learn/what-arithmetic-costs/">05 What arithmetic costs</a></li><li><a href="/learn/how-much-fits/">06 How much fits on the chip</a></li></ol></li><li><h3>Module 3 · Your t27 program</h3><p>Write a spec, test it, and see why compiling is not the same as being right.</p><ol><li><a href="/learn/hello-t27/">07 Hello, t27</a></li><li><a href="/learn/tests-are-the-spec/">08 Tests are part of the spec</a></li><li><a href="/learn/compiles-is-not-correct/">09 Compiles is not the same as correct</a></li></ol></li><li><h3>Module 4 · Inside t27c</h3><p>How the compiler reads a spec and what it writes, including the native t27b.</p><ol><li><a href="/learn/compiler-stages/">10 The stages of t27c</a></li><li><a href="/learn/t27c-on-a-real-machine/">11 t27c on a real machine</a></li><li><a href="/learn/seven-backends-and-t27b/">12 Seven backends, and t27b</a></li></ol></li><li><h3>Module 5 · From spec to hardware</h3><p>The Verilog t27c writes, the cells it becomes, and what one LUT does.</p><ol><li><a href="/learn/spec-to-verilog/">13 From a spec to Verilog</a></li><li><a href="/learn/count-the-cells/">14 Count the cells</a></li><li><a href="/learn/one-lut-up-close/">15 One LUT, up close</a></li></ol></li><li><h3>Module 6 · Reading synthesis</h3><p>What yosys reports about your design, and which warnings matter.</p><ol><li><a href="/learn/synthesis-warnings/">16 Synthesis warnings that matter</a></li><li class="on" aria-current="page"><a href="/learn/state-machines/">17 State machines</a></li><li><a href="/learn/where-the-luts-go/">18 Where the LUTs go</a></li></ol></li><li><h3>Module 7 · Place, route, timing</h3><p>Where the cells land on the die, and whether the clock is met.</p><ol><li><a href="/learn/placement/">19 Placement: where the cells land</a></li><li><a href="/learn/logic-depth/">20 Logic depth</a></li><li><a href="/learn/timing-slack/">21 Timing and slack</a></li></ol></li><li><h3>Module 8 · The bitstream</h3><p>How a routed design becomes the bits the chip loads, with no Vivado.</p><ol><li><a href="/learn/fasm-to-frames/">22 From FASM to frames</a></li><li><a href="/learn/a-build-receipt/">23 A build, itemised</a></li><li><a href="/learn/read-a-bitstream/">24 Read a bitstream</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>Ask the chip who it is, load the bits, and check them.</p><ol><li><a href="/learn/jtag-asks-the-chip/">25 JTAG asks the chip</a></li><li><a href="/learn/same-bits-on-the-board/">26 The same bits, on the board</a></li><li><a href="/learn/capstone/">27 Capstone: out-guess yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/status-and-fifos/index.html b/apps/website/public/learn/status-and-fifos/index.html
index 1151fe3521..b805b027d2 100644
--- a/apps/website/public/learn/status-and-fifos/index.html
+++ b/apps/website/public/learn/status-and-fifos/index.html
@@ -466,5 +466,6 @@ <h1>Status and FIFOs</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a bus is</h3><p>Why a bus exists at all: a conversation on wires, framed and addressed, and who is allowed to talk.</p><ol><li><a href="/learn/wires-that-talk/">01 Wires that talk</a></li><li><a href="/learn/framing-a-conversation/">02 Framing a conversation</a></li><li><a href="/learn/master-and-servant/">03 Master and servant</a></li></ol></li><li><h3>Module 2 · UART</h3><p>The two-wire bus with no clock: the frame, the divisor that sets the rate, and the status a driver polls.</p><ol><li><a href="/learn/the-frame/">04 The frame: start, data, stop</a></li><li><a href="/learn/baud-and-the-divisor/">05 Baud and the divisor</a></li><li class="on" aria-current="page"><a href="/learn/status-and-fifos/">06 Status and FIFOs</a></li></ol></li><li><h3>Module 3 · SPI</h3><p>The clocked conversation: four modes, a prescaler ladder, and a chip select per servant.</p><ol><li><a href="/learn/modes-cpol-cpha/">07 Modes: CPOL and CPHA</a></li><li><a href="/learn/the-prescaler-ladder/">08 The prescaler ladder</a></li><li><a href="/learn/chip-select-and-widths/">09 Chip select and widths</a></li></ol></li><li><h3>Module 4 · APB</h3><p>The register bus: PSEL and PENABLE, strobes and wait states, and how many address bits a peripheral count costs.</p><ol><li><a href="/learn/psel-and-penable/">10 PSEL and PENABLE</a></li><li><a href="/learn/strobes-and-waits/">11 Strobes and waits</a></li><li><a href="/learn/address-decode/">12 Address decode</a></li></ol></li><li><h3>Module 5 · AXI4</h3><p>The five channels: address, data and response in both directions, lite or full, bursts and IDs.</p><ol><li><a href="/learn/five-channels/">13 Five channels</a></li><li><a href="/learn/lite-or-full/">14 Lite or full</a></li><li><a href="/learn/bursts-and-ids/">15 Bursts and IDs</a></li></ol></li><li><h3>Module 6 · Memory</h3><p>What sits on the far side of every bus: memory maps, port kinds, and latency that a wait state must cover.</p><ol><li><a href="/learn/memory-maps/">16 Memory maps</a></li><li><a href="/learn/ports-and-kinds/">17 Ports and kinds</a></li><li><a href="/learn/latency/">18 Latency</a></li></ol></li><li><h3>Module 7 · Bridges</h3><p>Why designs grow more than one bus, and the packet bridge that moves work between them.</p><ol><li><a href="/learn/why-bridges/">19 Why bridges</a></li><li><a href="/learn/the-packet-bridge/">20 The packet bridge</a></li><li><a href="/learn/everything-in-one-bench/">21 Everything in one bench</a></li></ol></li><li><h3>Module 8 · Ethernet</h3><p>Frames, the frame check sequence, RGMII timing, and the pre-registered steps of a real bring-up.</p><ol><li><a href="/learn/frames-and-the-fcs/">22 Frames and the FCS</a></li><li><a href="/learn/rgmii-timing/">23 RGMII timing</a></li><li><a href="/learn/bring-up-on-the-bench/">24 Bring-up on the bench</a></li></ol></li><li><h3>Module 9 · The bench</h3><p>The discipline that guards real hardware: who holds the IO, taking and giving the lock, and what runs next.</p><ol><li><a href="/learn/who-holds-the-io/">25 Who holds the IO</a></li><li><a href="/learn/claim-and-release/">26 Claim and release</a></li><li><a href="/learn/capstone-what-runs-next/">27 Capstone: what runs next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/strobes-and-waits/index.html b/apps/website/public/learn/strobes-and-waits/index.html
index ded77ad2c1..ef7fde6976 100644
--- a/apps/website/public/learn/strobes-and-waits/index.html
+++ b/apps/website/public/learn/strobes-and-waits/index.html
@@ -417,5 +417,6 @@ <h1>Strobes and waits</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a bus is</h3><p>Why a bus exists at all: a conversation on wires, framed and addressed, and who is allowed to talk.</p><ol><li><a href="/learn/wires-that-talk/">01 Wires that talk</a></li><li><a href="/learn/framing-a-conversation/">02 Framing a conversation</a></li><li><a href="/learn/master-and-servant/">03 Master and servant</a></li></ol></li><li><h3>Module 2 · UART</h3><p>The two-wire bus with no clock: the frame, the divisor that sets the rate, and the status a driver polls.</p><ol><li><a href="/learn/the-frame/">04 The frame: start, data, stop</a></li><li><a href="/learn/baud-and-the-divisor/">05 Baud and the divisor</a></li><li><a href="/learn/status-and-fifos/">06 Status and FIFOs</a></li></ol></li><li><h3>Module 3 · SPI</h3><p>The clocked conversation: four modes, a prescaler ladder, and a chip select per servant.</p><ol><li><a href="/learn/modes-cpol-cpha/">07 Modes: CPOL and CPHA</a></li><li><a href="/learn/the-prescaler-ladder/">08 The prescaler ladder</a></li><li><a href="/learn/chip-select-and-widths/">09 Chip select and widths</a></li></ol></li><li><h3>Module 4 · APB</h3><p>The register bus: PSEL and PENABLE, strobes and wait states, and how many address bits a peripheral count costs.</p><ol><li><a href="/learn/psel-and-penable/">10 PSEL and PENABLE</a></li><li class="on" aria-current="page"><a href="/learn/strobes-and-waits/">11 Strobes and waits</a></li><li><a href="/learn/address-decode/">12 Address decode</a></li></ol></li><li><h3>Module 5 · AXI4</h3><p>The five channels: address, data and response in both directions, lite or full, bursts and IDs.</p><ol><li><a href="/learn/five-channels/">13 Five channels</a></li><li><a href="/learn/lite-or-full/">14 Lite or full</a></li><li><a href="/learn/bursts-and-ids/">15 Bursts and IDs</a></li></ol></li><li><h3>Module 6 · Memory</h3><p>What sits on the far side of every bus: memory maps, port kinds, and latency that a wait state must cover.</p><ol><li><a href="/learn/memory-maps/">16 Memory maps</a></li><li><a href="/learn/ports-and-kinds/">17 Ports and kinds</a></li><li><a href="/learn/latency/">18 Latency</a></li></ol></li><li><h3>Module 7 · Bridges</h3><p>Why designs grow more than one bus, and the packet bridge that moves work between them.</p><ol><li><a href="/learn/why-bridges/">19 Why bridges</a></li><li><a href="/learn/the-packet-bridge/">20 The packet bridge</a></li><li><a href="/learn/everything-in-one-bench/">21 Everything in one bench</a></li></ol></li><li><h3>Module 8 · Ethernet</h3><p>Frames, the frame check sequence, RGMII timing, and the pre-registered steps of a real bring-up.</p><ol><li><a href="/learn/frames-and-the-fcs/">22 Frames and the FCS</a></li><li><a href="/learn/rgmii-timing/">23 RGMII timing</a></li><li><a href="/learn/bring-up-on-the-bench/">24 Bring-up on the bench</a></li></ol></li><li><h3>Module 9 · The bench</h3><p>The discipline that guards real hardware: who holds the IO, taking and giving the lock, and what runs next.</p><ol><li><a href="/learn/who-holds-the-io/">25 Who holds the IO</a></li><li><a href="/learn/claim-and-release/">26 Claim and release</a></li><li><a href="/learn/capstone-what-runs-next/">27 Capstone: what runs next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/structs/index.html b/apps/website/public/learn/structs/index.html
index 7cdc08e65b..49b3d80e3f 100644
--- a/apps/website/public/learn/structs/index.html
+++ b/apps/website/public/learn/structs/index.html
@@ -91,5 +91,6 @@ <h1>Structs</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · A spec file</h3><p>What a .t27 file holds: a module line, prose lines and comments.</p><ol><li><a href="/learn/what-a-spec-is/">01 What a spec is</a></li><li><a href="/learn/the-module-line/">02 The module line</a></li><li><a href="/learn/comments-and-prose/">03 Comments and prose lines</a></li></ol></li><li><h3>Module 2 · Values and types</h3><p>Constants, the width of a whole number, and true, false and text.</p><ol><li><a href="/learn/constants/">04 Constants</a></li><li><a href="/learn/whole-numbers/">05 Whole numbers and their widths</a></li><li><a href="/learn/true-false-and-text/">06 True, false and text</a></li></ol></li><li><h3>Module 3 · Arrays, trits and expressions</h3><p>Lists of values, the three values of a trit, and what operators compute.</p><ol><li><a href="/learn/arrays/">07 Arrays</a></li><li><a href="/learn/trits/">08 Trits: minus one, zero, one</a></li><li><a href="/learn/expressions/">09 Expressions</a></li></ol></li><li><h3>Module 4 · Tests and functions</h3><p>A test block, several small tests in one spec, and a function with typed inputs.</p><ol><li><a href="/learn/a-test-block/">10 A test block</a></li><li><a href="/learn/many-tests/">11 Many tests in one spec</a></li><li><a href="/learn/functions/">12 Functions</a></li></ol></li><li><h3>Module 5 · Inside a function</h3><p>Names with let and var, choices with if and switch, and loops.</p><ol><li><a href="/learn/local-names/">13 Local names: let and var</a></li><li><a href="/learn/if-and-else/">14 Choices: if, else and switch</a></li><li><a href="/learn/loops/">15 Loops: while and for</a></li></ol></li><li><h3>Module 6 · Visibility and your own types</h3><p>What pub shares with other modules, and the structs and enums you declare.</p><ol><li><a href="/learn/pub-or-private/">16 pub or private</a></li><li class="on" aria-current="page"><a href="/learn/structs/">17 Structs</a></li><li><a href="/learn/enums/">18 Enums</a></li></ol></li><li><h3>Module 7 · Modules, rules and gen-ts</h3><p>Reading another module, stating a rule that always holds, and a spec turned into TypeScript.</p><ol><li><a href="/learn/use-other-modules/">19 Using other modules</a></li><li><a href="/learn/invariants/">20 Invariants and benches</a></li><li><a href="/learn/gen-ts/">21 gen-ts: a spec becomes TypeScript</a></li></ol></li><li><h3>Module 8 · The compiler and tri</h3><p>The backends t27c writes, the stages it reads a spec in, and the tri command.</p><ol><li><a href="/learn/seven-backends/">22 Seven backends</a></li><li><a href="/learn/stages-of-t27c/">23 The stages of t27c</a></li><li><a href="/learn/tri-commands/">24 tri: the command line</a></li></ol></li><li><h3>Module 9 · Errors, a program, what next</h3><p>Reading a compiler error, one small complete program, and the course to take next.</p><ol><li><a href="/learn/reading-errors/">25 Reading compiler errors</a></li><li><a href="/learn/a-small-program/">26 A small complete program</a></li><li><a href="/learn/where-next/">27 Where to go next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/sum-and-carry/index.html b/apps/website/public/learn/sum-and-carry/index.html
index 394a3bfe89..0888ad865f 100644
--- a/apps/website/public/learn/sum-and-carry/index.html
+++ b/apps/website/public/learn/sum-and-carry/index.html
@@ -148,5 +148,6 @@ <h1>Sum and carry</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Three values</h3><p>Why three, how balanced ternary writes every number without a sign, and what flipping and cutting trits do.</p><ol><li><a href="/learn/why-three/">01 Why three</a></li><li><a href="/learn/balanced-ternary/">02 Balanced ternary</a></li><li><a href="/learn/flip-and-cut/">03 Flip to negate, cut to round</a></li></ol></li><li><h3>Module 2 · Logic with unknown</h3><p>Kleene's three-valued gates, two gates binary has no twin for, and addition as a pair of tables.</p><ol><li><a href="/learn/kleene-logic/">04 Three-valued logic</a></li><li><a href="/learn/consensus-and-any/">05 Consensus and any</a></li><li class="on" aria-current="page"><a href="/learn/sum-and-carry/">06 Sum and carry</a></li></ol></li><li><h3>Module 3 · Adding trits</h3><p>The half adder, the full adder and a carry that ripples left, one place at a time.</p><ol><li><a href="/learn/half-adder/">07 The half adder</a></li><li><a href="/learn/full-adder/">08 The full adder</a></li><li><a href="/learn/ripple-carry/">09 Ripple carry</a></li></ol></li><li><h3>Module 4 · Multiplying without a multiplier</h3><p>Copy, drop or flip: a product by one trit, long multiplication, and a MAC that only adds.</p><ol><li><a href="/learn/times-a-trit/">10 Times one trit</a></li><li><a href="/learn/long-multiplication/">11 Long multiplication</a></li><li><a href="/learn/ternary-mac/">12 A ternary MAC</a></li></ol></li><li><h3>Module 5 · Trits in binary memory</h3><p>Two bits per trit, five trits per byte, and the bits a 27-trit word needs.</p><ol><li><a href="/learn/two-bits-per-trit/">13 Two bits per trit</a></li><li><a href="/learn/five-trits-a-byte/">14 Five trits in a byte</a></li><li><a href="/learn/a-word-of-27/">15 A word of 27 trits</a></li></ol></li><li><h3>Module 6 · The TRI-27 instruction word</h3><p>The 32-bit word the Trinity emulator decodes: its fields, its 47 opcodes and its 15-bit immediate.</p><ol><li><a href="/learn/the-instruction-word/">16 The instruction word</a></li><li><a href="/learn/the-opcode-map/">17 The opcode map</a></li><li><a href="/learn/the-immediate/">18 The immediate</a></li></ol></li><li><h3>Module 7 · A ternary machine</h3><p>An ALU built from this course's adders, a three-way jump, and a program you can step.</p><ol><li><a href="/learn/a-ternary-alu/">19 A ternary ALU</a></li><li><a href="/learn/three-way-jump/">20 A three-way jump</a></li><li><a href="/learn/run-a-program/">21 Run a program</a></li></ol></li><li><h3>Module 8 · Three answers</h3><p>Compare at the highest differing trit, find a number in thirds, sort with three-way compares.</p><ol><li><a href="/learn/compare-at-the-top/">22 Compare at the top</a></li><li><a href="/learn/questions-in-thirds/">23 Questions in thirds</a></li><li><a href="/learn/sort-by-three/">24 Sort with three-way compares</a></li></ol></li><li><h3>Module 9 · Ternary neurons</h3><p>Weights of -1, 0, +1: one neuron, a detector and a small layer, with no multiplier anywhere.</p><ol><li><a href="/learn/one-ternary-neuron/">25 One ternary neuron</a></li><li><a href="/learn/a-bar-detector/">26 A bar detector</a></li><li><a href="/learn/a-tiny-layer/">27 A tiny layer</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/surviving-mutants/index.html b/apps/website/public/learn/surviving-mutants/index.html
index 28006792da..22c93628e4 100644
--- a/apps/website/public/learn/surviving-mutants/index.html
+++ b/apps/website/public/learn/surviving-mutants/index.html
@@ -110,5 +110,6 @@ <h1>Surviving mutants</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Why verify</h3><p>Designs that compile and are wrong, the model that decides, and the plan written before the code.</p><ol><li><a href="/learn/bugs-that-compile/">01 Bugs that compile</a></li><li><a href="/learn/golden-model/">02 The golden model</a></li><li><a href="/learn/plan-before-code/">03 Plan before code</a></li></ol></li><li><h3>Module 2 · Testbenches</h3><p>Stimulus, checks and a verdict, written as one spec beside the design it judges.</p><ol><li><a href="/learn/anatomy-of-a-tb/">04 Anatomy of a testbench</a></li><li><a href="/learn/self-checking/">05 Self-checking, not staring</a></li><li><a href="/learn/directed-vs-random/">06 Directed versus random</a></li></ol></li><li><h3>Module 3 · Waveforms</h3><p>A trace of every signal, read the way a hardware engineer reads it, and two runs compared.</p><ol><li><a href="/learn/vcd-format/">07 The VCD format</a></li><li><a href="/learn/debug-with-waves/">08 Debug with waves</a></li><li><a href="/learn/compare-traces/">09 Compare two traces</a></li></ol></li><li><h3>Module 4 · Conformance vectors</h3><p>Cases with the answer written beside them, kept where the compiler can reach them.</p><ol><li><a href="/learn/what-a-vector-is/">10 What a vector is</a></li><li><a href="/learn/golden-ruler-vectors/">11 Golden ruler vectors</a></li><li><a href="/learn/end-to-end-scenarios/">12 End-to-end scenarios</a></li></ol></li><li><h3>Module 5 · Cosimulation</h3><p>Spec, simulator and board agreeing on the bench Artix-7 XC7A200T, and what to do when they do not.</p><ol><li><a href="/learn/three-way-agreement/">13 Three-way agreement</a></li><li><a href="/learn/simulator-internals/">14 Inside the simulator</a></li><li><a href="/learn/when-they-disagree/">15 When they disagree</a></li></ol></li><li><h3>Module 6 · Coverage</h3><p>What the tests touched: lines, toggles, states, and what that number hides.</p><ol><li><a href="/learn/line-and-toggle/">16 Line and toggle</a></li><li><a href="/learn/fsm-coverage/">17 FSM coverage</a></li><li><a href="/learn/coverage-lies/">18 Coverage lies</a></li></ol></li><li><h3>Module 7 · Formal</h3><p>Assertions that hold every cycle, bounded search for a counterexample, and why a proof needs induction.</p><ol><li><a href="/learn/assertions/">19 Assertions</a></li><li><a href="/learn/bounded-model-checking/">20 Bounded model checking</a></li><li><a href="/learn/induction/">21 Induction</a></li></ol></li><li><h3>Module 8 · Mutation</h3><p>Break the design on purpose and count what the tests catch.</p><ol><li><a href="/learn/kill-the-mutant/">22 Kill the mutant</a></li><li class="on" aria-current="page"><a href="/learn/surviving-mutants/">23 Surviving mutants</a></li><li><a href="/learn/test-strength/">24 Test strength</a></li></ol></li><li><h3>Module 9 · Sign-off</h3><p>One command, every receipt, a clean verdict you can show.</p><ol><li><a href="/learn/build-verify/">25 Build and verify</a></li><li><a href="/learn/board-selftest/">26 The board self-tests</a></li><li><a href="/learn/verify-capstone/">27 Capstone: verify a module whole</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/sync-or-async/index.html b/apps/website/public/learn/sync-or-async/index.html
index cf5acee9b8..d8e597953a 100644
--- a/apps/website/public/learn/sync-or-async/index.html
+++ b/apps/website/public/learn/sync-or-async/index.html
@@ -329,5 +329,6 @@ <h1>Sync or async</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a clock is</h3><p>One edge, one world: what shares a clock edge shares a world, the period and the jitter of a real edge, and where the clock enters a board.</p><ol><li><a href="/learn/one-edge-one-world/">01 One edge, one world</a></li><li><a href="/learn/period-and-jitter/">02 A period, an edge, and jitter</a></li><li><a href="/learn/where-the-clock-enters/">03 Where the clock enters</a></li></ol></li><li><h3>Module 2 · Clock trees</h3><p>Skew and insertion delay, the global buffer network, and the trap of gating a clock with logic.</p><ol><li><a href="/learn/skew-and-insertion/">04 Skew and insertion delay</a></li><li><a href="/learn/global-buffers/">05 The global buffer</a></li><li><a href="/learn/gated-clock-trap/">06 The gated-clock trap</a></li></ol></li><li><h3>Module 3 · PLL and MMCM</h3><p>Multiply and divide one clock into another, move its phase in steps of the VCO, and which clocks the analyzer treats as related.</p><ol><li><a href="/learn/multiply-and-divide/">07 Multiply and divide</a></li><li><a href="/learn/phase-and-lock/">08 Phase and lock</a></li><li><a href="/learn/related-or-not/">09 Related or not</a></li></ol></li><li><h3>Module 4 · Resets</h3><p>Assert asynchronously, release synchronously: the three reset kinds, the release pipe, and the tree a reset grows.</p><ol><li class="on" aria-current="page"><a href="/learn/sync-or-async/">10 Sync or async</a></li><li><a href="/learn/reset-release/">11 The release pipe</a></li><li><a href="/learn/reset-fanout/">12 Reset fanout</a></li></ol></li><li><h3>Module 5 · Metastability</h3><p>The setup-hold window, the mean time between failures in integer arithmetic, and the two flops that fix it.</p><ol><li><a href="/learn/setup-hold-window/">13 The setup-hold window</a></li><li><a href="/learn/mtbf/">14 Mean time between failures</a></li><li><a href="/learn/two-flop-sync/">15 The two-flop synchronizer</a></li></ol></li><li><h3>Module 6 · Crossing many bits</h3><p>Why a binary bus tears, why Gray code does not, and the handshake that moves a pulse between worlds.</p><ol><li><a href="/learn/why-binary-fails/">16 Why binary fails</a></li><li><a href="/learn/gray-code/">17 Gray code</a></li><li><a href="/learn/handshakes-and-pulses/">18 Handshakes and pulses</a></li></ol></li><li><h3>Module 7 · The asynchronous FIFO</h3><p>Pointers, flags and depth: the buffer that moves a stream between two clocks.</p><ol><li><a href="/learn/pointers-and-flags/">19 Pointers and flags</a></li><li><a href="/learn/depth-sizing/">20 Sizing the depth</a></li><li><a href="/learn/fifo-testbench/">21 The FIFO testbench</a></li></ol></li><li><h3>Module 8 · Constraints</h3><p>The lines that tell the analyzer what a clock is, which paths not to check, and what the pins must meet.</p><ol><li><a href="/learn/create-clock/">22 create_clock</a></li><li><a href="/learn/false-path-and-max-delay/">23 False paths and max delay</a></li><li><a href="/learn/io-timing/">24 IO timing</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>A CDC report, one crossing captured at the flip-flops, and the bitstream diff that closes the course.</p><ol><li><a href="/learn/cdc-report/">25 The CDC report</a></li><li><a href="/learn/capture-a-crossing/">26 Capture a crossing</a></li><li><a href="/learn/capstone-diff/">27 Capstone: change it and diff it</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/synthesis-warnings/index.html b/apps/website/public/learn/synthesis-warnings/index.html
index 0f90535db4..b701185d8e 100644
--- a/apps/website/public/learn/synthesis-warnings/index.html
+++ b/apps/website/public/learn/synthesis-warnings/index.html
@@ -195,5 +195,6 @@ <h1>Synthesis warnings that matter</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The chip</h3><p>What an FPGA is, which chip we use, and how its pins meet the board.</p><ol><li><a href="/learn/fpga-is-a-grid/">01 An FPGA is a grid you configure</a></li><li><a href="/learn/which-chip/">02 Which chip is this?</a></li><li><a href="/learn/pins-and-package/">03 Pins, balls and the package</a></li></ol></li><li><h3>Module 2 · Numbers in hardware</h3><p>Bits, trits and number formats, and what arithmetic costs in logic.</p><ol><li><a href="/learn/bits-and-trits/">04 Bits, trits and GF16</a></li><li><a href="/learn/what-arithmetic-costs/">05 What arithmetic costs</a></li><li><a href="/learn/how-much-fits/">06 How much fits on the chip</a></li></ol></li><li><h3>Module 3 · Your t27 program</h3><p>Write a spec, test it, and see why compiling is not the same as being right.</p><ol><li><a href="/learn/hello-t27/">07 Hello, t27</a></li><li><a href="/learn/tests-are-the-spec/">08 Tests are part of the spec</a></li><li><a href="/learn/compiles-is-not-correct/">09 Compiles is not the same as correct</a></li></ol></li><li><h3>Module 4 · Inside t27c</h3><p>How the compiler reads a spec and what it writes, including the native t27b.</p><ol><li><a href="/learn/compiler-stages/">10 The stages of t27c</a></li><li><a href="/learn/t27c-on-a-real-machine/">11 t27c on a real machine</a></li><li><a href="/learn/seven-backends-and-t27b/">12 Seven backends, and t27b</a></li></ol></li><li><h3>Module 5 · From spec to hardware</h3><p>The Verilog t27c writes, the cells it becomes, and what one LUT does.</p><ol><li><a href="/learn/spec-to-verilog/">13 From a spec to Verilog</a></li><li><a href="/learn/count-the-cells/">14 Count the cells</a></li><li><a href="/learn/one-lut-up-close/">15 One LUT, up close</a></li></ol></li><li><h3>Module 6 · Reading synthesis</h3><p>What yosys reports about your design, and which warnings matter.</p><ol><li class="on" aria-current="page"><a href="/learn/synthesis-warnings/">16 Synthesis warnings that matter</a></li><li><a href="/learn/state-machines/">17 State machines</a></li><li><a href="/learn/where-the-luts-go/">18 Where the LUTs go</a></li></ol></li><li><h3>Module 7 · Place, route, timing</h3><p>Where the cells land on the die, and whether the clock is met.</p><ol><li><a href="/learn/placement/">19 Placement: where the cells land</a></li><li><a href="/learn/logic-depth/">20 Logic depth</a></li><li><a href="/learn/timing-slack/">21 Timing and slack</a></li></ol></li><li><h3>Module 8 · The bitstream</h3><p>How a routed design becomes the bits the chip loads, with no Vivado.</p><ol><li><a href="/learn/fasm-to-frames/">22 From FASM to frames</a></li><li><a href="/learn/a-build-receipt/">23 A build, itemised</a></li><li><a href="/learn/read-a-bitstream/">24 Read a bitstream</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>Ask the chip who it is, load the bits, and check them.</p><ol><li><a href="/learn/jtag-asks-the-chip/">25 JTAG asks the chip</a></li><li><a href="/learn/same-bits-on-the-board/">26 The same bits, on the board</a></li><li><a href="/learn/capstone/">27 Capstone: out-guess yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/t27-number-format/index.html b/apps/website/public/learn/t27-number-format/index.html
index 69a074acc5..7a10a2ee43 100644
--- a/apps/website/public/learn/t27-number-format/index.html
+++ b/apps/website/public/learn/t27-number-format/index.html
@@ -112,5 +112,6 @@ <h1>T27: a number format of our own</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Lab: our own research</h3><p>A number format of our own, an honest scoreboard, and a model's tables multiplied on the board.</p><ol><li class="on" aria-current="page"><a href="/learn/t27-number-format/">01 T27: a number format of our own</a></li><li><a href="/learn/an-honest-scoreboard/">02 An honest scoreboard</a></li><li><a href="/learn/a-model-on-the-board/">03 A model's tables on the board</a></li></ol></li><li><h3>Module 2 · AI numbers: the MX block</h3><p>How AI chips keep weights in a few bits: one shared scale per block, the scale byte itself, and what one outlier does to its neighbours.</p><ol><li><a href="/learn/one-scale-per-block/">04 One scale for a block of weights</a></li><li><a href="/learn/scale-byte-native/">05 The scale byte on a real machine</a></li><li><a href="/learn/one-outlier/">06 One outlier, many zeros</a></li></ol></li><li><h3>Module 3 · Ternary weights</h3><p>Weights that are only minus, zero or plus a scale, the five rules a ternary alphabet must pass, and a test pass that checked nothing.</p><ol><li><a href="/learn/phi-scaled-ternary/">07 Weights of minus phi, zero and plus phi</a></li><li><a href="/learn/a-pass-that-checked-nothing/">08 A pass that checked nothing</a></li><li><a href="/learn/five-rules-for-weights/">09 Five rules for a weight alphabet</a></li></ol></li><li><h3>Module 4 · The Ternary Network Float</h3><p>A rule the compiler enforces before any test runs, and a 17-bit float whose exponent is four balanced trits.</p><ol><li><a href="/learn/blocked-is-not-failed/">10 Blocked is not failed</a></li><li><a href="/learn/four-trit-exponent/">11 A float whose exponent is four trits</a></li><li><a href="/learn/negate-twice/">12 Negate twice, get the number back</a></li></ol></li><li><h3>Module 5 · Arithmetic on signed numbers</h3><p>Multiply two signed numbers, add them when their signs differ, and do both at once in a multiply-accumulate.</p><ol><li><a href="/learn/a-sign-is-one-bit/">13 A sign is one bit</a></li><li><a href="/learn/opposite-signs-subtract/">14 Opposite signs subtract</a></li><li><a href="/learn/multiply-then-add/">15 Multiply, then add</a></li></ol></li><li><h3>Module 6 · Parts of a neuron</h3><p>A ReLU that bends at zero, a power of two for softmax, and an argmax that names the answer.</p><ol><li><a href="/learn/a-bend-at-zero/">16 A bend at zero</a></li><li><a href="/learn/two-to-the-x/">17 Two to the x</a></li><li><a href="/learn/pick-the-largest/">18 Pick the largest</a></li></ol></li><li><h3>Module 7 · Learning from a mistake</h3><p>A loss that prices a wrong guess in bits, one step that moves a weight against its gradient, and the hidden layer that XOR needs.</p><ol><li><a href="/learn/a-loss-in-bits/">19 A loss in bits</a></li><li><a href="/learn/one-step-downhill/">20 One step downhill</a></li><li><a href="/learn/xor-needs-a-bend/">21 XOR needs a bend</a></li></ol></li><li><h3>Module 8 · BitNet: ternary networks</h3><p>A threshold that squeezes a sum back to three values, one neuron that becomes a different function when its weights change, and a neuron that reads its inputs 27 trits at a time.</p><ol><li><a href="/learn/back-to-three-values/">22 Back to three values</a></li><li><a href="/learn/same-neuron-new-weights/">23 Same neuron, new weights</a></li><li><a href="/learn/a-neuron-in-chunks/">24 A neuron in chunks</a></li></ol></li><li><h3>Module 9 · The ternary MAC as a chip</h3><p>The 27-trit dot product as wires with no register, the same sum added into a register on every clock, and a small whole network to close the course.</p><ol><li><a href="/learn/a-dot-product-in-wires/">25 A dot product made of wires</a></li><li><a href="/learn/add-it-up-every-clock/">26 Add it up on every clock</a></li><li><a href="/learn/a-whole-network/">27 A whole network</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/t27c-on-a-real-machine/index.html b/apps/website/public/learn/t27c-on-a-real-machine/index.html
index d29eba4e96..2febbf9e2f 100644
--- a/apps/website/public/learn/t27c-on-a-real-machine/index.html
+++ b/apps/website/public/learn/t27c-on-a-real-machine/index.html
@@ -171,5 +171,6 @@ <h1>t27c on a real machine</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The chip</h3><p>What an FPGA is, which chip we use, and how its pins meet the board.</p><ol><li><a href="/learn/fpga-is-a-grid/">01 An FPGA is a grid you configure</a></li><li><a href="/learn/which-chip/">02 Which chip is this?</a></li><li><a href="/learn/pins-and-package/">03 Pins, balls and the package</a></li></ol></li><li><h3>Module 2 · Numbers in hardware</h3><p>Bits, trits and number formats, and what arithmetic costs in logic.</p><ol><li><a href="/learn/bits-and-trits/">04 Bits, trits and GF16</a></li><li><a href="/learn/what-arithmetic-costs/">05 What arithmetic costs</a></li><li><a href="/learn/how-much-fits/">06 How much fits on the chip</a></li></ol></li><li><h3>Module 3 · Your t27 program</h3><p>Write a spec, test it, and see why compiling is not the same as being right.</p><ol><li><a href="/learn/hello-t27/">07 Hello, t27</a></li><li><a href="/learn/tests-are-the-spec/">08 Tests are part of the spec</a></li><li><a href="/learn/compiles-is-not-correct/">09 Compiles is not the same as correct</a></li></ol></li><li><h3>Module 4 · Inside t27c</h3><p>How the compiler reads a spec and what it writes, including the native t27b.</p><ol><li><a href="/learn/compiler-stages/">10 The stages of t27c</a></li><li class="on" aria-current="page"><a href="/learn/t27c-on-a-real-machine/">11 t27c on a real machine</a></li><li><a href="/learn/seven-backends-and-t27b/">12 Seven backends, and t27b</a></li></ol></li><li><h3>Module 5 · From spec to hardware</h3><p>The Verilog t27c writes, the cells it becomes, and what one LUT does.</p><ol><li><a href="/learn/spec-to-verilog/">13 From a spec to Verilog</a></li><li><a href="/learn/count-the-cells/">14 Count the cells</a></li><li><a href="/learn/one-lut-up-close/">15 One LUT, up close</a></li></ol></li><li><h3>Module 6 · Reading synthesis</h3><p>What yosys reports about your design, and which warnings matter.</p><ol><li><a href="/learn/synthesis-warnings/">16 Synthesis warnings that matter</a></li><li><a href="/learn/state-machines/">17 State machines</a></li><li><a href="/learn/where-the-luts-go/">18 Where the LUTs go</a></li></ol></li><li><h3>Module 7 · Place, route, timing</h3><p>Where the cells land on the die, and whether the clock is met.</p><ol><li><a href="/learn/placement/">19 Placement: where the cells land</a></li><li><a href="/learn/logic-depth/">20 Logic depth</a></li><li><a href="/learn/timing-slack/">21 Timing and slack</a></li></ol></li><li><h3>Module 8 · The bitstream</h3><p>How a routed design becomes the bits the chip loads, with no Vivado.</p><ol><li><a href="/learn/fasm-to-frames/">22 From FASM to frames</a></li><li><a href="/learn/a-build-receipt/">23 A build, itemised</a></li><li><a href="/learn/read-a-bitstream/">24 Read a bitstream</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>Ask the chip who it is, load the bits, and check them.</p><ol><li><a href="/learn/jtag-asks-the-chip/">25 JTAG asks the chip</a></li><li><a href="/learn/same-bits-on-the-board/">26 The same bits, on the board</a></li><li><a href="/learn/capstone/">27 Capstone: out-guess yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/tables-and-interpolation/index.html b/apps/website/public/learn/tables-and-interpolation/index.html
index 52c7f26573..5043b5c8e7 100644
--- a/apps/website/public/learn/tables-and-interpolation/index.html
+++ b/apps/website/public/learn/tables-and-interpolation/index.html
@@ -439,5 +439,6 @@ <h1>Tables and interpolation</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Adding numbers</h3><p>Why addition is slow: the carry that walks up the word, the dedicated chain in every slice, prefix networks and carry-save trees.</p><ol><li><a href="/learn/the-carry-chain/">01 The carry chain</a></li><li><a href="/learn/prefix-adders/">02 Looking ahead: prefix adders</a></li><li><a href="/learn/carry-save/">03 Three numbers at once</a></li></ol></li><li><h3>Module 2 · Multiplying</h3><p>A product is a sum of shifted copies; Booth recoding halves them, and the DSP48E1 slice does the rest in one block.</p><ol><li><a href="/learn/shift-and-add/">04 Shift and add</a></li><li><a href="/learn/booths-trick/">05 Booth's trick</a></li><li><a href="/learn/the-dsp48e1-slice/">06 The DSP48E1 slice</a></li></ol></li><li><h3>Module 3 · Fixed point</h3><p>Where the binary point sits, what rounding does to a value and to its average, and what each bit of a quantizer buys.</p><ol><li><a href="/learn/q-formats/">07 Q formats</a></li><li><a href="/learn/rounding-and-saturation/">08 Rounding and saturation</a></li><li><a href="/learn/quantization-noise/">09 Quantization noise</a></li></ol></li><li><h3>Module 4 · Functions in hardware</h3><p>Sine, cosine, angle and length from shifts and adds, and when a table is the better answer.</p><ol><li><a href="/learn/rotating-by-shifts/">10 Rotating by shifts</a></li><li><a href="/learn/angle-and-magnitude/">11 Angle and magnitude</a></li><li class="on" aria-current="page"><a href="/learn/tables-and-interpolation/">12 Tables and interpolation</a></li></ol></li><li><h3>Module 5 · Signals and sampling</h3><p>What sampling does to a frequency, how an oscillator is built from an adder, and what a DFT bin measures.</p><ol><li><a href="/learn/sampling-and-aliasing/">13 Sampling and aliasing</a></li><li><a href="/learn/the-nco/">14 A numerically controlled oscillator</a></li><li><a href="/learn/bins-of-a-spectrum/">15 Bins of a spectrum</a></li></ol></li><li><h3>Module 6 · FIR filters</h3><p>The moving average, a windowed-sinc design with integer taps, and folding the taps onto DSP slices.</p><ol><li><a href="/learn/the-moving-average/">16 The moving average</a></li><li><a href="/learn/designing-taps/">17 Designing taps</a></li><li><a href="/learn/folding-taps-onto-slices/">18 Folding taps onto slices</a></li></ol></li><li><h3>Module 7 · Multirate</h3><p>Lowering the sample rate without folding noise in: decimation, the CIC filter and the polyphase form.</p><ol><li><a href="/learn/lowering-the-rate/">19 Lowering the rate</a></li><li><a href="/learn/the-cic-filter/">20 The CIC filter</a></li><li><a href="/learn/polyphase/">21 Polyphase</a></li></ol></li><li><h3>Module 8 · The FFT</h3><p>N log N instead of N^2: butterflies, the bit-reversed input order and the bits each stage adds.</p><ol><li><a href="/learn/the-butterfly/">22 The butterfly</a></li><li><a href="/learn/twiddles-and-bit-reversal/">23 Twiddles and bit reversal</a></li><li><a href="/learn/growth-through-the-stages/">24 Growth through the stages</a></li></ol></li><li><h3>Module 9 · On the bench</h3><p>From the arithmetic to the board: a correlator from the modem spec, a budget of slices and timing, and the capstone filter.</p><ol><li><a href="/learn/a-correlator-finds-the-frame/">25 A correlator finds the frame</a></li><li><a href="/learn/does-it-fit-and-close-timing/">26 Does it fit and close timing?</a></li><li><a href="/learn/capstone-a-filter-on-the-board/">27 Capstone: a filter on the board</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/ternary-computing/index.html b/apps/website/public/learn/ternary-computing/index.html
index 0e534023fd..3595b8a3ee 100644
--- a/apps/website/public/learn/ternary-computing/index.html
+++ b/apps/website/public/learn/ternary-computing/index.html
@@ -67,5 +67,6 @@ <h1>The ternary machine: from a trit to a running program</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Three values</h3><p>Why three, how balanced ternary writes every number without a sign, and what flipping and cutting trits do.</p><ol><li><a href="/learn/why-three/">01 Why three</a></li><li><a href="/learn/balanced-ternary/">02 Balanced ternary</a></li><li><a href="/learn/flip-and-cut/">03 Flip to negate, cut to round</a></li></ol></li><li><h3>Module 2 · Logic with unknown</h3><p>Kleene's three-valued gates, two gates binary has no twin for, and addition as a pair of tables.</p><ol><li><a href="/learn/kleene-logic/">04 Three-valued logic</a></li><li><a href="/learn/consensus-and-any/">05 Consensus and any</a></li><li><a href="/learn/sum-and-carry/">06 Sum and carry</a></li></ol></li><li><h3>Module 3 · Adding trits</h3><p>The half adder, the full adder and a carry that ripples left, one place at a time.</p><ol><li><a href="/learn/half-adder/">07 The half adder</a></li><li><a href="/learn/full-adder/">08 The full adder</a></li><li><a href="/learn/ripple-carry/">09 Ripple carry</a></li></ol></li><li><h3>Module 4 · Multiplying without a multiplier</h3><p>Copy, drop or flip: a product by one trit, long multiplication, and a MAC that only adds.</p><ol><li><a href="/learn/times-a-trit/">10 Times one trit</a></li><li><a href="/learn/long-multiplication/">11 Long multiplication</a></li><li><a href="/learn/ternary-mac/">12 A ternary MAC</a></li></ol></li><li><h3>Module 5 · Trits in binary memory</h3><p>Two bits per trit, five trits per byte, and the bits a 27-trit word needs.</p><ol><li><a href="/learn/two-bits-per-trit/">13 Two bits per trit</a></li><li><a href="/learn/five-trits-a-byte/">14 Five trits in a byte</a></li><li><a href="/learn/a-word-of-27/">15 A word of 27 trits</a></li></ol></li><li><h3>Module 6 · The TRI-27 instruction word</h3><p>The 32-bit word the Trinity emulator decodes: its fields, its 47 opcodes and its 15-bit immediate.</p><ol><li><a href="/learn/the-instruction-word/">16 The instruction word</a></li><li><a href="/learn/the-opcode-map/">17 The opcode map</a></li><li><a href="/learn/the-immediate/">18 The immediate</a></li></ol></li><li><h3>Module 7 · A ternary machine</h3><p>An ALU built from this course's adders, a three-way jump, and a program you can step.</p><ol><li><a href="/learn/a-ternary-alu/">19 A ternary ALU</a></li><li><a href="/learn/three-way-jump/">20 A three-way jump</a></li><li><a href="/learn/run-a-program/">21 Run a program</a></li></ol></li><li><h3>Module 8 · Three answers</h3><p>Compare at the highest differing trit, find a number in thirds, sort with three-way compares.</p><ol><li><a href="/learn/compare-at-the-top/">22 Compare at the top</a></li><li><a href="/learn/questions-in-thirds/">23 Questions in thirds</a></li><li><a href="/learn/sort-by-three/">24 Sort with three-way compares</a></li></ol></li><li><h3>Module 9 · Ternary neurons</h3><p>Weights of -1, 0, +1: one neuron, a detector and a small layer, with no multiplier anywhere.</p><ol><li><a href="/learn/one-ternary-neuron/">25 One ternary neuron</a></li><li><a href="/learn/a-bar-detector/">26 A bar detector</a></li><li><a href="/learn/a-tiny-layer/">27 A tiny layer</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/ternary-mac/index.html b/apps/website/public/learn/ternary-mac/index.html
index 6708ae7a87..00467e2187 100644
--- a/apps/website/public/learn/ternary-mac/index.html
+++ b/apps/website/public/learn/ternary-mac/index.html
@@ -115,5 +115,6 @@ <h1>A ternary MAC</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Three values</h3><p>Why three, how balanced ternary writes every number without a sign, and what flipping and cutting trits do.</p><ol><li><a href="/learn/why-three/">01 Why three</a></li><li><a href="/learn/balanced-ternary/">02 Balanced ternary</a></li><li><a href="/learn/flip-and-cut/">03 Flip to negate, cut to round</a></li></ol></li><li><h3>Module 2 · Logic with unknown</h3><p>Kleene's three-valued gates, two gates binary has no twin for, and addition as a pair of tables.</p><ol><li><a href="/learn/kleene-logic/">04 Three-valued logic</a></li><li><a href="/learn/consensus-and-any/">05 Consensus and any</a></li><li><a href="/learn/sum-and-carry/">06 Sum and carry</a></li></ol></li><li><h3>Module 3 · Adding trits</h3><p>The half adder, the full adder and a carry that ripples left, one place at a time.</p><ol><li><a href="/learn/half-adder/">07 The half adder</a></li><li><a href="/learn/full-adder/">08 The full adder</a></li><li><a href="/learn/ripple-carry/">09 Ripple carry</a></li></ol></li><li><h3>Module 4 · Multiplying without a multiplier</h3><p>Copy, drop or flip: a product by one trit, long multiplication, and a MAC that only adds.</p><ol><li><a href="/learn/times-a-trit/">10 Times one trit</a></li><li><a href="/learn/long-multiplication/">11 Long multiplication</a></li><li class="on" aria-current="page"><a href="/learn/ternary-mac/">12 A ternary MAC</a></li></ol></li><li><h3>Module 5 · Trits in binary memory</h3><p>Two bits per trit, five trits per byte, and the bits a 27-trit word needs.</p><ol><li><a href="/learn/two-bits-per-trit/">13 Two bits per trit</a></li><li><a href="/learn/five-trits-a-byte/">14 Five trits in a byte</a></li><li><a href="/learn/a-word-of-27/">15 A word of 27 trits</a></li></ol></li><li><h3>Module 6 · The TRI-27 instruction word</h3><p>The 32-bit word the Trinity emulator decodes: its fields, its 47 opcodes and its 15-bit immediate.</p><ol><li><a href="/learn/the-instruction-word/">16 The instruction word</a></li><li><a href="/learn/the-opcode-map/">17 The opcode map</a></li><li><a href="/learn/the-immediate/">18 The immediate</a></li></ol></li><li><h3>Module 7 · A ternary machine</h3><p>An ALU built from this course's adders, a three-way jump, and a program you can step.</p><ol><li><a href="/learn/a-ternary-alu/">19 A ternary ALU</a></li><li><a href="/learn/three-way-jump/">20 A three-way jump</a></li><li><a href="/learn/run-a-program/">21 Run a program</a></li></ol></li><li><h3>Module 8 · Three answers</h3><p>Compare at the highest differing trit, find a number in thirds, sort with three-way compares.</p><ol><li><a href="/learn/compare-at-the-top/">22 Compare at the top</a></li><li><a href="/learn/questions-in-thirds/">23 Questions in thirds</a></li><li><a href="/learn/sort-by-three/">24 Sort with three-way compares</a></li></ol></li><li><h3>Module 9 · Ternary neurons</h3><p>Weights of -1, 0, +1: one neuron, a detector and a small layer, with no multiplier anywhere.</p><ol><li><a href="/learn/one-ternary-neuron/">25 One ternary neuron</a></li><li><a href="/learn/a-bar-detector/">26 A bar detector</a></li><li><a href="/learn/a-tiny-layer/">27 A tiny layer</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/test-strength/index.html b/apps/website/public/learn/test-strength/index.html
index 5bc3a893b3..bef991f976 100644
--- a/apps/website/public/learn/test-strength/index.html
+++ b/apps/website/public/learn/test-strength/index.html
@@ -110,5 +110,6 @@ <h1>Test strength</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Why verify</h3><p>Designs that compile and are wrong, the model that decides, and the plan written before the code.</p><ol><li><a href="/learn/bugs-that-compile/">01 Bugs that compile</a></li><li><a href="/learn/golden-model/">02 The golden model</a></li><li><a href="/learn/plan-before-code/">03 Plan before code</a></li></ol></li><li><h3>Module 2 · Testbenches</h3><p>Stimulus, checks and a verdict, written as one spec beside the design it judges.</p><ol><li><a href="/learn/anatomy-of-a-tb/">04 Anatomy of a testbench</a></li><li><a href="/learn/self-checking/">05 Self-checking, not staring</a></li><li><a href="/learn/directed-vs-random/">06 Directed versus random</a></li></ol></li><li><h3>Module 3 · Waveforms</h3><p>A trace of every signal, read the way a hardware engineer reads it, and two runs compared.</p><ol><li><a href="/learn/vcd-format/">07 The VCD format</a></li><li><a href="/learn/debug-with-waves/">08 Debug with waves</a></li><li><a href="/learn/compare-traces/">09 Compare two traces</a></li></ol></li><li><h3>Module 4 · Conformance vectors</h3><p>Cases with the answer written beside them, kept where the compiler can reach them.</p><ol><li><a href="/learn/what-a-vector-is/">10 What a vector is</a></li><li><a href="/learn/golden-ruler-vectors/">11 Golden ruler vectors</a></li><li><a href="/learn/end-to-end-scenarios/">12 End-to-end scenarios</a></li></ol></li><li><h3>Module 5 · Cosimulation</h3><p>Spec, simulator and board agreeing on the bench Artix-7 XC7A200T, and what to do when they do not.</p><ol><li><a href="/learn/three-way-agreement/">13 Three-way agreement</a></li><li><a href="/learn/simulator-internals/">14 Inside the simulator</a></li><li><a href="/learn/when-they-disagree/">15 When they disagree</a></li></ol></li><li><h3>Module 6 · Coverage</h3><p>What the tests touched: lines, toggles, states, and what that number hides.</p><ol><li><a href="/learn/line-and-toggle/">16 Line and toggle</a></li><li><a href="/learn/fsm-coverage/">17 FSM coverage</a></li><li><a href="/learn/coverage-lies/">18 Coverage lies</a></li></ol></li><li><h3>Module 7 · Formal</h3><p>Assertions that hold every cycle, bounded search for a counterexample, and why a proof needs induction.</p><ol><li><a href="/learn/assertions/">19 Assertions</a></li><li><a href="/learn/bounded-model-checking/">20 Bounded model checking</a></li><li><a href="/learn/induction/">21 Induction</a></li></ol></li><li><h3>Module 8 · Mutation</h3><p>Break the design on purpose and count what the tests catch.</p><ol><li><a href="/learn/kill-the-mutant/">22 Kill the mutant</a></li><li><a href="/learn/surviving-mutants/">23 Surviving mutants</a></li><li class="on" aria-current="page"><a href="/learn/test-strength/">24 Test strength</a></li></ol></li><li><h3>Module 9 · Sign-off</h3><p>One command, every receipt, a clean verdict you can show.</p><ol><li><a href="/learn/build-verify/">25 Build and verify</a></li><li><a href="/learn/board-selftest/">26 The board self-tests</a></li><li><a href="/learn/verify-capstone/">27 Capstone: verify a module whole</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/tests-are-the-spec/index.html b/apps/website/public/learn/tests-are-the-spec/index.html
index da3426b51f..0627155ef4 100644
--- a/apps/website/public/learn/tests-are-the-spec/index.html
+++ b/apps/website/public/learn/tests-are-the-spec/index.html
@@ -156,5 +156,6 @@ <h1>Tests are part of the spec</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The chip</h3><p>What an FPGA is, which chip we use, and how its pins meet the board.</p><ol><li><a href="/learn/fpga-is-a-grid/">01 An FPGA is a grid you configure</a></li><li><a href="/learn/which-chip/">02 Which chip is this?</a></li><li><a href="/learn/pins-and-package/">03 Pins, balls and the package</a></li></ol></li><li><h3>Module 2 · Numbers in hardware</h3><p>Bits, trits and number formats, and what arithmetic costs in logic.</p><ol><li><a href="/learn/bits-and-trits/">04 Bits, trits and GF16</a></li><li><a href="/learn/what-arithmetic-costs/">05 What arithmetic costs</a></li><li><a href="/learn/how-much-fits/">06 How much fits on the chip</a></li></ol></li><li><h3>Module 3 · Your t27 program</h3><p>Write a spec, test it, and see why compiling is not the same as being right.</p><ol><li><a href="/learn/hello-t27/">07 Hello, t27</a></li><li class="on" aria-current="page"><a href="/learn/tests-are-the-spec/">08 Tests are part of the spec</a></li><li><a href="/learn/compiles-is-not-correct/">09 Compiles is not the same as correct</a></li></ol></li><li><h3>Module 4 · Inside t27c</h3><p>How the compiler reads a spec and what it writes, including the native t27b.</p><ol><li><a href="/learn/compiler-stages/">10 The stages of t27c</a></li><li><a href="/learn/t27c-on-a-real-machine/">11 t27c on a real machine</a></li><li><a href="/learn/seven-backends-and-t27b/">12 Seven backends, and t27b</a></li></ol></li><li><h3>Module 5 · From spec to hardware</h3><p>The Verilog t27c writes, the cells it becomes, and what one LUT does.</p><ol><li><a href="/learn/spec-to-verilog/">13 From a spec to Verilog</a></li><li><a href="/learn/count-the-cells/">14 Count the cells</a></li><li><a href="/learn/one-lut-up-close/">15 One LUT, up close</a></li></ol></li><li><h3>Module 6 · Reading synthesis</h3><p>What yosys reports about your design, and which warnings matter.</p><ol><li><a href="/learn/synthesis-warnings/">16 Synthesis warnings that matter</a></li><li><a href="/learn/state-machines/">17 State machines</a></li><li><a href="/learn/where-the-luts-go/">18 Where the LUTs go</a></li></ol></li><li><h3>Module 7 · Place, route, timing</h3><p>Where the cells land on the die, and whether the clock is met.</p><ol><li><a href="/learn/placement/">19 Placement: where the cells land</a></li><li><a href="/learn/logic-depth/">20 Logic depth</a></li><li><a href="/learn/timing-slack/">21 Timing and slack</a></li></ol></li><li><h3>Module 8 · The bitstream</h3><p>How a routed design becomes the bits the chip loads, with no Vivado.</p><ol><li><a href="/learn/fasm-to-frames/">22 From FASM to frames</a></li><li><a href="/learn/a-build-receipt/">23 A build, itemised</a></li><li><a href="/learn/read-a-bitstream/">24 Read a bitstream</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>Ask the chip who it is, load the bits, and check them.</p><ol><li><a href="/learn/jtag-asks-the-chip/">25 JTAG asks the chip</a></li><li><a href="/learn/same-bits-on-the-board/">26 The same bits, on the board</a></li><li><a href="/learn/capstone/">27 Capstone: out-guess yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/the-butterfly/index.html b/apps/website/public/learn/the-butterfly/index.html
index 354255fb92..8abf3b3fca 100644
--- a/apps/website/public/learn/the-butterfly/index.html
+++ b/apps/website/public/learn/the-butterfly/index.html
@@ -1122,5 +1122,6 @@ <h1>The butterfly</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Adding numbers</h3><p>Why addition is slow: the carry that walks up the word, the dedicated chain in every slice, prefix networks and carry-save trees.</p><ol><li><a href="/learn/the-carry-chain/">01 The carry chain</a></li><li><a href="/learn/prefix-adders/">02 Looking ahead: prefix adders</a></li><li><a href="/learn/carry-save/">03 Three numbers at once</a></li></ol></li><li><h3>Module 2 · Multiplying</h3><p>A product is a sum of shifted copies; Booth recoding halves them, and the DSP48E1 slice does the rest in one block.</p><ol><li><a href="/learn/shift-and-add/">04 Shift and add</a></li><li><a href="/learn/booths-trick/">05 Booth's trick</a></li><li><a href="/learn/the-dsp48e1-slice/">06 The DSP48E1 slice</a></li></ol></li><li><h3>Module 3 · Fixed point</h3><p>Where the binary point sits, what rounding does to a value and to its average, and what each bit of a quantizer buys.</p><ol><li><a href="/learn/q-formats/">07 Q formats</a></li><li><a href="/learn/rounding-and-saturation/">08 Rounding and saturation</a></li><li><a href="/learn/quantization-noise/">09 Quantization noise</a></li></ol></li><li><h3>Module 4 · Functions in hardware</h3><p>Sine, cosine, angle and length from shifts and adds, and when a table is the better answer.</p><ol><li><a href="/learn/rotating-by-shifts/">10 Rotating by shifts</a></li><li><a href="/learn/angle-and-magnitude/">11 Angle and magnitude</a></li><li><a href="/learn/tables-and-interpolation/">12 Tables and interpolation</a></li></ol></li><li><h3>Module 5 · Signals and sampling</h3><p>What sampling does to a frequency, how an oscillator is built from an adder, and what a DFT bin measures.</p><ol><li><a href="/learn/sampling-and-aliasing/">13 Sampling and aliasing</a></li><li><a href="/learn/the-nco/">14 A numerically controlled oscillator</a></li><li><a href="/learn/bins-of-a-spectrum/">15 Bins of a spectrum</a></li></ol></li><li><h3>Module 6 · FIR filters</h3><p>The moving average, a windowed-sinc design with integer taps, and folding the taps onto DSP slices.</p><ol><li><a href="/learn/the-moving-average/">16 The moving average</a></li><li><a href="/learn/designing-taps/">17 Designing taps</a></li><li><a href="/learn/folding-taps-onto-slices/">18 Folding taps onto slices</a></li></ol></li><li><h3>Module 7 · Multirate</h3><p>Lowering the sample rate without folding noise in: decimation, the CIC filter and the polyphase form.</p><ol><li><a href="/learn/lowering-the-rate/">19 Lowering the rate</a></li><li><a href="/learn/the-cic-filter/">20 The CIC filter</a></li><li><a href="/learn/polyphase/">21 Polyphase</a></li></ol></li><li><h3>Module 8 · The FFT</h3><p>N log N instead of N^2: butterflies, the bit-reversed input order and the bits each stage adds.</p><ol><li class="on" aria-current="page"><a href="/learn/the-butterfly/">22 The butterfly</a></li><li><a href="/learn/twiddles-and-bit-reversal/">23 Twiddles and bit reversal</a></li><li><a href="/learn/growth-through-the-stages/">24 Growth through the stages</a></li></ol></li><li><h3>Module 9 · On the bench</h3><p>From the arithmetic to the board: a correlator from the modem spec, a budget of slices and timing, and the capstone filter.</p><ol><li><a href="/learn/a-correlator-finds-the-frame/">25 A correlator finds the frame</a></li><li><a href="/learn/does-it-fit-and-close-timing/">26 Does it fit and close timing?</a></li><li><a href="/learn/capstone-a-filter-on-the-board/">27 Capstone: a filter on the board</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/the-carry-chain/index.html b/apps/website/public/learn/the-carry-chain/index.html
index b4c11c3da2..026dd4c188 100644
--- a/apps/website/public/learn/the-carry-chain/index.html
+++ b/apps/website/public/learn/the-carry-chain/index.html
@@ -554,5 +554,6 @@ <h1>The carry chain</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Adding numbers</h3><p>Why addition is slow: the carry that walks up the word, the dedicated chain in every slice, prefix networks and carry-save trees.</p><ol><li class="on" aria-current="page"><a href="/learn/the-carry-chain/">01 The carry chain</a></li><li><a href="/learn/prefix-adders/">02 Looking ahead: prefix adders</a></li><li><a href="/learn/carry-save/">03 Three numbers at once</a></li></ol></li><li><h3>Module 2 · Multiplying</h3><p>A product is a sum of shifted copies; Booth recoding halves them, and the DSP48E1 slice does the rest in one block.</p><ol><li><a href="/learn/shift-and-add/">04 Shift and add</a></li><li><a href="/learn/booths-trick/">05 Booth's trick</a></li><li><a href="/learn/the-dsp48e1-slice/">06 The DSP48E1 slice</a></li></ol></li><li><h3>Module 3 · Fixed point</h3><p>Where the binary point sits, what rounding does to a value and to its average, and what each bit of a quantizer buys.</p><ol><li><a href="/learn/q-formats/">07 Q formats</a></li><li><a href="/learn/rounding-and-saturation/">08 Rounding and saturation</a></li><li><a href="/learn/quantization-noise/">09 Quantization noise</a></li></ol></li><li><h3>Module 4 · Functions in hardware</h3><p>Sine, cosine, angle and length from shifts and adds, and when a table is the better answer.</p><ol><li><a href="/learn/rotating-by-shifts/">10 Rotating by shifts</a></li><li><a href="/learn/angle-and-magnitude/">11 Angle and magnitude</a></li><li><a href="/learn/tables-and-interpolation/">12 Tables and interpolation</a></li></ol></li><li><h3>Module 5 · Signals and sampling</h3><p>What sampling does to a frequency, how an oscillator is built from an adder, and what a DFT bin measures.</p><ol><li><a href="/learn/sampling-and-aliasing/">13 Sampling and aliasing</a></li><li><a href="/learn/the-nco/">14 A numerically controlled oscillator</a></li><li><a href="/learn/bins-of-a-spectrum/">15 Bins of a spectrum</a></li></ol></li><li><h3>Module 6 · FIR filters</h3><p>The moving average, a windowed-sinc design with integer taps, and folding the taps onto DSP slices.</p><ol><li><a href="/learn/the-moving-average/">16 The moving average</a></li><li><a href="/learn/designing-taps/">17 Designing taps</a></li><li><a href="/learn/folding-taps-onto-slices/">18 Folding taps onto slices</a></li></ol></li><li><h3>Module 7 · Multirate</h3><p>Lowering the sample rate without folding noise in: decimation, the CIC filter and the polyphase form.</p><ol><li><a href="/learn/lowering-the-rate/">19 Lowering the rate</a></li><li><a href="/learn/the-cic-filter/">20 The CIC filter</a></li><li><a href="/learn/polyphase/">21 Polyphase</a></li></ol></li><li><h3>Module 8 · The FFT</h3><p>N log N instead of N^2: butterflies, the bit-reversed input order and the bits each stage adds.</p><ol><li><a href="/learn/the-butterfly/">22 The butterfly</a></li><li><a href="/learn/twiddles-and-bit-reversal/">23 Twiddles and bit reversal</a></li><li><a href="/learn/growth-through-the-stages/">24 Growth through the stages</a></li></ol></li><li><h3>Module 9 · On the bench</h3><p>From the arithmetic to the board: a correlator from the modem spec, a budget of slices and timing, and the capstone filter.</p><ol><li><a href="/learn/a-correlator-finds-the-frame/">25 A correlator finds the frame</a></li><li><a href="/learn/does-it-fit-and-close-timing/">26 Does it fit and close timing?</a></li><li><a href="/learn/capstone-a-filter-on-the-board/">27 Capstone: a filter on the board</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/the-cic-filter/index.html b/apps/website/public/learn/the-cic-filter/index.html
index f521aae3cb..d5a021f417 100644
--- a/apps/website/public/learn/the-cic-filter/index.html
+++ b/apps/website/public/learn/the-cic-filter/index.html
@@ -1122,5 +1122,6 @@ <h1>The CIC filter</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Adding numbers</h3><p>Why addition is slow: the carry that walks up the word, the dedicated chain in every slice, prefix networks and carry-save trees.</p><ol><li><a href="/learn/the-carry-chain/">01 The carry chain</a></li><li><a href="/learn/prefix-adders/">02 Looking ahead: prefix adders</a></li><li><a href="/learn/carry-save/">03 Three numbers at once</a></li></ol></li><li><h3>Module 2 · Multiplying</h3><p>A product is a sum of shifted copies; Booth recoding halves them, and the DSP48E1 slice does the rest in one block.</p><ol><li><a href="/learn/shift-and-add/">04 Shift and add</a></li><li><a href="/learn/booths-trick/">05 Booth's trick</a></li><li><a href="/learn/the-dsp48e1-slice/">06 The DSP48E1 slice</a></li></ol></li><li><h3>Module 3 · Fixed point</h3><p>Where the binary point sits, what rounding does to a value and to its average, and what each bit of a quantizer buys.</p><ol><li><a href="/learn/q-formats/">07 Q formats</a></li><li><a href="/learn/rounding-and-saturation/">08 Rounding and saturation</a></li><li><a href="/learn/quantization-noise/">09 Quantization noise</a></li></ol></li><li><h3>Module 4 · Functions in hardware</h3><p>Sine, cosine, angle and length from shifts and adds, and when a table is the better answer.</p><ol><li><a href="/learn/rotating-by-shifts/">10 Rotating by shifts</a></li><li><a href="/learn/angle-and-magnitude/">11 Angle and magnitude</a></li><li><a href="/learn/tables-and-interpolation/">12 Tables and interpolation</a></li></ol></li><li><h3>Module 5 · Signals and sampling</h3><p>What sampling does to a frequency, how an oscillator is built from an adder, and what a DFT bin measures.</p><ol><li><a href="/learn/sampling-and-aliasing/">13 Sampling and aliasing</a></li><li><a href="/learn/the-nco/">14 A numerically controlled oscillator</a></li><li><a href="/learn/bins-of-a-spectrum/">15 Bins of a spectrum</a></li></ol></li><li><h3>Module 6 · FIR filters</h3><p>The moving average, a windowed-sinc design with integer taps, and folding the taps onto DSP slices.</p><ol><li><a href="/learn/the-moving-average/">16 The moving average</a></li><li><a href="/learn/designing-taps/">17 Designing taps</a></li><li><a href="/learn/folding-taps-onto-slices/">18 Folding taps onto slices</a></li></ol></li><li><h3>Module 7 · Multirate</h3><p>Lowering the sample rate without folding noise in: decimation, the CIC filter and the polyphase form.</p><ol><li><a href="/learn/lowering-the-rate/">19 Lowering the rate</a></li><li class="on" aria-current="page"><a href="/learn/the-cic-filter/">20 The CIC filter</a></li><li><a href="/learn/polyphase/">21 Polyphase</a></li></ol></li><li><h3>Module 8 · The FFT</h3><p>N log N instead of N^2: butterflies, the bit-reversed input order and the bits each stage adds.</p><ol><li><a href="/learn/the-butterfly/">22 The butterfly</a></li><li><a href="/learn/twiddles-and-bit-reversal/">23 Twiddles and bit reversal</a></li><li><a href="/learn/growth-through-the-stages/">24 Growth through the stages</a></li></ol></li><li><h3>Module 9 · On the bench</h3><p>From the arithmetic to the board: a correlator from the modem spec, a budget of slices and timing, and the capstone filter.</p><ol><li><a href="/learn/a-correlator-finds-the-frame/">25 A correlator finds the frame</a></li><li><a href="/learn/does-it-fit-and-close-timing/">26 Does it fit and close timing?</a></li><li><a href="/learn/capstone-a-filter-on-the-board/">27 Capstone: a filter on the board</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/the-dsp48e1-slice/index.html b/apps/website/public/learn/the-dsp48e1-slice/index.html
index eb7c825a44..3fa3b7c904 100644
--- a/apps/website/public/learn/the-dsp48e1-slice/index.html
+++ b/apps/website/public/learn/the-dsp48e1-slice/index.html
@@ -372,5 +372,6 @@ <h1>The DSP48E1 slice</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Adding numbers</h3><p>Why addition is slow: the carry that walks up the word, the dedicated chain in every slice, prefix networks and carry-save trees.</p><ol><li><a href="/learn/the-carry-chain/">01 The carry chain</a></li><li><a href="/learn/prefix-adders/">02 Looking ahead: prefix adders</a></li><li><a href="/learn/carry-save/">03 Three numbers at once</a></li></ol></li><li><h3>Module 2 · Multiplying</h3><p>A product is a sum of shifted copies; Booth recoding halves them, and the DSP48E1 slice does the rest in one block.</p><ol><li><a href="/learn/shift-and-add/">04 Shift and add</a></li><li><a href="/learn/booths-trick/">05 Booth's trick</a></li><li class="on" aria-current="page"><a href="/learn/the-dsp48e1-slice/">06 The DSP48E1 slice</a></li></ol></li><li><h3>Module 3 · Fixed point</h3><p>Where the binary point sits, what rounding does to a value and to its average, and what each bit of a quantizer buys.</p><ol><li><a href="/learn/q-formats/">07 Q formats</a></li><li><a href="/learn/rounding-and-saturation/">08 Rounding and saturation</a></li><li><a href="/learn/quantization-noise/">09 Quantization noise</a></li></ol></li><li><h3>Module 4 · Functions in hardware</h3><p>Sine, cosine, angle and length from shifts and adds, and when a table is the better answer.</p><ol><li><a href="/learn/rotating-by-shifts/">10 Rotating by shifts</a></li><li><a href="/learn/angle-and-magnitude/">11 Angle and magnitude</a></li><li><a href="/learn/tables-and-interpolation/">12 Tables and interpolation</a></li></ol></li><li><h3>Module 5 · Signals and sampling</h3><p>What sampling does to a frequency, how an oscillator is built from an adder, and what a DFT bin measures.</p><ol><li><a href="/learn/sampling-and-aliasing/">13 Sampling and aliasing</a></li><li><a href="/learn/the-nco/">14 A numerically controlled oscillator</a></li><li><a href="/learn/bins-of-a-spectrum/">15 Bins of a spectrum</a></li></ol></li><li><h3>Module 6 · FIR filters</h3><p>The moving average, a windowed-sinc design with integer taps, and folding the taps onto DSP slices.</p><ol><li><a href="/learn/the-moving-average/">16 The moving average</a></li><li><a href="/learn/designing-taps/">17 Designing taps</a></li><li><a href="/learn/folding-taps-onto-slices/">18 Folding taps onto slices</a></li></ol></li><li><h3>Module 7 · Multirate</h3><p>Lowering the sample rate without folding noise in: decimation, the CIC filter and the polyphase form.</p><ol><li><a href="/learn/lowering-the-rate/">19 Lowering the rate</a></li><li><a href="/learn/the-cic-filter/">20 The CIC filter</a></li><li><a href="/learn/polyphase/">21 Polyphase</a></li></ol></li><li><h3>Module 8 · The FFT</h3><p>N log N instead of N^2: butterflies, the bit-reversed input order and the bits each stage adds.</p><ol><li><a href="/learn/the-butterfly/">22 The butterfly</a></li><li><a href="/learn/twiddles-and-bit-reversal/">23 Twiddles and bit reversal</a></li><li><a href="/learn/growth-through-the-stages/">24 Growth through the stages</a></li></ol></li><li><h3>Module 9 · On the bench</h3><p>From the arithmetic to the board: a correlator from the modem spec, a budget of slices and timing, and the capstone filter.</p><ol><li><a href="/learn/a-correlator-finds-the-frame/">25 A correlator finds the frame</a></li><li><a href="/learn/does-it-fit-and-close-timing/">26 Does it fit and close timing?</a></li><li><a href="/learn/capstone-a-filter-on-the-board/">27 Capstone: a filter on the board</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/the-frame/index.html b/apps/website/public/learn/the-frame/index.html
index a4fdbdf2a7..fd5985ab61 100644
--- a/apps/website/public/learn/the-frame/index.html
+++ b/apps/website/public/learn/the-frame/index.html
@@ -299,5 +299,6 @@ <h1>The frame: start, data, stop</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a bus is</h3><p>Why a bus exists at all: a conversation on wires, framed and addressed, and who is allowed to talk.</p><ol><li><a href="/learn/wires-that-talk/">01 Wires that talk</a></li><li><a href="/learn/framing-a-conversation/">02 Framing a conversation</a></li><li><a href="/learn/master-and-servant/">03 Master and servant</a></li></ol></li><li><h3>Module 2 · UART</h3><p>The two-wire bus with no clock: the frame, the divisor that sets the rate, and the status a driver polls.</p><ol><li class="on" aria-current="page"><a href="/learn/the-frame/">04 The frame: start, data, stop</a></li><li><a href="/learn/baud-and-the-divisor/">05 Baud and the divisor</a></li><li><a href="/learn/status-and-fifos/">06 Status and FIFOs</a></li></ol></li><li><h3>Module 3 · SPI</h3><p>The clocked conversation: four modes, a prescaler ladder, and a chip select per servant.</p><ol><li><a href="/learn/modes-cpol-cpha/">07 Modes: CPOL and CPHA</a></li><li><a href="/learn/the-prescaler-ladder/">08 The prescaler ladder</a></li><li><a href="/learn/chip-select-and-widths/">09 Chip select and widths</a></li></ol></li><li><h3>Module 4 · APB</h3><p>The register bus: PSEL and PENABLE, strobes and wait states, and how many address bits a peripheral count costs.</p><ol><li><a href="/learn/psel-and-penable/">10 PSEL and PENABLE</a></li><li><a href="/learn/strobes-and-waits/">11 Strobes and waits</a></li><li><a href="/learn/address-decode/">12 Address decode</a></li></ol></li><li><h3>Module 5 · AXI4</h3><p>The five channels: address, data and response in both directions, lite or full, bursts and IDs.</p><ol><li><a href="/learn/five-channels/">13 Five channels</a></li><li><a href="/learn/lite-or-full/">14 Lite or full</a></li><li><a href="/learn/bursts-and-ids/">15 Bursts and IDs</a></li></ol></li><li><h3>Module 6 · Memory</h3><p>What sits on the far side of every bus: memory maps, port kinds, and latency that a wait state must cover.</p><ol><li><a href="/learn/memory-maps/">16 Memory maps</a></li><li><a href="/learn/ports-and-kinds/">17 Ports and kinds</a></li><li><a href="/learn/latency/">18 Latency</a></li></ol></li><li><h3>Module 7 · Bridges</h3><p>Why designs grow more than one bus, and the packet bridge that moves work between them.</p><ol><li><a href="/learn/why-bridges/">19 Why bridges</a></li><li><a href="/learn/the-packet-bridge/">20 The packet bridge</a></li><li><a href="/learn/everything-in-one-bench/">21 Everything in one bench</a></li></ol></li><li><h3>Module 8 · Ethernet</h3><p>Frames, the frame check sequence, RGMII timing, and the pre-registered steps of a real bring-up.</p><ol><li><a href="/learn/frames-and-the-fcs/">22 Frames and the FCS</a></li><li><a href="/learn/rgmii-timing/">23 RGMII timing</a></li><li><a href="/learn/bring-up-on-the-bench/">24 Bring-up on the bench</a></li></ol></li><li><h3>Module 9 · The bench</h3><p>The discipline that guards real hardware: who holds the IO, taking and giving the lock, and what runs next.</p><ol><li><a href="/learn/who-holds-the-io/">25 Who holds the IO</a></li><li><a href="/learn/claim-and-release/">26 Claim and release</a></li><li><a href="/learn/capstone-what-runs-next/">27 Capstone: what runs next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/the-immediate/index.html b/apps/website/public/learn/the-immediate/index.html
index 5fe63597c1..c0b8240d3a 100644
--- a/apps/website/public/learn/the-immediate/index.html
+++ b/apps/website/public/learn/the-immediate/index.html
@@ -764,5 +764,6 @@ <h1>The immediate</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Three values</h3><p>Why three, how balanced ternary writes every number without a sign, and what flipping and cutting trits do.</p><ol><li><a href="/learn/why-three/">01 Why three</a></li><li><a href="/learn/balanced-ternary/">02 Balanced ternary</a></li><li><a href="/learn/flip-and-cut/">03 Flip to negate, cut to round</a></li></ol></li><li><h3>Module 2 · Logic with unknown</h3><p>Kleene's three-valued gates, two gates binary has no twin for, and addition as a pair of tables.</p><ol><li><a href="/learn/kleene-logic/">04 Three-valued logic</a></li><li><a href="/learn/consensus-and-any/">05 Consensus and any</a></li><li><a href="/learn/sum-and-carry/">06 Sum and carry</a></li></ol></li><li><h3>Module 3 · Adding trits</h3><p>The half adder, the full adder and a carry that ripples left, one place at a time.</p><ol><li><a href="/learn/half-adder/">07 The half adder</a></li><li><a href="/learn/full-adder/">08 The full adder</a></li><li><a href="/learn/ripple-carry/">09 Ripple carry</a></li></ol></li><li><h3>Module 4 · Multiplying without a multiplier</h3><p>Copy, drop or flip: a product by one trit, long multiplication, and a MAC that only adds.</p><ol><li><a href="/learn/times-a-trit/">10 Times one trit</a></li><li><a href="/learn/long-multiplication/">11 Long multiplication</a></li><li><a href="/learn/ternary-mac/">12 A ternary MAC</a></li></ol></li><li><h3>Module 5 · Trits in binary memory</h3><p>Two bits per trit, five trits per byte, and the bits a 27-trit word needs.</p><ol><li><a href="/learn/two-bits-per-trit/">13 Two bits per trit</a></li><li><a href="/learn/five-trits-a-byte/">14 Five trits in a byte</a></li><li><a href="/learn/a-word-of-27/">15 A word of 27 trits</a></li></ol></li><li><h3>Module 6 · The TRI-27 instruction word</h3><p>The 32-bit word the Trinity emulator decodes: its fields, its 47 opcodes and its 15-bit immediate.</p><ol><li><a href="/learn/the-instruction-word/">16 The instruction word</a></li><li><a href="/learn/the-opcode-map/">17 The opcode map</a></li><li class="on" aria-current="page"><a href="/learn/the-immediate/">18 The immediate</a></li></ol></li><li><h3>Module 7 · A ternary machine</h3><p>An ALU built from this course's adders, a three-way jump, and a program you can step.</p><ol><li><a href="/learn/a-ternary-alu/">19 A ternary ALU</a></li><li><a href="/learn/three-way-jump/">20 A three-way jump</a></li><li><a href="/learn/run-a-program/">21 Run a program</a></li></ol></li><li><h3>Module 8 · Three answers</h3><p>Compare at the highest differing trit, find a number in thirds, sort with three-way compares.</p><ol><li><a href="/learn/compare-at-the-top/">22 Compare at the top</a></li><li><a href="/learn/questions-in-thirds/">23 Questions in thirds</a></li><li><a href="/learn/sort-by-three/">24 Sort with three-way compares</a></li></ol></li><li><h3>Module 9 · Ternary neurons</h3><p>Weights of -1, 0, +1: one neuron, a detector and a small layer, with no multiplier anywhere.</p><ol><li><a href="/learn/one-ternary-neuron/">25 One ternary neuron</a></li><li><a href="/learn/a-bar-detector/">26 A bar detector</a></li><li><a href="/learn/a-tiny-layer/">27 A tiny layer</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/the-instruction-word/index.html b/apps/website/public/learn/the-instruction-word/index.html
index 813a8051d8..336203a94a 100644
--- a/apps/website/public/learn/the-instruction-word/index.html
+++ b/apps/website/public/learn/the-instruction-word/index.html
@@ -548,5 +548,6 @@ <h1>The instruction word</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Three values</h3><p>Why three, how balanced ternary writes every number without a sign, and what flipping and cutting trits do.</p><ol><li><a href="/learn/why-three/">01 Why three</a></li><li><a href="/learn/balanced-ternary/">02 Balanced ternary</a></li><li><a href="/learn/flip-and-cut/">03 Flip to negate, cut to round</a></li></ol></li><li><h3>Module 2 · Logic with unknown</h3><p>Kleene's three-valued gates, two gates binary has no twin for, and addition as a pair of tables.</p><ol><li><a href="/learn/kleene-logic/">04 Three-valued logic</a></li><li><a href="/learn/consensus-and-any/">05 Consensus and any</a></li><li><a href="/learn/sum-and-carry/">06 Sum and carry</a></li></ol></li><li><h3>Module 3 · Adding trits</h3><p>The half adder, the full adder and a carry that ripples left, one place at a time.</p><ol><li><a href="/learn/half-adder/">07 The half adder</a></li><li><a href="/learn/full-adder/">08 The full adder</a></li><li><a href="/learn/ripple-carry/">09 Ripple carry</a></li></ol></li><li><h3>Module 4 · Multiplying without a multiplier</h3><p>Copy, drop or flip: a product by one trit, long multiplication, and a MAC that only adds.</p><ol><li><a href="/learn/times-a-trit/">10 Times one trit</a></li><li><a href="/learn/long-multiplication/">11 Long multiplication</a></li><li><a href="/learn/ternary-mac/">12 A ternary MAC</a></li></ol></li><li><h3>Module 5 · Trits in binary memory</h3><p>Two bits per trit, five trits per byte, and the bits a 27-trit word needs.</p><ol><li><a href="/learn/two-bits-per-trit/">13 Two bits per trit</a></li><li><a href="/learn/five-trits-a-byte/">14 Five trits in a byte</a></li><li><a href="/learn/a-word-of-27/">15 A word of 27 trits</a></li></ol></li><li><h3>Module 6 · The TRI-27 instruction word</h3><p>The 32-bit word the Trinity emulator decodes: its fields, its 47 opcodes and its 15-bit immediate.</p><ol><li class="on" aria-current="page"><a href="/learn/the-instruction-word/">16 The instruction word</a></li><li><a href="/learn/the-opcode-map/">17 The opcode map</a></li><li><a href="/learn/the-immediate/">18 The immediate</a></li></ol></li><li><h3>Module 7 · A ternary machine</h3><p>An ALU built from this course's adders, a three-way jump, and a program you can step.</p><ol><li><a href="/learn/a-ternary-alu/">19 A ternary ALU</a></li><li><a href="/learn/three-way-jump/">20 A three-way jump</a></li><li><a href="/learn/run-a-program/">21 Run a program</a></li></ol></li><li><h3>Module 8 · Three answers</h3><p>Compare at the highest differing trit, find a number in thirds, sort with three-way compares.</p><ol><li><a href="/learn/compare-at-the-top/">22 Compare at the top</a></li><li><a href="/learn/questions-in-thirds/">23 Questions in thirds</a></li><li><a href="/learn/sort-by-three/">24 Sort with three-way compares</a></li></ol></li><li><h3>Module 9 · Ternary neurons</h3><p>Weights of -1, 0, +1: one neuron, a detector and a small layer, with no multiplier anywhere.</p><ol><li><a href="/learn/one-ternary-neuron/">25 One ternary neuron</a></li><li><a href="/learn/a-bar-detector/">26 A bar detector</a></li><li><a href="/learn/a-tiny-layer/">27 A tiny layer</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/the-module-line/index.html b/apps/website/public/learn/the-module-line/index.html
index 47812a0a13..51695e095e 100644
--- a/apps/website/public/learn/the-module-line/index.html
+++ b/apps/website/public/learn/the-module-line/index.html
@@ -81,5 +81,6 @@ <h1>The module line</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · A spec file</h3><p>What a .t27 file holds: a module line, prose lines and comments.</p><ol><li><a href="/learn/what-a-spec-is/">01 What a spec is</a></li><li class="on" aria-current="page"><a href="/learn/the-module-line/">02 The module line</a></li><li><a href="/learn/comments-and-prose/">03 Comments and prose lines</a></li></ol></li><li><h3>Module 2 · Values and types</h3><p>Constants, the width of a whole number, and true, false and text.</p><ol><li><a href="/learn/constants/">04 Constants</a></li><li><a href="/learn/whole-numbers/">05 Whole numbers and their widths</a></li><li><a href="/learn/true-false-and-text/">06 True, false and text</a></li></ol></li><li><h3>Module 3 · Arrays, trits and expressions</h3><p>Lists of values, the three values of a trit, and what operators compute.</p><ol><li><a href="/learn/arrays/">07 Arrays</a></li><li><a href="/learn/trits/">08 Trits: minus one, zero, one</a></li><li><a href="/learn/expressions/">09 Expressions</a></li></ol></li><li><h3>Module 4 · Tests and functions</h3><p>A test block, several small tests in one spec, and a function with typed inputs.</p><ol><li><a href="/learn/a-test-block/">10 A test block</a></li><li><a href="/learn/many-tests/">11 Many tests in one spec</a></li><li><a href="/learn/functions/">12 Functions</a></li></ol></li><li><h3>Module 5 · Inside a function</h3><p>Names with let and var, choices with if and switch, and loops.</p><ol><li><a href="/learn/local-names/">13 Local names: let and var</a></li><li><a href="/learn/if-and-else/">14 Choices: if, else and switch</a></li><li><a href="/learn/loops/">15 Loops: while and for</a></li></ol></li><li><h3>Module 6 · Visibility and your own types</h3><p>What pub shares with other modules, and the structs and enums you declare.</p><ol><li><a href="/learn/pub-or-private/">16 pub or private</a></li><li><a href="/learn/structs/">17 Structs</a></li><li><a href="/learn/enums/">18 Enums</a></li></ol></li><li><h3>Module 7 · Modules, rules and gen-ts</h3><p>Reading another module, stating a rule that always holds, and a spec turned into TypeScript.</p><ol><li><a href="/learn/use-other-modules/">19 Using other modules</a></li><li><a href="/learn/invariants/">20 Invariants and benches</a></li><li><a href="/learn/gen-ts/">21 gen-ts: a spec becomes TypeScript</a></li></ol></li><li><h3>Module 8 · The compiler and tri</h3><p>The backends t27c writes, the stages it reads a spec in, and the tri command.</p><ol><li><a href="/learn/seven-backends/">22 Seven backends</a></li><li><a href="/learn/stages-of-t27c/">23 The stages of t27c</a></li><li><a href="/learn/tri-commands/">24 tri: the command line</a></li></ol></li><li><h3>Module 9 · Errors, a program, what next</h3><p>Reading a compiler error, one small complete program, and the course to take next.</p><ol><li><a href="/learn/reading-errors/">25 Reading compiler errors</a></li><li><a href="/learn/a-small-program/">26 A small complete program</a></li><li><a href="/learn/where-next/">27 Where to go next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/the-moving-average/index.html b/apps/website/public/learn/the-moving-average/index.html
index 8454245761..f6344b4860 100644
--- a/apps/website/public/learn/the-moving-average/index.html
+++ b/apps/website/public/learn/the-moving-average/index.html
@@ -1122,5 +1122,6 @@ <h1>The moving average</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Adding numbers</h3><p>Why addition is slow: the carry that walks up the word, the dedicated chain in every slice, prefix networks and carry-save trees.</p><ol><li><a href="/learn/the-carry-chain/">01 The carry chain</a></li><li><a href="/learn/prefix-adders/">02 Looking ahead: prefix adders</a></li><li><a href="/learn/carry-save/">03 Three numbers at once</a></li></ol></li><li><h3>Module 2 · Multiplying</h3><p>A product is a sum of shifted copies; Booth recoding halves them, and the DSP48E1 slice does the rest in one block.</p><ol><li><a href="/learn/shift-and-add/">04 Shift and add</a></li><li><a href="/learn/booths-trick/">05 Booth's trick</a></li><li><a href="/learn/the-dsp48e1-slice/">06 The DSP48E1 slice</a></li></ol></li><li><h3>Module 3 · Fixed point</h3><p>Where the binary point sits, what rounding does to a value and to its average, and what each bit of a quantizer buys.</p><ol><li><a href="/learn/q-formats/">07 Q formats</a></li><li><a href="/learn/rounding-and-saturation/">08 Rounding and saturation</a></li><li><a href="/learn/quantization-noise/">09 Quantization noise</a></li></ol></li><li><h3>Module 4 · Functions in hardware</h3><p>Sine, cosine, angle and length from shifts and adds, and when a table is the better answer.</p><ol><li><a href="/learn/rotating-by-shifts/">10 Rotating by shifts</a></li><li><a href="/learn/angle-and-magnitude/">11 Angle and magnitude</a></li><li><a href="/learn/tables-and-interpolation/">12 Tables and interpolation</a></li></ol></li><li><h3>Module 5 · Signals and sampling</h3><p>What sampling does to a frequency, how an oscillator is built from an adder, and what a DFT bin measures.</p><ol><li><a href="/learn/sampling-and-aliasing/">13 Sampling and aliasing</a></li><li><a href="/learn/the-nco/">14 A numerically controlled oscillator</a></li><li><a href="/learn/bins-of-a-spectrum/">15 Bins of a spectrum</a></li></ol></li><li><h3>Module 6 · FIR filters</h3><p>The moving average, a windowed-sinc design with integer taps, and folding the taps onto DSP slices.</p><ol><li class="on" aria-current="page"><a href="/learn/the-moving-average/">16 The moving average</a></li><li><a href="/learn/designing-taps/">17 Designing taps</a></li><li><a href="/learn/folding-taps-onto-slices/">18 Folding taps onto slices</a></li></ol></li><li><h3>Module 7 · Multirate</h3><p>Lowering the sample rate without folding noise in: decimation, the CIC filter and the polyphase form.</p><ol><li><a href="/learn/lowering-the-rate/">19 Lowering the rate</a></li><li><a href="/learn/the-cic-filter/">20 The CIC filter</a></li><li><a href="/learn/polyphase/">21 Polyphase</a></li></ol></li><li><h3>Module 8 · The FFT</h3><p>N log N instead of N^2: butterflies, the bit-reversed input order and the bits each stage adds.</p><ol><li><a href="/learn/the-butterfly/">22 The butterfly</a></li><li><a href="/learn/twiddles-and-bit-reversal/">23 Twiddles and bit reversal</a></li><li><a href="/learn/growth-through-the-stages/">24 Growth through the stages</a></li></ol></li><li><h3>Module 9 · On the bench</h3><p>From the arithmetic to the board: a correlator from the modem spec, a budget of slices and timing, and the capstone filter.</p><ol><li><a href="/learn/a-correlator-finds-the-frame/">25 A correlator finds the frame</a></li><li><a href="/learn/does-it-fit-and-close-timing/">26 Does it fit and close timing?</a></li><li><a href="/learn/capstone-a-filter-on-the-board/">27 Capstone: a filter on the board</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/the-nco/index.html b/apps/website/public/learn/the-nco/index.html
index f494e077e8..0f77992529 100644
--- a/apps/website/public/learn/the-nco/index.html
+++ b/apps/website/public/learn/the-nco/index.html
@@ -1122,5 +1122,6 @@ <h1>A numerically controlled oscillator</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Adding numbers</h3><p>Why addition is slow: the carry that walks up the word, the dedicated chain in every slice, prefix networks and carry-save trees.</p><ol><li><a href="/learn/the-carry-chain/">01 The carry chain</a></li><li><a href="/learn/prefix-adders/">02 Looking ahead: prefix adders</a></li><li><a href="/learn/carry-save/">03 Three numbers at once</a></li></ol></li><li><h3>Module 2 · Multiplying</h3><p>A product is a sum of shifted copies; Booth recoding halves them, and the DSP48E1 slice does the rest in one block.</p><ol><li><a href="/learn/shift-and-add/">04 Shift and add</a></li><li><a href="/learn/booths-trick/">05 Booth's trick</a></li><li><a href="/learn/the-dsp48e1-slice/">06 The DSP48E1 slice</a></li></ol></li><li><h3>Module 3 · Fixed point</h3><p>Where the binary point sits, what rounding does to a value and to its average, and what each bit of a quantizer buys.</p><ol><li><a href="/learn/q-formats/">07 Q formats</a></li><li><a href="/learn/rounding-and-saturation/">08 Rounding and saturation</a></li><li><a href="/learn/quantization-noise/">09 Quantization noise</a></li></ol></li><li><h3>Module 4 · Functions in hardware</h3><p>Sine, cosine, angle and length from shifts and adds, and when a table is the better answer.</p><ol><li><a href="/learn/rotating-by-shifts/">10 Rotating by shifts</a></li><li><a href="/learn/angle-and-magnitude/">11 Angle and magnitude</a></li><li><a href="/learn/tables-and-interpolation/">12 Tables and interpolation</a></li></ol></li><li><h3>Module 5 · Signals and sampling</h3><p>What sampling does to a frequency, how an oscillator is built from an adder, and what a DFT bin measures.</p><ol><li><a href="/learn/sampling-and-aliasing/">13 Sampling and aliasing</a></li><li class="on" aria-current="page"><a href="/learn/the-nco/">14 A numerically controlled oscillator</a></li><li><a href="/learn/bins-of-a-spectrum/">15 Bins of a spectrum</a></li></ol></li><li><h3>Module 6 · FIR filters</h3><p>The moving average, a windowed-sinc design with integer taps, and folding the taps onto DSP slices.</p><ol><li><a href="/learn/the-moving-average/">16 The moving average</a></li><li><a href="/learn/designing-taps/">17 Designing taps</a></li><li><a href="/learn/folding-taps-onto-slices/">18 Folding taps onto slices</a></li></ol></li><li><h3>Module 7 · Multirate</h3><p>Lowering the sample rate without folding noise in: decimation, the CIC filter and the polyphase form.</p><ol><li><a href="/learn/lowering-the-rate/">19 Lowering the rate</a></li><li><a href="/learn/the-cic-filter/">20 The CIC filter</a></li><li><a href="/learn/polyphase/">21 Polyphase</a></li></ol></li><li><h3>Module 8 · The FFT</h3><p>N log N instead of N^2: butterflies, the bit-reversed input order and the bits each stage adds.</p><ol><li><a href="/learn/the-butterfly/">22 The butterfly</a></li><li><a href="/learn/twiddles-and-bit-reversal/">23 Twiddles and bit reversal</a></li><li><a href="/learn/growth-through-the-stages/">24 Growth through the stages</a></li></ol></li><li><h3>Module 9 · On the bench</h3><p>From the arithmetic to the board: a correlator from the modem spec, a budget of slices and timing, and the capstone filter.</p><ol><li><a href="/learn/a-correlator-finds-the-frame/">25 A correlator finds the frame</a></li><li><a href="/learn/does-it-fit-and-close-timing/">26 Does it fit and close timing?</a></li><li><a href="/learn/capstone-a-filter-on-the-board/">27 Capstone: a filter on the board</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/the-opcode-map/index.html b/apps/website/public/learn/the-opcode-map/index.html
index 3096c82a98..c4193049ea 100644
--- a/apps/website/public/learn/the-opcode-map/index.html
+++ b/apps/website/public/learn/the-opcode-map/index.html
@@ -304,5 +304,6 @@ <h1>The opcode map</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Three values</h3><p>Why three, how balanced ternary writes every number without a sign, and what flipping and cutting trits do.</p><ol><li><a href="/learn/why-three/">01 Why three</a></li><li><a href="/learn/balanced-ternary/">02 Balanced ternary</a></li><li><a href="/learn/flip-and-cut/">03 Flip to negate, cut to round</a></li></ol></li><li><h3>Module 2 · Logic with unknown</h3><p>Kleene's three-valued gates, two gates binary has no twin for, and addition as a pair of tables.</p><ol><li><a href="/learn/kleene-logic/">04 Three-valued logic</a></li><li><a href="/learn/consensus-and-any/">05 Consensus and any</a></li><li><a href="/learn/sum-and-carry/">06 Sum and carry</a></li></ol></li><li><h3>Module 3 · Adding trits</h3><p>The half adder, the full adder and a carry that ripples left, one place at a time.</p><ol><li><a href="/learn/half-adder/">07 The half adder</a></li><li><a href="/learn/full-adder/">08 The full adder</a></li><li><a href="/learn/ripple-carry/">09 Ripple carry</a></li></ol></li><li><h3>Module 4 · Multiplying without a multiplier</h3><p>Copy, drop or flip: a product by one trit, long multiplication, and a MAC that only adds.</p><ol><li><a href="/learn/times-a-trit/">10 Times one trit</a></li><li><a href="/learn/long-multiplication/">11 Long multiplication</a></li><li><a href="/learn/ternary-mac/">12 A ternary MAC</a></li></ol></li><li><h3>Module 5 · Trits in binary memory</h3><p>Two bits per trit, five trits per byte, and the bits a 27-trit word needs.</p><ol><li><a href="/learn/two-bits-per-trit/">13 Two bits per trit</a></li><li><a href="/learn/five-trits-a-byte/">14 Five trits in a byte</a></li><li><a href="/learn/a-word-of-27/">15 A word of 27 trits</a></li></ol></li><li><h3>Module 6 · The TRI-27 instruction word</h3><p>The 32-bit word the Trinity emulator decodes: its fields, its 47 opcodes and its 15-bit immediate.</p><ol><li><a href="/learn/the-instruction-word/">16 The instruction word</a></li><li class="on" aria-current="page"><a href="/learn/the-opcode-map/">17 The opcode map</a></li><li><a href="/learn/the-immediate/">18 The immediate</a></li></ol></li><li><h3>Module 7 · A ternary machine</h3><p>An ALU built from this course's adders, a three-way jump, and a program you can step.</p><ol><li><a href="/learn/a-ternary-alu/">19 A ternary ALU</a></li><li><a href="/learn/three-way-jump/">20 A three-way jump</a></li><li><a href="/learn/run-a-program/">21 Run a program</a></li></ol></li><li><h3>Module 8 · Three answers</h3><p>Compare at the highest differing trit, find a number in thirds, sort with three-way compares.</p><ol><li><a href="/learn/compare-at-the-top/">22 Compare at the top</a></li><li><a href="/learn/questions-in-thirds/">23 Questions in thirds</a></li><li><a href="/learn/sort-by-three/">24 Sort with three-way compares</a></li></ol></li><li><h3>Module 9 · Ternary neurons</h3><p>Weights of -1, 0, +1: one neuron, a detector and a small layer, with no multiplier anywhere.</p><ol><li><a href="/learn/one-ternary-neuron/">25 One ternary neuron</a></li><li><a href="/learn/a-bar-detector/">26 A bar detector</a></li><li><a href="/learn/a-tiny-layer/">27 A tiny layer</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/the-packet-bridge/index.html b/apps/website/public/learn/the-packet-bridge/index.html
index 34fa0c1ed6..410e088b29 100644
--- a/apps/website/public/learn/the-packet-bridge/index.html
+++ b/apps/website/public/learn/the-packet-bridge/index.html
@@ -593,5 +593,6 @@ <h1>The packet bridge</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a bus is</h3><p>Why a bus exists at all: a conversation on wires, framed and addressed, and who is allowed to talk.</p><ol><li><a href="/learn/wires-that-talk/">01 Wires that talk</a></li><li><a href="/learn/framing-a-conversation/">02 Framing a conversation</a></li><li><a href="/learn/master-and-servant/">03 Master and servant</a></li></ol></li><li><h3>Module 2 · UART</h3><p>The two-wire bus with no clock: the frame, the divisor that sets the rate, and the status a driver polls.</p><ol><li><a href="/learn/the-frame/">04 The frame: start, data, stop</a></li><li><a href="/learn/baud-and-the-divisor/">05 Baud and the divisor</a></li><li><a href="/learn/status-and-fifos/">06 Status and FIFOs</a></li></ol></li><li><h3>Module 3 · SPI</h3><p>The clocked conversation: four modes, a prescaler ladder, and a chip select per servant.</p><ol><li><a href="/learn/modes-cpol-cpha/">07 Modes: CPOL and CPHA</a></li><li><a href="/learn/the-prescaler-ladder/">08 The prescaler ladder</a></li><li><a href="/learn/chip-select-and-widths/">09 Chip select and widths</a></li></ol></li><li><h3>Module 4 · APB</h3><p>The register bus: PSEL and PENABLE, strobes and wait states, and how many address bits a peripheral count costs.</p><ol><li><a href="/learn/psel-and-penable/">10 PSEL and PENABLE</a></li><li><a href="/learn/strobes-and-waits/">11 Strobes and waits</a></li><li><a href="/learn/address-decode/">12 Address decode</a></li></ol></li><li><h3>Module 5 · AXI4</h3><p>The five channels: address, data and response in both directions, lite or full, bursts and IDs.</p><ol><li><a href="/learn/five-channels/">13 Five channels</a></li><li><a href="/learn/lite-or-full/">14 Lite or full</a></li><li><a href="/learn/bursts-and-ids/">15 Bursts and IDs</a></li></ol></li><li><h3>Module 6 · Memory</h3><p>What sits on the far side of every bus: memory maps, port kinds, and latency that a wait state must cover.</p><ol><li><a href="/learn/memory-maps/">16 Memory maps</a></li><li><a href="/learn/ports-and-kinds/">17 Ports and kinds</a></li><li><a href="/learn/latency/">18 Latency</a></li></ol></li><li><h3>Module 7 · Bridges</h3><p>Why designs grow more than one bus, and the packet bridge that moves work between them.</p><ol><li><a href="/learn/why-bridges/">19 Why bridges</a></li><li class="on" aria-current="page"><a href="/learn/the-packet-bridge/">20 The packet bridge</a></li><li><a href="/learn/everything-in-one-bench/">21 Everything in one bench</a></li></ol></li><li><h3>Module 8 · Ethernet</h3><p>Frames, the frame check sequence, RGMII timing, and the pre-registered steps of a real bring-up.</p><ol><li><a href="/learn/frames-and-the-fcs/">22 Frames and the FCS</a></li><li><a href="/learn/rgmii-timing/">23 RGMII timing</a></li><li><a href="/learn/bring-up-on-the-bench/">24 Bring-up on the bench</a></li></ol></li><li><h3>Module 9 · The bench</h3><p>The discipline that guards real hardware: who holds the IO, taking and giving the lock, and what runs next.</p><ol><li><a href="/learn/who-holds-the-io/">25 Who holds the IO</a></li><li><a href="/learn/claim-and-release/">26 Claim and release</a></li><li><a href="/learn/capstone-what-runs-next/">27 Capstone: what runs next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/the-prescaler-ladder/index.html b/apps/website/public/learn/the-prescaler-ladder/index.html
index 1d37075bc1..9cde30bb4b 100644
--- a/apps/website/public/learn/the-prescaler-ladder/index.html
+++ b/apps/website/public/learn/the-prescaler-ladder/index.html
@@ -209,5 +209,6 @@ <h1>The prescaler ladder</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a bus is</h3><p>Why a bus exists at all: a conversation on wires, framed and addressed, and who is allowed to talk.</p><ol><li><a href="/learn/wires-that-talk/">01 Wires that talk</a></li><li><a href="/learn/framing-a-conversation/">02 Framing a conversation</a></li><li><a href="/learn/master-and-servant/">03 Master and servant</a></li></ol></li><li><h3>Module 2 · UART</h3><p>The two-wire bus with no clock: the frame, the divisor that sets the rate, and the status a driver polls.</p><ol><li><a href="/learn/the-frame/">04 The frame: start, data, stop</a></li><li><a href="/learn/baud-and-the-divisor/">05 Baud and the divisor</a></li><li><a href="/learn/status-and-fifos/">06 Status and FIFOs</a></li></ol></li><li><h3>Module 3 · SPI</h3><p>The clocked conversation: four modes, a prescaler ladder, and a chip select per servant.</p><ol><li><a href="/learn/modes-cpol-cpha/">07 Modes: CPOL and CPHA</a></li><li class="on" aria-current="page"><a href="/learn/the-prescaler-ladder/">08 The prescaler ladder</a></li><li><a href="/learn/chip-select-and-widths/">09 Chip select and widths</a></li></ol></li><li><h3>Module 4 · APB</h3><p>The register bus: PSEL and PENABLE, strobes and wait states, and how many address bits a peripheral count costs.</p><ol><li><a href="/learn/psel-and-penable/">10 PSEL and PENABLE</a></li><li><a href="/learn/strobes-and-waits/">11 Strobes and waits</a></li><li><a href="/learn/address-decode/">12 Address decode</a></li></ol></li><li><h3>Module 5 · AXI4</h3><p>The five channels: address, data and response in both directions, lite or full, bursts and IDs.</p><ol><li><a href="/learn/five-channels/">13 Five channels</a></li><li><a href="/learn/lite-or-full/">14 Lite or full</a></li><li><a href="/learn/bursts-and-ids/">15 Bursts and IDs</a></li></ol></li><li><h3>Module 6 · Memory</h3><p>What sits on the far side of every bus: memory maps, port kinds, and latency that a wait state must cover.</p><ol><li><a href="/learn/memory-maps/">16 Memory maps</a></li><li><a href="/learn/ports-and-kinds/">17 Ports and kinds</a></li><li><a href="/learn/latency/">18 Latency</a></li></ol></li><li><h3>Module 7 · Bridges</h3><p>Why designs grow more than one bus, and the packet bridge that moves work between them.</p><ol><li><a href="/learn/why-bridges/">19 Why bridges</a></li><li><a href="/learn/the-packet-bridge/">20 The packet bridge</a></li><li><a href="/learn/everything-in-one-bench/">21 Everything in one bench</a></li></ol></li><li><h3>Module 8 · Ethernet</h3><p>Frames, the frame check sequence, RGMII timing, and the pre-registered steps of a real bring-up.</p><ol><li><a href="/learn/frames-and-the-fcs/">22 Frames and the FCS</a></li><li><a href="/learn/rgmii-timing/">23 RGMII timing</a></li><li><a href="/learn/bring-up-on-the-bench/">24 Bring-up on the bench</a></li></ol></li><li><h3>Module 9 · The bench</h3><p>The discipline that guards real hardware: who holds the IO, taking and giving the lock, and what runs next.</p><ol><li><a href="/learn/who-holds-the-io/">25 Who holds the IO</a></li><li><a href="/learn/claim-and-release/">26 Claim and release</a></li><li><a href="/learn/capstone-what-runs-next/">27 Capstone: what runs next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/three-way-agreement/index.html b/apps/website/public/learn/three-way-agreement/index.html
index f980796e11..4530ce5957 100644
--- a/apps/website/public/learn/three-way-agreement/index.html
+++ b/apps/website/public/learn/three-way-agreement/index.html
@@ -107,5 +107,6 @@ <h1>Three-way agreement</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Why verify</h3><p>Designs that compile and are wrong, the model that decides, and the plan written before the code.</p><ol><li><a href="/learn/bugs-that-compile/">01 Bugs that compile</a></li><li><a href="/learn/golden-model/">02 The golden model</a></li><li><a href="/learn/plan-before-code/">03 Plan before code</a></li></ol></li><li><h3>Module 2 · Testbenches</h3><p>Stimulus, checks and a verdict, written as one spec beside the design it judges.</p><ol><li><a href="/learn/anatomy-of-a-tb/">04 Anatomy of a testbench</a></li><li><a href="/learn/self-checking/">05 Self-checking, not staring</a></li><li><a href="/learn/directed-vs-random/">06 Directed versus random</a></li></ol></li><li><h3>Module 3 · Waveforms</h3><p>A trace of every signal, read the way a hardware engineer reads it, and two runs compared.</p><ol><li><a href="/learn/vcd-format/">07 The VCD format</a></li><li><a href="/learn/debug-with-waves/">08 Debug with waves</a></li><li><a href="/learn/compare-traces/">09 Compare two traces</a></li></ol></li><li><h3>Module 4 · Conformance vectors</h3><p>Cases with the answer written beside them, kept where the compiler can reach them.</p><ol><li><a href="/learn/what-a-vector-is/">10 What a vector is</a></li><li><a href="/learn/golden-ruler-vectors/">11 Golden ruler vectors</a></li><li><a href="/learn/end-to-end-scenarios/">12 End-to-end scenarios</a></li></ol></li><li><h3>Module 5 · Cosimulation</h3><p>Spec, simulator and board agreeing on the bench Artix-7 XC7A200T, and what to do when they do not.</p><ol><li class="on" aria-current="page"><a href="/learn/three-way-agreement/">13 Three-way agreement</a></li><li><a href="/learn/simulator-internals/">14 Inside the simulator</a></li><li><a href="/learn/when-they-disagree/">15 When they disagree</a></li></ol></li><li><h3>Module 6 · Coverage</h3><p>What the tests touched: lines, toggles, states, and what that number hides.</p><ol><li><a href="/learn/line-and-toggle/">16 Line and toggle</a></li><li><a href="/learn/fsm-coverage/">17 FSM coverage</a></li><li><a href="/learn/coverage-lies/">18 Coverage lies</a></li></ol></li><li><h3>Module 7 · Formal</h3><p>Assertions that hold every cycle, bounded search for a counterexample, and why a proof needs induction.</p><ol><li><a href="/learn/assertions/">19 Assertions</a></li><li><a href="/learn/bounded-model-checking/">20 Bounded model checking</a></li><li><a href="/learn/induction/">21 Induction</a></li></ol></li><li><h3>Module 8 · Mutation</h3><p>Break the design on purpose and count what the tests catch.</p><ol><li><a href="/learn/kill-the-mutant/">22 Kill the mutant</a></li><li><a href="/learn/surviving-mutants/">23 Surviving mutants</a></li><li><a href="/learn/test-strength/">24 Test strength</a></li></ol></li><li><h3>Module 9 · Sign-off</h3><p>One command, every receipt, a clean verdict you can show.</p><ol><li><a href="/learn/build-verify/">25 Build and verify</a></li><li><a href="/learn/board-selftest/">26 The board self-tests</a></li><li><a href="/learn/verify-capstone/">27 Capstone: verify a module whole</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/three-way-jump/index.html b/apps/website/public/learn/three-way-jump/index.html
index 39b44711e6..9b1ed81f71 100644
--- a/apps/website/public/learn/three-way-jump/index.html
+++ b/apps/website/public/learn/three-way-jump/index.html
@@ -769,5 +769,6 @@ <h1>A three-way jump</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Three values</h3><p>Why three, how balanced ternary writes every number without a sign, and what flipping and cutting trits do.</p><ol><li><a href="/learn/why-three/">01 Why three</a></li><li><a href="/learn/balanced-ternary/">02 Balanced ternary</a></li><li><a href="/learn/flip-and-cut/">03 Flip to negate, cut to round</a></li></ol></li><li><h3>Module 2 · Logic with unknown</h3><p>Kleene's three-valued gates, two gates binary has no twin for, and addition as a pair of tables.</p><ol><li><a href="/learn/kleene-logic/">04 Three-valued logic</a></li><li><a href="/learn/consensus-and-any/">05 Consensus and any</a></li><li><a href="/learn/sum-and-carry/">06 Sum and carry</a></li></ol></li><li><h3>Module 3 · Adding trits</h3><p>The half adder, the full adder and a carry that ripples left, one place at a time.</p><ol><li><a href="/learn/half-adder/">07 The half adder</a></li><li><a href="/learn/full-adder/">08 The full adder</a></li><li><a href="/learn/ripple-carry/">09 Ripple carry</a></li></ol></li><li><h3>Module 4 · Multiplying without a multiplier</h3><p>Copy, drop or flip: a product by one trit, long multiplication, and a MAC that only adds.</p><ol><li><a href="/learn/times-a-trit/">10 Times one trit</a></li><li><a href="/learn/long-multiplication/">11 Long multiplication</a></li><li><a href="/learn/ternary-mac/">12 A ternary MAC</a></li></ol></li><li><h3>Module 5 · Trits in binary memory</h3><p>Two bits per trit, five trits per byte, and the bits a 27-trit word needs.</p><ol><li><a href="/learn/two-bits-per-trit/">13 Two bits per trit</a></li><li><a href="/learn/five-trits-a-byte/">14 Five trits in a byte</a></li><li><a href="/learn/a-word-of-27/">15 A word of 27 trits</a></li></ol></li><li><h3>Module 6 · The TRI-27 instruction word</h3><p>The 32-bit word the Trinity emulator decodes: its fields, its 47 opcodes and its 15-bit immediate.</p><ol><li><a href="/learn/the-instruction-word/">16 The instruction word</a></li><li><a href="/learn/the-opcode-map/">17 The opcode map</a></li><li><a href="/learn/the-immediate/">18 The immediate</a></li></ol></li><li><h3>Module 7 · A ternary machine</h3><p>An ALU built from this course's adders, a three-way jump, and a program you can step.</p><ol><li><a href="/learn/a-ternary-alu/">19 A ternary ALU</a></li><li class="on" aria-current="page"><a href="/learn/three-way-jump/">20 A three-way jump</a></li><li><a href="/learn/run-a-program/">21 Run a program</a></li></ol></li><li><h3>Module 8 · Three answers</h3><p>Compare at the highest differing trit, find a number in thirds, sort with three-way compares.</p><ol><li><a href="/learn/compare-at-the-top/">22 Compare at the top</a></li><li><a href="/learn/questions-in-thirds/">23 Questions in thirds</a></li><li><a href="/learn/sort-by-three/">24 Sort with three-way compares</a></li></ol></li><li><h3>Module 9 · Ternary neurons</h3><p>Weights of -1, 0, +1: one neuron, a detector and a small layer, with no multiplier anywhere.</p><ol><li><a href="/learn/one-ternary-neuron/">25 One ternary neuron</a></li><li><a href="/learn/a-bar-detector/">26 A bar detector</a></li><li><a href="/learn/a-tiny-layer/">27 A tiny layer</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/times-a-trit/index.html b/apps/website/public/learn/times-a-trit/index.html
index 71fc8bad2e..327e80a2df 100644
--- a/apps/website/public/learn/times-a-trit/index.html
+++ b/apps/website/public/learn/times-a-trit/index.html
@@ -769,5 +769,6 @@ <h1>Times one trit</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Three values</h3><p>Why three, how balanced ternary writes every number without a sign, and what flipping and cutting trits do.</p><ol><li><a href="/learn/why-three/">01 Why three</a></li><li><a href="/learn/balanced-ternary/">02 Balanced ternary</a></li><li><a href="/learn/flip-and-cut/">03 Flip to negate, cut to round</a></li></ol></li><li><h3>Module 2 · Logic with unknown</h3><p>Kleene's three-valued gates, two gates binary has no twin for, and addition as a pair of tables.</p><ol><li><a href="/learn/kleene-logic/">04 Three-valued logic</a></li><li><a href="/learn/consensus-and-any/">05 Consensus and any</a></li><li><a href="/learn/sum-and-carry/">06 Sum and carry</a></li></ol></li><li><h3>Module 3 · Adding trits</h3><p>The half adder, the full adder and a carry that ripples left, one place at a time.</p><ol><li><a href="/learn/half-adder/">07 The half adder</a></li><li><a href="/learn/full-adder/">08 The full adder</a></li><li><a href="/learn/ripple-carry/">09 Ripple carry</a></li></ol></li><li><h3>Module 4 · Multiplying without a multiplier</h3><p>Copy, drop or flip: a product by one trit, long multiplication, and a MAC that only adds.</p><ol><li class="on" aria-current="page"><a href="/learn/times-a-trit/">10 Times one trit</a></li><li><a href="/learn/long-multiplication/">11 Long multiplication</a></li><li><a href="/learn/ternary-mac/">12 A ternary MAC</a></li></ol></li><li><h3>Module 5 · Trits in binary memory</h3><p>Two bits per trit, five trits per byte, and the bits a 27-trit word needs.</p><ol><li><a href="/learn/two-bits-per-trit/">13 Two bits per trit</a></li><li><a href="/learn/five-trits-a-byte/">14 Five trits in a byte</a></li><li><a href="/learn/a-word-of-27/">15 A word of 27 trits</a></li></ol></li><li><h3>Module 6 · The TRI-27 instruction word</h3><p>The 32-bit word the Trinity emulator decodes: its fields, its 47 opcodes and its 15-bit immediate.</p><ol><li><a href="/learn/the-instruction-word/">16 The instruction word</a></li><li><a href="/learn/the-opcode-map/">17 The opcode map</a></li><li><a href="/learn/the-immediate/">18 The immediate</a></li></ol></li><li><h3>Module 7 · A ternary machine</h3><p>An ALU built from this course's adders, a three-way jump, and a program you can step.</p><ol><li><a href="/learn/a-ternary-alu/">19 A ternary ALU</a></li><li><a href="/learn/three-way-jump/">20 A three-way jump</a></li><li><a href="/learn/run-a-program/">21 Run a program</a></li></ol></li><li><h3>Module 8 · Three answers</h3><p>Compare at the highest differing trit, find a number in thirds, sort with three-way compares.</p><ol><li><a href="/learn/compare-at-the-top/">22 Compare at the top</a></li><li><a href="/learn/questions-in-thirds/">23 Questions in thirds</a></li><li><a href="/learn/sort-by-three/">24 Sort with three-way compares</a></li></ol></li><li><h3>Module 9 · Ternary neurons</h3><p>Weights of -1, 0, +1: one neuron, a detector and a small layer, with no multiplier anywhere.</p><ol><li><a href="/learn/one-ternary-neuron/">25 One ternary neuron</a></li><li><a href="/learn/a-bar-detector/">26 A bar detector</a></li><li><a href="/learn/a-tiny-layer/">27 A tiny layer</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/timing-slack/index.html b/apps/website/public/learn/timing-slack/index.html
index f35966a0a1..a9665f2484 100644
--- a/apps/website/public/learn/timing-slack/index.html
+++ b/apps/website/public/learn/timing-slack/index.html
@@ -294,5 +294,6 @@ <h1>Timing and slack</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The chip</h3><p>What an FPGA is, which chip we use, and how its pins meet the board.</p><ol><li><a href="/learn/fpga-is-a-grid/">01 An FPGA is a grid you configure</a></li><li><a href="/learn/which-chip/">02 Which chip is this?</a></li><li><a href="/learn/pins-and-package/">03 Pins, balls and the package</a></li></ol></li><li><h3>Module 2 · Numbers in hardware</h3><p>Bits, trits and number formats, and what arithmetic costs in logic.</p><ol><li><a href="/learn/bits-and-trits/">04 Bits, trits and GF16</a></li><li><a href="/learn/what-arithmetic-costs/">05 What arithmetic costs</a></li><li><a href="/learn/how-much-fits/">06 How much fits on the chip</a></li></ol></li><li><h3>Module 3 · Your t27 program</h3><p>Write a spec, test it, and see why compiling is not the same as being right.</p><ol><li><a href="/learn/hello-t27/">07 Hello, t27</a></li><li><a href="/learn/tests-are-the-spec/">08 Tests are part of the spec</a></li><li><a href="/learn/compiles-is-not-correct/">09 Compiles is not the same as correct</a></li></ol></li><li><h3>Module 4 · Inside t27c</h3><p>How the compiler reads a spec and what it writes, including the native t27b.</p><ol><li><a href="/learn/compiler-stages/">10 The stages of t27c</a></li><li><a href="/learn/t27c-on-a-real-machine/">11 t27c on a real machine</a></li><li><a href="/learn/seven-backends-and-t27b/">12 Seven backends, and t27b</a></li></ol></li><li><h3>Module 5 · From spec to hardware</h3><p>The Verilog t27c writes, the cells it becomes, and what one LUT does.</p><ol><li><a href="/learn/spec-to-verilog/">13 From a spec to Verilog</a></li><li><a href="/learn/count-the-cells/">14 Count the cells</a></li><li><a href="/learn/one-lut-up-close/">15 One LUT, up close</a></li></ol></li><li><h3>Module 6 · Reading synthesis</h3><p>What yosys reports about your design, and which warnings matter.</p><ol><li><a href="/learn/synthesis-warnings/">16 Synthesis warnings that matter</a></li><li><a href="/learn/state-machines/">17 State machines</a></li><li><a href="/learn/where-the-luts-go/">18 Where the LUTs go</a></li></ol></li><li><h3>Module 7 · Place, route, timing</h3><p>Where the cells land on the die, and whether the clock is met.</p><ol><li><a href="/learn/placement/">19 Placement: where the cells land</a></li><li><a href="/learn/logic-depth/">20 Logic depth</a></li><li class="on" aria-current="page"><a href="/learn/timing-slack/">21 Timing and slack</a></li></ol></li><li><h3>Module 8 · The bitstream</h3><p>How a routed design becomes the bits the chip loads, with no Vivado.</p><ol><li><a href="/learn/fasm-to-frames/">22 From FASM to frames</a></li><li><a href="/learn/a-build-receipt/">23 A build, itemised</a></li><li><a href="/learn/read-a-bitstream/">24 Read a bitstream</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>Ask the chip who it is, load the bits, and check them.</p><ol><li><a href="/learn/jtag-asks-the-chip/">25 JTAG asks the chip</a></li><li><a href="/learn/same-bits-on-the-board/">26 The same bits, on the board</a></li><li><a href="/learn/capstone/">27 Capstone: out-guess yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/tri-commands/index.html b/apps/website/public/learn/tri-commands/index.html
index b59ebd5d4b..87a2fe163a 100644
--- a/apps/website/public/learn/tri-commands/index.html
+++ b/apps/website/public/learn/tri-commands/index.html
@@ -85,5 +85,6 @@ <h1>tri: the command line</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · A spec file</h3><p>What a .t27 file holds: a module line, prose lines and comments.</p><ol><li><a href="/learn/what-a-spec-is/">01 What a spec is</a></li><li><a href="/learn/the-module-line/">02 The module line</a></li><li><a href="/learn/comments-and-prose/">03 Comments and prose lines</a></li></ol></li><li><h3>Module 2 · Values and types</h3><p>Constants, the width of a whole number, and true, false and text.</p><ol><li><a href="/learn/constants/">04 Constants</a></li><li><a href="/learn/whole-numbers/">05 Whole numbers and their widths</a></li><li><a href="/learn/true-false-and-text/">06 True, false and text</a></li></ol></li><li><h3>Module 3 · Arrays, trits and expressions</h3><p>Lists of values, the three values of a trit, and what operators compute.</p><ol><li><a href="/learn/arrays/">07 Arrays</a></li><li><a href="/learn/trits/">08 Trits: minus one, zero, one</a></li><li><a href="/learn/expressions/">09 Expressions</a></li></ol></li><li><h3>Module 4 · Tests and functions</h3><p>A test block, several small tests in one spec, and a function with typed inputs.</p><ol><li><a href="/learn/a-test-block/">10 A test block</a></li><li><a href="/learn/many-tests/">11 Many tests in one spec</a></li><li><a href="/learn/functions/">12 Functions</a></li></ol></li><li><h3>Module 5 · Inside a function</h3><p>Names with let and var, choices with if and switch, and loops.</p><ol><li><a href="/learn/local-names/">13 Local names: let and var</a></li><li><a href="/learn/if-and-else/">14 Choices: if, else and switch</a></li><li><a href="/learn/loops/">15 Loops: while and for</a></li></ol></li><li><h3>Module 6 · Visibility and your own types</h3><p>What pub shares with other modules, and the structs and enums you declare.</p><ol><li><a href="/learn/pub-or-private/">16 pub or private</a></li><li><a href="/learn/structs/">17 Structs</a></li><li><a href="/learn/enums/">18 Enums</a></li></ol></li><li><h3>Module 7 · Modules, rules and gen-ts</h3><p>Reading another module, stating a rule that always holds, and a spec turned into TypeScript.</p><ol><li><a href="/learn/use-other-modules/">19 Using other modules</a></li><li><a href="/learn/invariants/">20 Invariants and benches</a></li><li><a href="/learn/gen-ts/">21 gen-ts: a spec becomes TypeScript</a></li></ol></li><li><h3>Module 8 · The compiler and tri</h3><p>The backends t27c writes, the stages it reads a spec in, and the tri command.</p><ol><li><a href="/learn/seven-backends/">22 Seven backends</a></li><li><a href="/learn/stages-of-t27c/">23 The stages of t27c</a></li><li class="on" aria-current="page"><a href="/learn/tri-commands/">24 tri: the command line</a></li></ol></li><li><h3>Module 9 · Errors, a program, what next</h3><p>Reading a compiler error, one small complete program, and the course to take next.</p><ol><li><a href="/learn/reading-errors/">25 Reading compiler errors</a></li><li><a href="/learn/a-small-program/">26 A small complete program</a></li><li><a href="/learn/where-next/">27 Where to go next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/trits/index.html b/apps/website/public/learn/trits/index.html
index 5305319acc..e3703fa964 100644
--- a/apps/website/public/learn/trits/index.html
+++ b/apps/website/public/learn/trits/index.html
@@ -89,5 +89,6 @@ <h1>Trits: minus one, zero, one</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · A spec file</h3><p>What a .t27 file holds: a module line, prose lines and comments.</p><ol><li><a href="/learn/what-a-spec-is/">01 What a spec is</a></li><li><a href="/learn/the-module-line/">02 The module line</a></li><li><a href="/learn/comments-and-prose/">03 Comments and prose lines</a></li></ol></li><li><h3>Module 2 · Values and types</h3><p>Constants, the width of a whole number, and true, false and text.</p><ol><li><a href="/learn/constants/">04 Constants</a></li><li><a href="/learn/whole-numbers/">05 Whole numbers and their widths</a></li><li><a href="/learn/true-false-and-text/">06 True, false and text</a></li></ol></li><li><h3>Module 3 · Arrays, trits and expressions</h3><p>Lists of values, the three values of a trit, and what operators compute.</p><ol><li><a href="/learn/arrays/">07 Arrays</a></li><li class="on" aria-current="page"><a href="/learn/trits/">08 Trits: minus one, zero, one</a></li><li><a href="/learn/expressions/">09 Expressions</a></li></ol></li><li><h3>Module 4 · Tests and functions</h3><p>A test block, several small tests in one spec, and a function with typed inputs.</p><ol><li><a href="/learn/a-test-block/">10 A test block</a></li><li><a href="/learn/many-tests/">11 Many tests in one spec</a></li><li><a href="/learn/functions/">12 Functions</a></li></ol></li><li><h3>Module 5 · Inside a function</h3><p>Names with let and var, choices with if and switch, and loops.</p><ol><li><a href="/learn/local-names/">13 Local names: let and var</a></li><li><a href="/learn/if-and-else/">14 Choices: if, else and switch</a></li><li><a href="/learn/loops/">15 Loops: while and for</a></li></ol></li><li><h3>Module 6 · Visibility and your own types</h3><p>What pub shares with other modules, and the structs and enums you declare.</p><ol><li><a href="/learn/pub-or-private/">16 pub or private</a></li><li><a href="/learn/structs/">17 Structs</a></li><li><a href="/learn/enums/">18 Enums</a></li></ol></li><li><h3>Module 7 · Modules, rules and gen-ts</h3><p>Reading another module, stating a rule that always holds, and a spec turned into TypeScript.</p><ol><li><a href="/learn/use-other-modules/">19 Using other modules</a></li><li><a href="/learn/invariants/">20 Invariants and benches</a></li><li><a href="/learn/gen-ts/">21 gen-ts: a spec becomes TypeScript</a></li></ol></li><li><h3>Module 8 · The compiler and tri</h3><p>The backends t27c writes, the stages it reads a spec in, and the tri command.</p><ol><li><a href="/learn/seven-backends/">22 Seven backends</a></li><li><a href="/learn/stages-of-t27c/">23 The stages of t27c</a></li><li><a href="/learn/tri-commands/">24 tri: the command line</a></li></ol></li><li><h3>Module 9 · Errors, a program, what next</h3><p>Reading a compiler error, one small complete program, and the course to take next.</p><ol><li><a href="/learn/reading-errors/">25 Reading compiler errors</a></li><li><a href="/learn/a-small-program/">26 A small complete program</a></li><li><a href="/learn/where-next/">27 Where to go next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/true-false-and-text/index.html b/apps/website/public/learn/true-false-and-text/index.html
index acd8cf68ce..9982645075 100644
--- a/apps/website/public/learn/true-false-and-text/index.html
+++ b/apps/website/public/learn/true-false-and-text/index.html
@@ -86,5 +86,6 @@ <h1>True, false and text</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · A spec file</h3><p>What a .t27 file holds: a module line, prose lines and comments.</p><ol><li><a href="/learn/what-a-spec-is/">01 What a spec is</a></li><li><a href="/learn/the-module-line/">02 The module line</a></li><li><a href="/learn/comments-and-prose/">03 Comments and prose lines</a></li></ol></li><li><h3>Module 2 · Values and types</h3><p>Constants, the width of a whole number, and true, false and text.</p><ol><li><a href="/learn/constants/">04 Constants</a></li><li><a href="/learn/whole-numbers/">05 Whole numbers and their widths</a></li><li class="on" aria-current="page"><a href="/learn/true-false-and-text/">06 True, false and text</a></li></ol></li><li><h3>Module 3 · Arrays, trits and expressions</h3><p>Lists of values, the three values of a trit, and what operators compute.</p><ol><li><a href="/learn/arrays/">07 Arrays</a></li><li><a href="/learn/trits/">08 Trits: minus one, zero, one</a></li><li><a href="/learn/expressions/">09 Expressions</a></li></ol></li><li><h3>Module 4 · Tests and functions</h3><p>A test block, several small tests in one spec, and a function with typed inputs.</p><ol><li><a href="/learn/a-test-block/">10 A test block</a></li><li><a href="/learn/many-tests/">11 Many tests in one spec</a></li><li><a href="/learn/functions/">12 Functions</a></li></ol></li><li><h3>Module 5 · Inside a function</h3><p>Names with let and var, choices with if and switch, and loops.</p><ol><li><a href="/learn/local-names/">13 Local names: let and var</a></li><li><a href="/learn/if-and-else/">14 Choices: if, else and switch</a></li><li><a href="/learn/loops/">15 Loops: while and for</a></li></ol></li><li><h3>Module 6 · Visibility and your own types</h3><p>What pub shares with other modules, and the structs and enums you declare.</p><ol><li><a href="/learn/pub-or-private/">16 pub or private</a></li><li><a href="/learn/structs/">17 Structs</a></li><li><a href="/learn/enums/">18 Enums</a></li></ol></li><li><h3>Module 7 · Modules, rules and gen-ts</h3><p>Reading another module, stating a rule that always holds, and a spec turned into TypeScript.</p><ol><li><a href="/learn/use-other-modules/">19 Using other modules</a></li><li><a href="/learn/invariants/">20 Invariants and benches</a></li><li><a href="/learn/gen-ts/">21 gen-ts: a spec becomes TypeScript</a></li></ol></li><li><h3>Module 8 · The compiler and tri</h3><p>The backends t27c writes, the stages it reads a spec in, and the tri command.</p><ol><li><a href="/learn/seven-backends/">22 Seven backends</a></li><li><a href="/learn/stages-of-t27c/">23 The stages of t27c</a></li><li><a href="/learn/tri-commands/">24 tri: the command line</a></li></ol></li><li><h3>Module 9 · Errors, a program, what next</h3><p>Reading a compiler error, one small complete program, and the course to take next.</p><ol><li><a href="/learn/reading-errors/">25 Reading compiler errors</a></li><li><a href="/learn/a-small-program/">26 A small complete program</a></li><li><a href="/learn/where-next/">27 Where to go next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/twiddles-and-bit-reversal/index.html b/apps/website/public/learn/twiddles-and-bit-reversal/index.html
index fbdde97c20..e24416def0 100644
--- a/apps/website/public/learn/twiddles-and-bit-reversal/index.html
+++ b/apps/website/public/learn/twiddles-and-bit-reversal/index.html
@@ -1122,5 +1122,6 @@ <h1>Twiddles and bit reversal</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Adding numbers</h3><p>Why addition is slow: the carry that walks up the word, the dedicated chain in every slice, prefix networks and carry-save trees.</p><ol><li><a href="/learn/the-carry-chain/">01 The carry chain</a></li><li><a href="/learn/prefix-adders/">02 Looking ahead: prefix adders</a></li><li><a href="/learn/carry-save/">03 Three numbers at once</a></li></ol></li><li><h3>Module 2 · Multiplying</h3><p>A product is a sum of shifted copies; Booth recoding halves them, and the DSP48E1 slice does the rest in one block.</p><ol><li><a href="/learn/shift-and-add/">04 Shift and add</a></li><li><a href="/learn/booths-trick/">05 Booth's trick</a></li><li><a href="/learn/the-dsp48e1-slice/">06 The DSP48E1 slice</a></li></ol></li><li><h3>Module 3 · Fixed point</h3><p>Where the binary point sits, what rounding does to a value and to its average, and what each bit of a quantizer buys.</p><ol><li><a href="/learn/q-formats/">07 Q formats</a></li><li><a href="/learn/rounding-and-saturation/">08 Rounding and saturation</a></li><li><a href="/learn/quantization-noise/">09 Quantization noise</a></li></ol></li><li><h3>Module 4 · Functions in hardware</h3><p>Sine, cosine, angle and length from shifts and adds, and when a table is the better answer.</p><ol><li><a href="/learn/rotating-by-shifts/">10 Rotating by shifts</a></li><li><a href="/learn/angle-and-magnitude/">11 Angle and magnitude</a></li><li><a href="/learn/tables-and-interpolation/">12 Tables and interpolation</a></li></ol></li><li><h3>Module 5 · Signals and sampling</h3><p>What sampling does to a frequency, how an oscillator is built from an adder, and what a DFT bin measures.</p><ol><li><a href="/learn/sampling-and-aliasing/">13 Sampling and aliasing</a></li><li><a href="/learn/the-nco/">14 A numerically controlled oscillator</a></li><li><a href="/learn/bins-of-a-spectrum/">15 Bins of a spectrum</a></li></ol></li><li><h3>Module 6 · FIR filters</h3><p>The moving average, a windowed-sinc design with integer taps, and folding the taps onto DSP slices.</p><ol><li><a href="/learn/the-moving-average/">16 The moving average</a></li><li><a href="/learn/designing-taps/">17 Designing taps</a></li><li><a href="/learn/folding-taps-onto-slices/">18 Folding taps onto slices</a></li></ol></li><li><h3>Module 7 · Multirate</h3><p>Lowering the sample rate without folding noise in: decimation, the CIC filter and the polyphase form.</p><ol><li><a href="/learn/lowering-the-rate/">19 Lowering the rate</a></li><li><a href="/learn/the-cic-filter/">20 The CIC filter</a></li><li><a href="/learn/polyphase/">21 Polyphase</a></li></ol></li><li><h3>Module 8 · The FFT</h3><p>N log N instead of N^2: butterflies, the bit-reversed input order and the bits each stage adds.</p><ol><li><a href="/learn/the-butterfly/">22 The butterfly</a></li><li class="on" aria-current="page"><a href="/learn/twiddles-and-bit-reversal/">23 Twiddles and bit reversal</a></li><li><a href="/learn/growth-through-the-stages/">24 Growth through the stages</a></li></ol></li><li><h3>Module 9 · On the bench</h3><p>From the arithmetic to the board: a correlator from the modem spec, a budget of slices and timing, and the capstone filter.</p><ol><li><a href="/learn/a-correlator-finds-the-frame/">25 A correlator finds the frame</a></li><li><a href="/learn/does-it-fit-and-close-timing/">26 Does it fit and close timing?</a></li><li><a href="/learn/capstone-a-filter-on-the-board/">27 Capstone: a filter on the board</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/two-bits-per-trit/index.html b/apps/website/public/learn/two-bits-per-trit/index.html
index 35e0371a14..b3bb367231 100644
--- a/apps/website/public/learn/two-bits-per-trit/index.html
+++ b/apps/website/public/learn/two-bits-per-trit/index.html
@@ -769,5 +769,6 @@ <h1>Two bits per trit</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Three values</h3><p>Why three, how balanced ternary writes every number without a sign, and what flipping and cutting trits do.</p><ol><li><a href="/learn/why-three/">01 Why three</a></li><li><a href="/learn/balanced-ternary/">02 Balanced ternary</a></li><li><a href="/learn/flip-and-cut/">03 Flip to negate, cut to round</a></li></ol></li><li><h3>Module 2 · Logic with unknown</h3><p>Kleene's three-valued gates, two gates binary has no twin for, and addition as a pair of tables.</p><ol><li><a href="/learn/kleene-logic/">04 Three-valued logic</a></li><li><a href="/learn/consensus-and-any/">05 Consensus and any</a></li><li><a href="/learn/sum-and-carry/">06 Sum and carry</a></li></ol></li><li><h3>Module 3 · Adding trits</h3><p>The half adder, the full adder and a carry that ripples left, one place at a time.</p><ol><li><a href="/learn/half-adder/">07 The half adder</a></li><li><a href="/learn/full-adder/">08 The full adder</a></li><li><a href="/learn/ripple-carry/">09 Ripple carry</a></li></ol></li><li><h3>Module 4 · Multiplying without a multiplier</h3><p>Copy, drop or flip: a product by one trit, long multiplication, and a MAC that only adds.</p><ol><li><a href="/learn/times-a-trit/">10 Times one trit</a></li><li><a href="/learn/long-multiplication/">11 Long multiplication</a></li><li><a href="/learn/ternary-mac/">12 A ternary MAC</a></li></ol></li><li><h3>Module 5 · Trits in binary memory</h3><p>Two bits per trit, five trits per byte, and the bits a 27-trit word needs.</p><ol><li class="on" aria-current="page"><a href="/learn/two-bits-per-trit/">13 Two bits per trit</a></li><li><a href="/learn/five-trits-a-byte/">14 Five trits in a byte</a></li><li><a href="/learn/a-word-of-27/">15 A word of 27 trits</a></li></ol></li><li><h3>Module 6 · The TRI-27 instruction word</h3><p>The 32-bit word the Trinity emulator decodes: its fields, its 47 opcodes and its 15-bit immediate.</p><ol><li><a href="/learn/the-instruction-word/">16 The instruction word</a></li><li><a href="/learn/the-opcode-map/">17 The opcode map</a></li><li><a href="/learn/the-immediate/">18 The immediate</a></li></ol></li><li><h3>Module 7 · A ternary machine</h3><p>An ALU built from this course's adders, a three-way jump, and a program you can step.</p><ol><li><a href="/learn/a-ternary-alu/">19 A ternary ALU</a></li><li><a href="/learn/three-way-jump/">20 A three-way jump</a></li><li><a href="/learn/run-a-program/">21 Run a program</a></li></ol></li><li><h3>Module 8 · Three answers</h3><p>Compare at the highest differing trit, find a number in thirds, sort with three-way compares.</p><ol><li><a href="/learn/compare-at-the-top/">22 Compare at the top</a></li><li><a href="/learn/questions-in-thirds/">23 Questions in thirds</a></li><li><a href="/learn/sort-by-three/">24 Sort with three-way compares</a></li></ol></li><li><h3>Module 9 · Ternary neurons</h3><p>Weights of -1, 0, +1: one neuron, a detector and a small layer, with no multiplier anywhere.</p><ol><li><a href="/learn/one-ternary-neuron/">25 One ternary neuron</a></li><li><a href="/learn/a-bar-detector/">26 A bar detector</a></li><li><a href="/learn/a-tiny-layer/">27 A tiny layer</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/two-flop-sync/index.html b/apps/website/public/learn/two-flop-sync/index.html
index 6f0eda76dd..f06bb36621 100644
--- a/apps/website/public/learn/two-flop-sync/index.html
+++ b/apps/website/public/learn/two-flop-sync/index.html
@@ -300,5 +300,6 @@ <h1>The two-flop synchronizer</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a clock is</h3><p>One edge, one world: what shares a clock edge shares a world, the period and the jitter of a real edge, and where the clock enters a board.</p><ol><li><a href="/learn/one-edge-one-world/">01 One edge, one world</a></li><li><a href="/learn/period-and-jitter/">02 A period, an edge, and jitter</a></li><li><a href="/learn/where-the-clock-enters/">03 Where the clock enters</a></li></ol></li><li><h3>Module 2 · Clock trees</h3><p>Skew and insertion delay, the global buffer network, and the trap of gating a clock with logic.</p><ol><li><a href="/learn/skew-and-insertion/">04 Skew and insertion delay</a></li><li><a href="/learn/global-buffers/">05 The global buffer</a></li><li><a href="/learn/gated-clock-trap/">06 The gated-clock trap</a></li></ol></li><li><h3>Module 3 · PLL and MMCM</h3><p>Multiply and divide one clock into another, move its phase in steps of the VCO, and which clocks the analyzer treats as related.</p><ol><li><a href="/learn/multiply-and-divide/">07 Multiply and divide</a></li><li><a href="/learn/phase-and-lock/">08 Phase and lock</a></li><li><a href="/learn/related-or-not/">09 Related or not</a></li></ol></li><li><h3>Module 4 · Resets</h3><p>Assert asynchronously, release synchronously: the three reset kinds, the release pipe, and the tree a reset grows.</p><ol><li><a href="/learn/sync-or-async/">10 Sync or async</a></li><li><a href="/learn/reset-release/">11 The release pipe</a></li><li><a href="/learn/reset-fanout/">12 Reset fanout</a></li></ol></li><li><h3>Module 5 · Metastability</h3><p>The setup-hold window, the mean time between failures in integer arithmetic, and the two flops that fix it.</p><ol><li><a href="/learn/setup-hold-window/">13 The setup-hold window</a></li><li><a href="/learn/mtbf/">14 Mean time between failures</a></li><li class="on" aria-current="page"><a href="/learn/two-flop-sync/">15 The two-flop synchronizer</a></li></ol></li><li><h3>Module 6 · Crossing many bits</h3><p>Why a binary bus tears, why Gray code does not, and the handshake that moves a pulse between worlds.</p><ol><li><a href="/learn/why-binary-fails/">16 Why binary fails</a></li><li><a href="/learn/gray-code/">17 Gray code</a></li><li><a href="/learn/handshakes-and-pulses/">18 Handshakes and pulses</a></li></ol></li><li><h3>Module 7 · The asynchronous FIFO</h3><p>Pointers, flags and depth: the buffer that moves a stream between two clocks.</p><ol><li><a href="/learn/pointers-and-flags/">19 Pointers and flags</a></li><li><a href="/learn/depth-sizing/">20 Sizing the depth</a></li><li><a href="/learn/fifo-testbench/">21 The FIFO testbench</a></li></ol></li><li><h3>Module 8 · Constraints</h3><p>The lines that tell the analyzer what a clock is, which paths not to check, and what the pins must meet.</p><ol><li><a href="/learn/create-clock/">22 create_clock</a></li><li><a href="/learn/false-path-and-max-delay/">23 False paths and max delay</a></li><li><a href="/learn/io-timing/">24 IO timing</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>A CDC report, one crossing captured at the flip-flops, and the bitstream diff that closes the course.</p><ol><li><a href="/learn/cdc-report/">25 The CDC report</a></li><li><a href="/learn/capture-a-crossing/">26 Capture a crossing</a></li><li><a href="/learn/capstone-diff/">27 Capstone: change it and diff it</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/two-products-one-sum/index.html b/apps/website/public/learn/two-products-one-sum/index.html
index 92538ba57d..e52506a340 100644
--- a/apps/website/public/learn/two-products-one-sum/index.html
+++ b/apps/website/public/learn/two-products-one-sum/index.html
@@ -145,5 +145,6 @@ <h1>Two products, one sum</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The rule and its numbers</h3><p>One rule splits every width, the ratio it aims at, and the Lucas numbers behind the 3.</p><ol><li><a href="/learn/a-float-cut-by-phi/">01 A float cut by phi</a></li><li><a href="/learn/phi-as-a-ratio/">02 Phi as a ratio</a></li><li><a href="/learn/lucas-numbers-stay-whole/">03 Lucas numbers stay whole</a></li></ol></li><li><h3>Module 2 · Why phi, why three</h3><p>Why the split is phi, why base three, and how a spec checks GF16 keeps phi.</p><ol><li><a href="/learn/why-the-split-is-phi/">04 Why the split is phi</a></li><li><a href="/learn/why-base-three/">05 Why base three</a></li><li><a href="/learn/phi-in-sixteen-bits/">06 Phi in sixteen bits</a></li></ol></li><li><h3>Module 3 · The small rungs: GF4 to GF8</h3><p>GF4, GF6 and GF8, the fewest bits, where rounding to whole bits costs the most.</p><ol><li><a href="/learn/gf4-four-bits/">07 GF4: four bits</a></li><li><a href="/learn/gf6-six-bits/">08 GF6: six bits</a></li><li><a href="/learn/gf8-one-byte/">09 GF8: one byte</a></li></ol></li><li><h3>Module 4 · Ten to fourteen bits</h3><p>GF10, GF12 and GF14, and how the distance from 1 / phi moves as the word grows.</p><ol><li><a href="/learn/gf10-ten-bits/">10 GF10: ten bits</a></li><li><a href="/learn/gf12-twelve-bits/">11 GF12: twelve bits</a></li><li><a href="/learn/gf14-fourteen-bits/">12 GF14: fourteen bits</a></li></ol></li><li><h3>Module 5 · GF16 at work</h3><p>The primary 16-bit format, a two-term dot product in GF-T16, then GF20 and GF24.</p><ol><li><a href="/learn/gf16-the-primary/">13 GF16: the primary format</a></li><li class="on" aria-current="page"><a href="/learn/two-products-one-sum/">14 Two products, one sum</a></li><li><a href="/learn/gf24-twenty-four-bits/">15 GF20 and GF24</a></li></ol></li><li><h3>Module 6 · GF32 to GF64</h3><p>GF32 beside IEEE single, GF48 with no IEEE twin, GF64 beside IEEE double.</p><ol><li><a href="/learn/gf32-a-single/">16 GF32: a single</a></li><li><a href="/learn/gf48-forty-eight-bits/">17 GF48: forty-eight bits</a></li><li><a href="/learn/gf64-a-double/">18 GF64: a double</a></li></ol></li><li><h3>Module 7 · GF96 to GF256</h3><p>GF96, GF128 and GF256, where the specs hold the layout with invariants.</p><ol><li><a href="/learn/gf96-ninety-six-bits/">19 GF96: ninety-six bits</a></li><li><a href="/learn/gf128-a-quad/">20 GF128: a quad</a></li><li><a href="/learn/gf256-two-five-six/">21 GF256: two hundred fifty-six bits</a></li></ol></li><li><h3>Module 8 · The widest rungs, then trits</h3><p>GF512 and GF1024, the two widest rungs, then GF-T8, where the exponent moves to trits.</p><ol><li><a href="/learn/gf512-five-twelve/">22 GF512: five hundred twelve bits</a></li><li><a href="/learn/gf1024-one-kilobit/">23 GF1024: one kilobit</a></li><li><a href="/learn/gft8-exponent-in-trits/">24 GF-T8: an exponent in trits</a></li></ol></li><li><h3>Module 9 · More trits, then the decode</h3><p>GF-T16 and GF-T32, then why fixed fields decode in parallel and a posit does not.</p><ol><li><a href="/learn/gft16-six-trits/">25 GF-T16: six trits</a></li><li><a href="/learn/gft32-twelve-trits/">26 GF-T32: twelve trits</a></li><li><a href="/learn/decode-in-one-step/">27 Decode in one step</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/two-to-the-x/index.html b/apps/website/public/learn/two-to-the-x/index.html
index 2018d1c684..4d080ab618 100644
--- a/apps/website/public/learn/two-to-the-x/index.html
+++ b/apps/website/public/learn/two-to-the-x/index.html
@@ -137,5 +137,6 @@ <h1>Two to the x</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Lab: our own research</h3><p>A number format of our own, an honest scoreboard, and a model's tables multiplied on the board.</p><ol><li><a href="/learn/t27-number-format/">01 T27: a number format of our own</a></li><li><a href="/learn/an-honest-scoreboard/">02 An honest scoreboard</a></li><li><a href="/learn/a-model-on-the-board/">03 A model's tables on the board</a></li></ol></li><li><h3>Module 2 · AI numbers: the MX block</h3><p>How AI chips keep weights in a few bits: one shared scale per block, the scale byte itself, and what one outlier does to its neighbours.</p><ol><li><a href="/learn/one-scale-per-block/">04 One scale for a block of weights</a></li><li><a href="/learn/scale-byte-native/">05 The scale byte on a real machine</a></li><li><a href="/learn/one-outlier/">06 One outlier, many zeros</a></li></ol></li><li><h3>Module 3 · Ternary weights</h3><p>Weights that are only minus, zero or plus a scale, the five rules a ternary alphabet must pass, and a test pass that checked nothing.</p><ol><li><a href="/learn/phi-scaled-ternary/">07 Weights of minus phi, zero and plus phi</a></li><li><a href="/learn/a-pass-that-checked-nothing/">08 A pass that checked nothing</a></li><li><a href="/learn/five-rules-for-weights/">09 Five rules for a weight alphabet</a></li></ol></li><li><h3>Module 4 · The Ternary Network Float</h3><p>A rule the compiler enforces before any test runs, and a 17-bit float whose exponent is four balanced trits.</p><ol><li><a href="/learn/blocked-is-not-failed/">10 Blocked is not failed</a></li><li><a href="/learn/four-trit-exponent/">11 A float whose exponent is four trits</a></li><li><a href="/learn/negate-twice/">12 Negate twice, get the number back</a></li></ol></li><li><h3>Module 5 · Arithmetic on signed numbers</h3><p>Multiply two signed numbers, add them when their signs differ, and do both at once in a multiply-accumulate.</p><ol><li><a href="/learn/a-sign-is-one-bit/">13 A sign is one bit</a></li><li><a href="/learn/opposite-signs-subtract/">14 Opposite signs subtract</a></li><li><a href="/learn/multiply-then-add/">15 Multiply, then add</a></li></ol></li><li><h3>Module 6 · Parts of a neuron</h3><p>A ReLU that bends at zero, a power of two for softmax, and an argmax that names the answer.</p><ol><li><a href="/learn/a-bend-at-zero/">16 A bend at zero</a></li><li class="on" aria-current="page"><a href="/learn/two-to-the-x/">17 Two to the x</a></li><li><a href="/learn/pick-the-largest/">18 Pick the largest</a></li></ol></li><li><h3>Module 7 · Learning from a mistake</h3><p>A loss that prices a wrong guess in bits, one step that moves a weight against its gradient, and the hidden layer that XOR needs.</p><ol><li><a href="/learn/a-loss-in-bits/">19 A loss in bits</a></li><li><a href="/learn/one-step-downhill/">20 One step downhill</a></li><li><a href="/learn/xor-needs-a-bend/">21 XOR needs a bend</a></li></ol></li><li><h3>Module 8 · BitNet: ternary networks</h3><p>A threshold that squeezes a sum back to three values, one neuron that becomes a different function when its weights change, and a neuron that reads its inputs 27 trits at a time.</p><ol><li><a href="/learn/back-to-three-values/">22 Back to three values</a></li><li><a href="/learn/same-neuron-new-weights/">23 Same neuron, new weights</a></li><li><a href="/learn/a-neuron-in-chunks/">24 A neuron in chunks</a></li></ol></li><li><h3>Module 9 · The ternary MAC as a chip</h3><p>The 27-trit dot product as wires with no register, the same sum added into a register on every clock, and a small whole network to close the course.</p><ol><li><a href="/learn/a-dot-product-in-wires/">25 A dot product made of wires</a></li><li><a href="/learn/add-it-up-every-clock/">26 Add it up on every clock</a></li><li><a href="/learn/a-whole-network/">27 A whole network</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/use-other-modules/index.html b/apps/website/public/learn/use-other-modules/index.html
index 72304a68ff..2700b2794a 100644
--- a/apps/website/public/learn/use-other-modules/index.html
+++ b/apps/website/public/learn/use-other-modules/index.html
@@ -83,5 +83,6 @@ <h1>Using other modules</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · A spec file</h3><p>What a .t27 file holds: a module line, prose lines and comments.</p><ol><li><a href="/learn/what-a-spec-is/">01 What a spec is</a></li><li><a href="/learn/the-module-line/">02 The module line</a></li><li><a href="/learn/comments-and-prose/">03 Comments and prose lines</a></li></ol></li><li><h3>Module 2 · Values and types</h3><p>Constants, the width of a whole number, and true, false and text.</p><ol><li><a href="/learn/constants/">04 Constants</a></li><li><a href="/learn/whole-numbers/">05 Whole numbers and their widths</a></li><li><a href="/learn/true-false-and-text/">06 True, false and text</a></li></ol></li><li><h3>Module 3 · Arrays, trits and expressions</h3><p>Lists of values, the three values of a trit, and what operators compute.</p><ol><li><a href="/learn/arrays/">07 Arrays</a></li><li><a href="/learn/trits/">08 Trits: minus one, zero, one</a></li><li><a href="/learn/expressions/">09 Expressions</a></li></ol></li><li><h3>Module 4 · Tests and functions</h3><p>A test block, several small tests in one spec, and a function with typed inputs.</p><ol><li><a href="/learn/a-test-block/">10 A test block</a></li><li><a href="/learn/many-tests/">11 Many tests in one spec</a></li><li><a href="/learn/functions/">12 Functions</a></li></ol></li><li><h3>Module 5 · Inside a function</h3><p>Names with let and var, choices with if and switch, and loops.</p><ol><li><a href="/learn/local-names/">13 Local names: let and var</a></li><li><a href="/learn/if-and-else/">14 Choices: if, else and switch</a></li><li><a href="/learn/loops/">15 Loops: while and for</a></li></ol></li><li><h3>Module 6 · Visibility and your own types</h3><p>What pub shares with other modules, and the structs and enums you declare.</p><ol><li><a href="/learn/pub-or-private/">16 pub or private</a></li><li><a href="/learn/structs/">17 Structs</a></li><li><a href="/learn/enums/">18 Enums</a></li></ol></li><li><h3>Module 7 · Modules, rules and gen-ts</h3><p>Reading another module, stating a rule that always holds, and a spec turned into TypeScript.</p><ol><li class="on" aria-current="page"><a href="/learn/use-other-modules/">19 Using other modules</a></li><li><a href="/learn/invariants/">20 Invariants and benches</a></li><li><a href="/learn/gen-ts/">21 gen-ts: a spec becomes TypeScript</a></li></ol></li><li><h3>Module 8 · The compiler and tri</h3><p>The backends t27c writes, the stages it reads a spec in, and the tri command.</p><ol><li><a href="/learn/seven-backends/">22 Seven backends</a></li><li><a href="/learn/stages-of-t27c/">23 The stages of t27c</a></li><li><a href="/learn/tri-commands/">24 tri: the command line</a></li></ol></li><li><h3>Module 9 · Errors, a program, what next</h3><p>Reading a compiler error, one small complete program, and the course to take next.</p><ol><li><a href="/learn/reading-errors/">25 Reading compiler errors</a></li><li><a href="/learn/a-small-program/">26 A small complete program</a></li><li><a href="/learn/where-next/">27 Where to go next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/vcd-format/index.html b/apps/website/public/learn/vcd-format/index.html
index d728c83471..8a41c48818 100644
--- a/apps/website/public/learn/vcd-format/index.html
+++ b/apps/website/public/learn/vcd-format/index.html
@@ -362,5 +362,6 @@ <h1>The VCD format</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Why verify</h3><p>Designs that compile and are wrong, the model that decides, and the plan written before the code.</p><ol><li><a href="/learn/bugs-that-compile/">01 Bugs that compile</a></li><li><a href="/learn/golden-model/">02 The golden model</a></li><li><a href="/learn/plan-before-code/">03 Plan before code</a></li></ol></li><li><h3>Module 2 · Testbenches</h3><p>Stimulus, checks and a verdict, written as one spec beside the design it judges.</p><ol><li><a href="/learn/anatomy-of-a-tb/">04 Anatomy of a testbench</a></li><li><a href="/learn/self-checking/">05 Self-checking, not staring</a></li><li><a href="/learn/directed-vs-random/">06 Directed versus random</a></li></ol></li><li><h3>Module 3 · Waveforms</h3><p>A trace of every signal, read the way a hardware engineer reads it, and two runs compared.</p><ol><li class="on" aria-current="page"><a href="/learn/vcd-format/">07 The VCD format</a></li><li><a href="/learn/debug-with-waves/">08 Debug with waves</a></li><li><a href="/learn/compare-traces/">09 Compare two traces</a></li></ol></li><li><h3>Module 4 · Conformance vectors</h3><p>Cases with the answer written beside them, kept where the compiler can reach them.</p><ol><li><a href="/learn/what-a-vector-is/">10 What a vector is</a></li><li><a href="/learn/golden-ruler-vectors/">11 Golden ruler vectors</a></li><li><a href="/learn/end-to-end-scenarios/">12 End-to-end scenarios</a></li></ol></li><li><h3>Module 5 · Cosimulation</h3><p>Spec, simulator and board agreeing on the bench Artix-7 XC7A200T, and what to do when they do not.</p><ol><li><a href="/learn/three-way-agreement/">13 Three-way agreement</a></li><li><a href="/learn/simulator-internals/">14 Inside the simulator</a></li><li><a href="/learn/when-they-disagree/">15 When they disagree</a></li></ol></li><li><h3>Module 6 · Coverage</h3><p>What the tests touched: lines, toggles, states, and what that number hides.</p><ol><li><a href="/learn/line-and-toggle/">16 Line and toggle</a></li><li><a href="/learn/fsm-coverage/">17 FSM coverage</a></li><li><a href="/learn/coverage-lies/">18 Coverage lies</a></li></ol></li><li><h3>Module 7 · Formal</h3><p>Assertions that hold every cycle, bounded search for a counterexample, and why a proof needs induction.</p><ol><li><a href="/learn/assertions/">19 Assertions</a></li><li><a href="/learn/bounded-model-checking/">20 Bounded model checking</a></li><li><a href="/learn/induction/">21 Induction</a></li></ol></li><li><h3>Module 8 · Mutation</h3><p>Break the design on purpose and count what the tests catch.</p><ol><li><a href="/learn/kill-the-mutant/">22 Kill the mutant</a></li><li><a href="/learn/surviving-mutants/">23 Surviving mutants</a></li><li><a href="/learn/test-strength/">24 Test strength</a></li></ol></li><li><h3>Module 9 · Sign-off</h3><p>One command, every receipt, a clean verdict you can show.</p><ol><li><a href="/learn/build-verify/">25 Build and verify</a></li><li><a href="/learn/board-selftest/">26 The board self-tests</a></li><li><a href="/learn/verify-capstone/">27 Capstone: verify a module whole</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/verify-capstone/index.html b/apps/website/public/learn/verify-capstone/index.html
index 09509c798d..11f2665c3a 100644
--- a/apps/website/public/learn/verify-capstone/index.html
+++ b/apps/website/public/learn/verify-capstone/index.html
@@ -202,5 +202,6 @@ <h1>Capstone: verify a module whole</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Why verify</h3><p>Designs that compile and are wrong, the model that decides, and the plan written before the code.</p><ol><li><a href="/learn/bugs-that-compile/">01 Bugs that compile</a></li><li><a href="/learn/golden-model/">02 The golden model</a></li><li><a href="/learn/plan-before-code/">03 Plan before code</a></li></ol></li><li><h3>Module 2 · Testbenches</h3><p>Stimulus, checks and a verdict, written as one spec beside the design it judges.</p><ol><li><a href="/learn/anatomy-of-a-tb/">04 Anatomy of a testbench</a></li><li><a href="/learn/self-checking/">05 Self-checking, not staring</a></li><li><a href="/learn/directed-vs-random/">06 Directed versus random</a></li></ol></li><li><h3>Module 3 · Waveforms</h3><p>A trace of every signal, read the way a hardware engineer reads it, and two runs compared.</p><ol><li><a href="/learn/vcd-format/">07 The VCD format</a></li><li><a href="/learn/debug-with-waves/">08 Debug with waves</a></li><li><a href="/learn/compare-traces/">09 Compare two traces</a></li></ol></li><li><h3>Module 4 · Conformance vectors</h3><p>Cases with the answer written beside them, kept where the compiler can reach them.</p><ol><li><a href="/learn/what-a-vector-is/">10 What a vector is</a></li><li><a href="/learn/golden-ruler-vectors/">11 Golden ruler vectors</a></li><li><a href="/learn/end-to-end-scenarios/">12 End-to-end scenarios</a></li></ol></li><li><h3>Module 5 · Cosimulation</h3><p>Spec, simulator and board agreeing on the bench Artix-7 XC7A200T, and what to do when they do not.</p><ol><li><a href="/learn/three-way-agreement/">13 Three-way agreement</a></li><li><a href="/learn/simulator-internals/">14 Inside the simulator</a></li><li><a href="/learn/when-they-disagree/">15 When they disagree</a></li></ol></li><li><h3>Module 6 · Coverage</h3><p>What the tests touched: lines, toggles, states, and what that number hides.</p><ol><li><a href="/learn/line-and-toggle/">16 Line and toggle</a></li><li><a href="/learn/fsm-coverage/">17 FSM coverage</a></li><li><a href="/learn/coverage-lies/">18 Coverage lies</a></li></ol></li><li><h3>Module 7 · Formal</h3><p>Assertions that hold every cycle, bounded search for a counterexample, and why a proof needs induction.</p><ol><li><a href="/learn/assertions/">19 Assertions</a></li><li><a href="/learn/bounded-model-checking/">20 Bounded model checking</a></li><li><a href="/learn/induction/">21 Induction</a></li></ol></li><li><h3>Module 8 · Mutation</h3><p>Break the design on purpose and count what the tests catch.</p><ol><li><a href="/learn/kill-the-mutant/">22 Kill the mutant</a></li><li><a href="/learn/surviving-mutants/">23 Surviving mutants</a></li><li><a href="/learn/test-strength/">24 Test strength</a></li></ol></li><li><h3>Module 9 · Sign-off</h3><p>One command, every receipt, a clean verdict you can show.</p><ol><li><a href="/learn/build-verify/">25 Build and verify</a></li><li><a href="/learn/board-selftest/">26 The board self-tests</a></li><li class="on" aria-current="page"><a href="/learn/verify-capstone/">27 Capstone: verify a module whole</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/verify-hardware/index.html b/apps/website/public/learn/verify-hardware/index.html
index 311bb4abe5..ac7bcc20f9 100644
--- a/apps/website/public/learn/verify-hardware/index.html
+++ b/apps/website/public/learn/verify-hardware/index.html
@@ -67,5 +67,6 @@ <h1>Verifying hardware with t27: testbenches, waveforms, formal, mutation</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Why verify</h3><p>Designs that compile and are wrong, the model that decides, and the plan written before the code.</p><ol><li><a href="/learn/bugs-that-compile/">01 Bugs that compile</a></li><li><a href="/learn/golden-model/">02 The golden model</a></li><li><a href="/learn/plan-before-code/">03 Plan before code</a></li></ol></li><li><h3>Module 2 · Testbenches</h3><p>Stimulus, checks and a verdict, written as one spec beside the design it judges.</p><ol><li><a href="/learn/anatomy-of-a-tb/">04 Anatomy of a testbench</a></li><li><a href="/learn/self-checking/">05 Self-checking, not staring</a></li><li><a href="/learn/directed-vs-random/">06 Directed versus random</a></li></ol></li><li><h3>Module 3 · Waveforms</h3><p>A trace of every signal, read the way a hardware engineer reads it, and two runs compared.</p><ol><li><a href="/learn/vcd-format/">07 The VCD format</a></li><li><a href="/learn/debug-with-waves/">08 Debug with waves</a></li><li><a href="/learn/compare-traces/">09 Compare two traces</a></li></ol></li><li><h3>Module 4 · Conformance vectors</h3><p>Cases with the answer written beside them, kept where the compiler can reach them.</p><ol><li><a href="/learn/what-a-vector-is/">10 What a vector is</a></li><li><a href="/learn/golden-ruler-vectors/">11 Golden ruler vectors</a></li><li><a href="/learn/end-to-end-scenarios/">12 End-to-end scenarios</a></li></ol></li><li><h3>Module 5 · Cosimulation</h3><p>Spec, simulator and board agreeing on the bench Artix-7 XC7A200T, and what to do when they do not.</p><ol><li><a href="/learn/three-way-agreement/">13 Three-way agreement</a></li><li><a href="/learn/simulator-internals/">14 Inside the simulator</a></li><li><a href="/learn/when-they-disagree/">15 When they disagree</a></li></ol></li><li><h3>Module 6 · Coverage</h3><p>What the tests touched: lines, toggles, states, and what that number hides.</p><ol><li><a href="/learn/line-and-toggle/">16 Line and toggle</a></li><li><a href="/learn/fsm-coverage/">17 FSM coverage</a></li><li><a href="/learn/coverage-lies/">18 Coverage lies</a></li></ol></li><li><h3>Module 7 · Formal</h3><p>Assertions that hold every cycle, bounded search for a counterexample, and why a proof needs induction.</p><ol><li><a href="/learn/assertions/">19 Assertions</a></li><li><a href="/learn/bounded-model-checking/">20 Bounded model checking</a></li><li><a href="/learn/induction/">21 Induction</a></li></ol></li><li><h3>Module 8 · Mutation</h3><p>Break the design on purpose and count what the tests catch.</p><ol><li><a href="/learn/kill-the-mutant/">22 Kill the mutant</a></li><li><a href="/learn/surviving-mutants/">23 Surviving mutants</a></li><li><a href="/learn/test-strength/">24 Test strength</a></li></ol></li><li><h3>Module 9 · Sign-off</h3><p>One command, every receipt, a clean verdict you can show.</p><ol><li><a href="/learn/build-verify/">25 Build and verify</a></li><li><a href="/learn/board-selftest/">26 The board self-tests</a></li><li><a href="/learn/verify-capstone/">27 Capstone: verify a module whole</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/what-a-spec-is/index.html b/apps/website/public/learn/what-a-spec-is/index.html
index fddff83081..ee255deb5a 100644
--- a/apps/website/public/learn/what-a-spec-is/index.html
+++ b/apps/website/public/learn/what-a-spec-is/index.html
@@ -83,5 +83,6 @@ <h1>What a spec is</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · A spec file</h3><p>What a .t27 file holds: a module line, prose lines and comments.</p><ol><li class="on" aria-current="page"><a href="/learn/what-a-spec-is/">01 What a spec is</a></li><li><a href="/learn/the-module-line/">02 The module line</a></li><li><a href="/learn/comments-and-prose/">03 Comments and prose lines</a></li></ol></li><li><h3>Module 2 · Values and types</h3><p>Constants, the width of a whole number, and true, false and text.</p><ol><li><a href="/learn/constants/">04 Constants</a></li><li><a href="/learn/whole-numbers/">05 Whole numbers and their widths</a></li><li><a href="/learn/true-false-and-text/">06 True, false and text</a></li></ol></li><li><h3>Module 3 · Arrays, trits and expressions</h3><p>Lists of values, the three values of a trit, and what operators compute.</p><ol><li><a href="/learn/arrays/">07 Arrays</a></li><li><a href="/learn/trits/">08 Trits: minus one, zero, one</a></li><li><a href="/learn/expressions/">09 Expressions</a></li></ol></li><li><h3>Module 4 · Tests and functions</h3><p>A test block, several small tests in one spec, and a function with typed inputs.</p><ol><li><a href="/learn/a-test-block/">10 A test block</a></li><li><a href="/learn/many-tests/">11 Many tests in one spec</a></li><li><a href="/learn/functions/">12 Functions</a></li></ol></li><li><h3>Module 5 · Inside a function</h3><p>Names with let and var, choices with if and switch, and loops.</p><ol><li><a href="/learn/local-names/">13 Local names: let and var</a></li><li><a href="/learn/if-and-else/">14 Choices: if, else and switch</a></li><li><a href="/learn/loops/">15 Loops: while and for</a></li></ol></li><li><h3>Module 6 · Visibility and your own types</h3><p>What pub shares with other modules, and the structs and enums you declare.</p><ol><li><a href="/learn/pub-or-private/">16 pub or private</a></li><li><a href="/learn/structs/">17 Structs</a></li><li><a href="/learn/enums/">18 Enums</a></li></ol></li><li><h3>Module 7 · Modules, rules and gen-ts</h3><p>Reading another module, stating a rule that always holds, and a spec turned into TypeScript.</p><ol><li><a href="/learn/use-other-modules/">19 Using other modules</a></li><li><a href="/learn/invariants/">20 Invariants and benches</a></li><li><a href="/learn/gen-ts/">21 gen-ts: a spec becomes TypeScript</a></li></ol></li><li><h3>Module 8 · The compiler and tri</h3><p>The backends t27c writes, the stages it reads a spec in, and the tri command.</p><ol><li><a href="/learn/seven-backends/">22 Seven backends</a></li><li><a href="/learn/stages-of-t27c/">23 The stages of t27c</a></li><li><a href="/learn/tri-commands/">24 tri: the command line</a></li></ol></li><li><h3>Module 9 · Errors, a program, what next</h3><p>Reading a compiler error, one small complete program, and the course to take next.</p><ol><li><a href="/learn/reading-errors/">25 Reading compiler errors</a></li><li><a href="/learn/a-small-program/">26 A small complete program</a></li><li><a href="/learn/where-next/">27 Where to go next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/what-a-vector-is/index.html b/apps/website/public/learn/what-a-vector-is/index.html
index 91018790ce..0b02b28aa0 100644
--- a/apps/website/public/learn/what-a-vector-is/index.html
+++ b/apps/website/public/learn/what-a-vector-is/index.html
@@ -565,5 +565,6 @@ <h1>What a vector is</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Why verify</h3><p>Designs that compile and are wrong, the model that decides, and the plan written before the code.</p><ol><li><a href="/learn/bugs-that-compile/">01 Bugs that compile</a></li><li><a href="/learn/golden-model/">02 The golden model</a></li><li><a href="/learn/plan-before-code/">03 Plan before code</a></li></ol></li><li><h3>Module 2 · Testbenches</h3><p>Stimulus, checks and a verdict, written as one spec beside the design it judges.</p><ol><li><a href="/learn/anatomy-of-a-tb/">04 Anatomy of a testbench</a></li><li><a href="/learn/self-checking/">05 Self-checking, not staring</a></li><li><a href="/learn/directed-vs-random/">06 Directed versus random</a></li></ol></li><li><h3>Module 3 · Waveforms</h3><p>A trace of every signal, read the way a hardware engineer reads it, and two runs compared.</p><ol><li><a href="/learn/vcd-format/">07 The VCD format</a></li><li><a href="/learn/debug-with-waves/">08 Debug with waves</a></li><li><a href="/learn/compare-traces/">09 Compare two traces</a></li></ol></li><li><h3>Module 4 · Conformance vectors</h3><p>Cases with the answer written beside them, kept where the compiler can reach them.</p><ol><li class="on" aria-current="page"><a href="/learn/what-a-vector-is/">10 What a vector is</a></li><li><a href="/learn/golden-ruler-vectors/">11 Golden ruler vectors</a></li><li><a href="/learn/end-to-end-scenarios/">12 End-to-end scenarios</a></li></ol></li><li><h3>Module 5 · Cosimulation</h3><p>Spec, simulator and board agreeing on the bench Artix-7 XC7A200T, and what to do when they do not.</p><ol><li><a href="/learn/three-way-agreement/">13 Three-way agreement</a></li><li><a href="/learn/simulator-internals/">14 Inside the simulator</a></li><li><a href="/learn/when-they-disagree/">15 When they disagree</a></li></ol></li><li><h3>Module 6 · Coverage</h3><p>What the tests touched: lines, toggles, states, and what that number hides.</p><ol><li><a href="/learn/line-and-toggle/">16 Line and toggle</a></li><li><a href="/learn/fsm-coverage/">17 FSM coverage</a></li><li><a href="/learn/coverage-lies/">18 Coverage lies</a></li></ol></li><li><h3>Module 7 · Formal</h3><p>Assertions that hold every cycle, bounded search for a counterexample, and why a proof needs induction.</p><ol><li><a href="/learn/assertions/">19 Assertions</a></li><li><a href="/learn/bounded-model-checking/">20 Bounded model checking</a></li><li><a href="/learn/induction/">21 Induction</a></li></ol></li><li><h3>Module 8 · Mutation</h3><p>Break the design on purpose and count what the tests catch.</p><ol><li><a href="/learn/kill-the-mutant/">22 Kill the mutant</a></li><li><a href="/learn/surviving-mutants/">23 Surviving mutants</a></li><li><a href="/learn/test-strength/">24 Test strength</a></li></ol></li><li><h3>Module 9 · Sign-off</h3><p>One command, every receipt, a clean verdict you can show.</p><ol><li><a href="/learn/build-verify/">25 Build and verify</a></li><li><a href="/learn/board-selftest/">26 The board self-tests</a></li><li><a href="/learn/verify-capstone/">27 Capstone: verify a module whole</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/what-arithmetic-costs/index.html b/apps/website/public/learn/what-arithmetic-costs/index.html
index cd9a5de133..ee0ae3d440 100644
--- a/apps/website/public/learn/what-arithmetic-costs/index.html
+++ b/apps/website/public/learn/what-arithmetic-costs/index.html
@@ -185,5 +185,6 @@ <h1>What arithmetic costs</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The chip</h3><p>What an FPGA is, which chip we use, and how its pins meet the board.</p><ol><li><a href="/learn/fpga-is-a-grid/">01 An FPGA is a grid you configure</a></li><li><a href="/learn/which-chip/">02 Which chip is this?</a></li><li><a href="/learn/pins-and-package/">03 Pins, balls and the package</a></li></ol></li><li><h3>Module 2 · Numbers in hardware</h3><p>Bits, trits and number formats, and what arithmetic costs in logic.</p><ol><li><a href="/learn/bits-and-trits/">04 Bits, trits and GF16</a></li><li class="on" aria-current="page"><a href="/learn/what-arithmetic-costs/">05 What arithmetic costs</a></li><li><a href="/learn/how-much-fits/">06 How much fits on the chip</a></li></ol></li><li><h3>Module 3 · Your t27 program</h3><p>Write a spec, test it, and see why compiling is not the same as being right.</p><ol><li><a href="/learn/hello-t27/">07 Hello, t27</a></li><li><a href="/learn/tests-are-the-spec/">08 Tests are part of the spec</a></li><li><a href="/learn/compiles-is-not-correct/">09 Compiles is not the same as correct</a></li></ol></li><li><h3>Module 4 · Inside t27c</h3><p>How the compiler reads a spec and what it writes, including the native t27b.</p><ol><li><a href="/learn/compiler-stages/">10 The stages of t27c</a></li><li><a href="/learn/t27c-on-a-real-machine/">11 t27c on a real machine</a></li><li><a href="/learn/seven-backends-and-t27b/">12 Seven backends, and t27b</a></li></ol></li><li><h3>Module 5 · From spec to hardware</h3><p>The Verilog t27c writes, the cells it becomes, and what one LUT does.</p><ol><li><a href="/learn/spec-to-verilog/">13 From a spec to Verilog</a></li><li><a href="/learn/count-the-cells/">14 Count the cells</a></li><li><a href="/learn/one-lut-up-close/">15 One LUT, up close</a></li></ol></li><li><h3>Module 6 · Reading synthesis</h3><p>What yosys reports about your design, and which warnings matter.</p><ol><li><a href="/learn/synthesis-warnings/">16 Synthesis warnings that matter</a></li><li><a href="/learn/state-machines/">17 State machines</a></li><li><a href="/learn/where-the-luts-go/">18 Where the LUTs go</a></li></ol></li><li><h3>Module 7 · Place, route, timing</h3><p>Where the cells land on the die, and whether the clock is met.</p><ol><li><a href="/learn/placement/">19 Placement: where the cells land</a></li><li><a href="/learn/logic-depth/">20 Logic depth</a></li><li><a href="/learn/timing-slack/">21 Timing and slack</a></li></ol></li><li><h3>Module 8 · The bitstream</h3><p>How a routed design becomes the bits the chip loads, with no Vivado.</p><ol><li><a href="/learn/fasm-to-frames/">22 From FASM to frames</a></li><li><a href="/learn/a-build-receipt/">23 A build, itemised</a></li><li><a href="/learn/read-a-bitstream/">24 Read a bitstream</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>Ask the chip who it is, load the bits, and check them.</p><ol><li><a href="/learn/jtag-asks-the-chip/">25 JTAG asks the chip</a></li><li><a href="/learn/same-bits-on-the-board/">26 The same bits, on the board</a></li><li><a href="/learn/capstone/">27 Capstone: out-guess yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/when-they-disagree/index.html b/apps/website/public/learn/when-they-disagree/index.html
index 15e42ded9e..c8d5f42f45 100644
--- a/apps/website/public/learn/when-they-disagree/index.html
+++ b/apps/website/public/learn/when-they-disagree/index.html
@@ -107,5 +107,6 @@ <h1>When they disagree</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Why verify</h3><p>Designs that compile and are wrong, the model that decides, and the plan written before the code.</p><ol><li><a href="/learn/bugs-that-compile/">01 Bugs that compile</a></li><li><a href="/learn/golden-model/">02 The golden model</a></li><li><a href="/learn/plan-before-code/">03 Plan before code</a></li></ol></li><li><h3>Module 2 · Testbenches</h3><p>Stimulus, checks and a verdict, written as one spec beside the design it judges.</p><ol><li><a href="/learn/anatomy-of-a-tb/">04 Anatomy of a testbench</a></li><li><a href="/learn/self-checking/">05 Self-checking, not staring</a></li><li><a href="/learn/directed-vs-random/">06 Directed versus random</a></li></ol></li><li><h3>Module 3 · Waveforms</h3><p>A trace of every signal, read the way a hardware engineer reads it, and two runs compared.</p><ol><li><a href="/learn/vcd-format/">07 The VCD format</a></li><li><a href="/learn/debug-with-waves/">08 Debug with waves</a></li><li><a href="/learn/compare-traces/">09 Compare two traces</a></li></ol></li><li><h3>Module 4 · Conformance vectors</h3><p>Cases with the answer written beside them, kept where the compiler can reach them.</p><ol><li><a href="/learn/what-a-vector-is/">10 What a vector is</a></li><li><a href="/learn/golden-ruler-vectors/">11 Golden ruler vectors</a></li><li><a href="/learn/end-to-end-scenarios/">12 End-to-end scenarios</a></li></ol></li><li><h3>Module 5 · Cosimulation</h3><p>Spec, simulator and board agreeing on the bench Artix-7 XC7A200T, and what to do when they do not.</p><ol><li><a href="/learn/three-way-agreement/">13 Three-way agreement</a></li><li><a href="/learn/simulator-internals/">14 Inside the simulator</a></li><li class="on" aria-current="page"><a href="/learn/when-they-disagree/">15 When they disagree</a></li></ol></li><li><h3>Module 6 · Coverage</h3><p>What the tests touched: lines, toggles, states, and what that number hides.</p><ol><li><a href="/learn/line-and-toggle/">16 Line and toggle</a></li><li><a href="/learn/fsm-coverage/">17 FSM coverage</a></li><li><a href="/learn/coverage-lies/">18 Coverage lies</a></li></ol></li><li><h3>Module 7 · Formal</h3><p>Assertions that hold every cycle, bounded search for a counterexample, and why a proof needs induction.</p><ol><li><a href="/learn/assertions/">19 Assertions</a></li><li><a href="/learn/bounded-model-checking/">20 Bounded model checking</a></li><li><a href="/learn/induction/">21 Induction</a></li></ol></li><li><h3>Module 8 · Mutation</h3><p>Break the design on purpose and count what the tests catch.</p><ol><li><a href="/learn/kill-the-mutant/">22 Kill the mutant</a></li><li><a href="/learn/surviving-mutants/">23 Surviving mutants</a></li><li><a href="/learn/test-strength/">24 Test strength</a></li></ol></li><li><h3>Module 9 · Sign-off</h3><p>One command, every receipt, a clean verdict you can show.</p><ol><li><a href="/learn/build-verify/">25 Build and verify</a></li><li><a href="/learn/board-selftest/">26 The board self-tests</a></li><li><a href="/learn/verify-capstone/">27 Capstone: verify a module whole</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/where-next/index.html b/apps/website/public/learn/where-next/index.html
index eb98fd503e..670a27cf03 100644
--- a/apps/website/public/learn/where-next/index.html
+++ b/apps/website/public/learn/where-next/index.html
@@ -84,5 +84,6 @@ <h1>Where to go next</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · A spec file</h3><p>What a .t27 file holds: a module line, prose lines and comments.</p><ol><li><a href="/learn/what-a-spec-is/">01 What a spec is</a></li><li><a href="/learn/the-module-line/">02 The module line</a></li><li><a href="/learn/comments-and-prose/">03 Comments and prose lines</a></li></ol></li><li><h3>Module 2 · Values and types</h3><p>Constants, the width of a whole number, and true, false and text.</p><ol><li><a href="/learn/constants/">04 Constants</a></li><li><a href="/learn/whole-numbers/">05 Whole numbers and their widths</a></li><li><a href="/learn/true-false-and-text/">06 True, false and text</a></li></ol></li><li><h3>Module 3 · Arrays, trits and expressions</h3><p>Lists of values, the three values of a trit, and what operators compute.</p><ol><li><a href="/learn/arrays/">07 Arrays</a></li><li><a href="/learn/trits/">08 Trits: minus one, zero, one</a></li><li><a href="/learn/expressions/">09 Expressions</a></li></ol></li><li><h3>Module 4 · Tests and functions</h3><p>A test block, several small tests in one spec, and a function with typed inputs.</p><ol><li><a href="/learn/a-test-block/">10 A test block</a></li><li><a href="/learn/many-tests/">11 Many tests in one spec</a></li><li><a href="/learn/functions/">12 Functions</a></li></ol></li><li><h3>Module 5 · Inside a function</h3><p>Names with let and var, choices with if and switch, and loops.</p><ol><li><a href="/learn/local-names/">13 Local names: let and var</a></li><li><a href="/learn/if-and-else/">14 Choices: if, else and switch</a></li><li><a href="/learn/loops/">15 Loops: while and for</a></li></ol></li><li><h3>Module 6 · Visibility and your own types</h3><p>What pub shares with other modules, and the structs and enums you declare.</p><ol><li><a href="/learn/pub-or-private/">16 pub or private</a></li><li><a href="/learn/structs/">17 Structs</a></li><li><a href="/learn/enums/">18 Enums</a></li></ol></li><li><h3>Module 7 · Modules, rules and gen-ts</h3><p>Reading another module, stating a rule that always holds, and a spec turned into TypeScript.</p><ol><li><a href="/learn/use-other-modules/">19 Using other modules</a></li><li><a href="/learn/invariants/">20 Invariants and benches</a></li><li><a href="/learn/gen-ts/">21 gen-ts: a spec becomes TypeScript</a></li></ol></li><li><h3>Module 8 · The compiler and tri</h3><p>The backends t27c writes, the stages it reads a spec in, and the tri command.</p><ol><li><a href="/learn/seven-backends/">22 Seven backends</a></li><li><a href="/learn/stages-of-t27c/">23 The stages of t27c</a></li><li><a href="/learn/tri-commands/">24 tri: the command line</a></li></ol></li><li><h3>Module 9 · Errors, a program, what next</h3><p>Reading a compiler error, one small complete program, and the course to take next.</p><ol><li><a href="/learn/reading-errors/">25 Reading compiler errors</a></li><li><a href="/learn/a-small-program/">26 A small complete program</a></li><li class="on" aria-current="page"><a href="/learn/where-next/">27 Where to go next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/where-the-clock-enters/index.html b/apps/website/public/learn/where-the-clock-enters/index.html
index 2e4db0c1d3..fb18281224 100644
--- a/apps/website/public/learn/where-the-clock-enters/index.html
+++ b/apps/website/public/learn/where-the-clock-enters/index.html
@@ -367,5 +367,6 @@ <h1>Where the clock enters</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a clock is</h3><p>One edge, one world: what shares a clock edge shares a world, the period and the jitter of a real edge, and where the clock enters a board.</p><ol><li><a href="/learn/one-edge-one-world/">01 One edge, one world</a></li><li><a href="/learn/period-and-jitter/">02 A period, an edge, and jitter</a></li><li class="on" aria-current="page"><a href="/learn/where-the-clock-enters/">03 Where the clock enters</a></li></ol></li><li><h3>Module 2 · Clock trees</h3><p>Skew and insertion delay, the global buffer network, and the trap of gating a clock with logic.</p><ol><li><a href="/learn/skew-and-insertion/">04 Skew and insertion delay</a></li><li><a href="/learn/global-buffers/">05 The global buffer</a></li><li><a href="/learn/gated-clock-trap/">06 The gated-clock trap</a></li></ol></li><li><h3>Module 3 · PLL and MMCM</h3><p>Multiply and divide one clock into another, move its phase in steps of the VCO, and which clocks the analyzer treats as related.</p><ol><li><a href="/learn/multiply-and-divide/">07 Multiply and divide</a></li><li><a href="/learn/phase-and-lock/">08 Phase and lock</a></li><li><a href="/learn/related-or-not/">09 Related or not</a></li></ol></li><li><h3>Module 4 · Resets</h3><p>Assert asynchronously, release synchronously: the three reset kinds, the release pipe, and the tree a reset grows.</p><ol><li><a href="/learn/sync-or-async/">10 Sync or async</a></li><li><a href="/learn/reset-release/">11 The release pipe</a></li><li><a href="/learn/reset-fanout/">12 Reset fanout</a></li></ol></li><li><h3>Module 5 · Metastability</h3><p>The setup-hold window, the mean time between failures in integer arithmetic, and the two flops that fix it.</p><ol><li><a href="/learn/setup-hold-window/">13 The setup-hold window</a></li><li><a href="/learn/mtbf/">14 Mean time between failures</a></li><li><a href="/learn/two-flop-sync/">15 The two-flop synchronizer</a></li></ol></li><li><h3>Module 6 · Crossing many bits</h3><p>Why a binary bus tears, why Gray code does not, and the handshake that moves a pulse between worlds.</p><ol><li><a href="/learn/why-binary-fails/">16 Why binary fails</a></li><li><a href="/learn/gray-code/">17 Gray code</a></li><li><a href="/learn/handshakes-and-pulses/">18 Handshakes and pulses</a></li></ol></li><li><h3>Module 7 · The asynchronous FIFO</h3><p>Pointers, flags and depth: the buffer that moves a stream between two clocks.</p><ol><li><a href="/learn/pointers-and-flags/">19 Pointers and flags</a></li><li><a href="/learn/depth-sizing/">20 Sizing the depth</a></li><li><a href="/learn/fifo-testbench/">21 The FIFO testbench</a></li></ol></li><li><h3>Module 8 · Constraints</h3><p>The lines that tell the analyzer what a clock is, which paths not to check, and what the pins must meet.</p><ol><li><a href="/learn/create-clock/">22 create_clock</a></li><li><a href="/learn/false-path-and-max-delay/">23 False paths and max delay</a></li><li><a href="/learn/io-timing/">24 IO timing</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>A CDC report, one crossing captured at the flip-flops, and the bitstream diff that closes the course.</p><ol><li><a href="/learn/cdc-report/">25 The CDC report</a></li><li><a href="/learn/capture-a-crossing/">26 Capture a crossing</a></li><li><a href="/learn/capstone-diff/">27 Capstone: change it and diff it</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/where-the-luts-go/index.html b/apps/website/public/learn/where-the-luts-go/index.html
index d4401861dc..accdaba014 100644
--- a/apps/website/public/learn/where-the-luts-go/index.html
+++ b/apps/website/public/learn/where-the-luts-go/index.html
@@ -330,5 +330,6 @@ <h1>Where the LUTs go</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The chip</h3><p>What an FPGA is, which chip we use, and how its pins meet the board.</p><ol><li><a href="/learn/fpga-is-a-grid/">01 An FPGA is a grid you configure</a></li><li><a href="/learn/which-chip/">02 Which chip is this?</a></li><li><a href="/learn/pins-and-package/">03 Pins, balls and the package</a></li></ol></li><li><h3>Module 2 · Numbers in hardware</h3><p>Bits, trits and number formats, and what arithmetic costs in logic.</p><ol><li><a href="/learn/bits-and-trits/">04 Bits, trits and GF16</a></li><li><a href="/learn/what-arithmetic-costs/">05 What arithmetic costs</a></li><li><a href="/learn/how-much-fits/">06 How much fits on the chip</a></li></ol></li><li><h3>Module 3 · Your t27 program</h3><p>Write a spec, test it, and see why compiling is not the same as being right.</p><ol><li><a href="/learn/hello-t27/">07 Hello, t27</a></li><li><a href="/learn/tests-are-the-spec/">08 Tests are part of the spec</a></li><li><a href="/learn/compiles-is-not-correct/">09 Compiles is not the same as correct</a></li></ol></li><li><h3>Module 4 · Inside t27c</h3><p>How the compiler reads a spec and what it writes, including the native t27b.</p><ol><li><a href="/learn/compiler-stages/">10 The stages of t27c</a></li><li><a href="/learn/t27c-on-a-real-machine/">11 t27c on a real machine</a></li><li><a href="/learn/seven-backends-and-t27b/">12 Seven backends, and t27b</a></li></ol></li><li><h3>Module 5 · From spec to hardware</h3><p>The Verilog t27c writes, the cells it becomes, and what one LUT does.</p><ol><li><a href="/learn/spec-to-verilog/">13 From a spec to Verilog</a></li><li><a href="/learn/count-the-cells/">14 Count the cells</a></li><li><a href="/learn/one-lut-up-close/">15 One LUT, up close</a></li></ol></li><li><h3>Module 6 · Reading synthesis</h3><p>What yosys reports about your design, and which warnings matter.</p><ol><li><a href="/learn/synthesis-warnings/">16 Synthesis warnings that matter</a></li><li><a href="/learn/state-machines/">17 State machines</a></li><li class="on" aria-current="page"><a href="/learn/where-the-luts-go/">18 Where the LUTs go</a></li></ol></li><li><h3>Module 7 · Place, route, timing</h3><p>Where the cells land on the die, and whether the clock is met.</p><ol><li><a href="/learn/placement/">19 Placement: where the cells land</a></li><li><a href="/learn/logic-depth/">20 Logic depth</a></li><li><a href="/learn/timing-slack/">21 Timing and slack</a></li></ol></li><li><h3>Module 8 · The bitstream</h3><p>How a routed design becomes the bits the chip loads, with no Vivado.</p><ol><li><a href="/learn/fasm-to-frames/">22 From FASM to frames</a></li><li><a href="/learn/a-build-receipt/">23 A build, itemised</a></li><li><a href="/learn/read-a-bitstream/">24 Read a bitstream</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>Ask the chip who it is, load the bits, and check them.</p><ol><li><a href="/learn/jtag-asks-the-chip/">25 JTAG asks the chip</a></li><li><a href="/learn/same-bits-on-the-board/">26 The same bits, on the board</a></li><li><a href="/learn/capstone/">27 Capstone: out-guess yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/which-chip/index.html b/apps/website/public/learn/which-chip/index.html
index 1be7b58568..16bc4c7bb3 100644
--- a/apps/website/public/learn/which-chip/index.html
+++ b/apps/website/public/learn/which-chip/index.html
@@ -113,5 +113,6 @@ <h1>Which chip is this?</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The chip</h3><p>What an FPGA is, which chip we use, and how its pins meet the board.</p><ol><li><a href="/learn/fpga-is-a-grid/">01 An FPGA is a grid you configure</a></li><li class="on" aria-current="page"><a href="/learn/which-chip/">02 Which chip is this?</a></li><li><a href="/learn/pins-and-package/">03 Pins, balls and the package</a></li></ol></li><li><h3>Module 2 · Numbers in hardware</h3><p>Bits, trits and number formats, and what arithmetic costs in logic.</p><ol><li><a href="/learn/bits-and-trits/">04 Bits, trits and GF16</a></li><li><a href="/learn/what-arithmetic-costs/">05 What arithmetic costs</a></li><li><a href="/learn/how-much-fits/">06 How much fits on the chip</a></li></ol></li><li><h3>Module 3 · Your t27 program</h3><p>Write a spec, test it, and see why compiling is not the same as being right.</p><ol><li><a href="/learn/hello-t27/">07 Hello, t27</a></li><li><a href="/learn/tests-are-the-spec/">08 Tests are part of the spec</a></li><li><a href="/learn/compiles-is-not-correct/">09 Compiles is not the same as correct</a></li></ol></li><li><h3>Module 4 · Inside t27c</h3><p>How the compiler reads a spec and what it writes, including the native t27b.</p><ol><li><a href="/learn/compiler-stages/">10 The stages of t27c</a></li><li><a href="/learn/t27c-on-a-real-machine/">11 t27c on a real machine</a></li><li><a href="/learn/seven-backends-and-t27b/">12 Seven backends, and t27b</a></li></ol></li><li><h3>Module 5 · From spec to hardware</h3><p>The Verilog t27c writes, the cells it becomes, and what one LUT does.</p><ol><li><a href="/learn/spec-to-verilog/">13 From a spec to Verilog</a></li><li><a href="/learn/count-the-cells/">14 Count the cells</a></li><li><a href="/learn/one-lut-up-close/">15 One LUT, up close</a></li></ol></li><li><h3>Module 6 · Reading synthesis</h3><p>What yosys reports about your design, and which warnings matter.</p><ol><li><a href="/learn/synthesis-warnings/">16 Synthesis warnings that matter</a></li><li><a href="/learn/state-machines/">17 State machines</a></li><li><a href="/learn/where-the-luts-go/">18 Where the LUTs go</a></li></ol></li><li><h3>Module 7 · Place, route, timing</h3><p>Where the cells land on the die, and whether the clock is met.</p><ol><li><a href="/learn/placement/">19 Placement: where the cells land</a></li><li><a href="/learn/logic-depth/">20 Logic depth</a></li><li><a href="/learn/timing-slack/">21 Timing and slack</a></li></ol></li><li><h3>Module 8 · The bitstream</h3><p>How a routed design becomes the bits the chip loads, with no Vivado.</p><ol><li><a href="/learn/fasm-to-frames/">22 From FASM to frames</a></li><li><a href="/learn/a-build-receipt/">23 A build, itemised</a></li><li><a href="/learn/read-a-bitstream/">24 Read a bitstream</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>Ask the chip who it is, load the bits, and check them.</p><ol><li><a href="/learn/jtag-asks-the-chip/">25 JTAG asks the chip</a></li><li><a href="/learn/same-bits-on-the-board/">26 The same bits, on the board</a></li><li><a href="/learn/capstone/">27 Capstone: out-guess yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/who-holds-the-io/index.html b/apps/website/public/learn/who-holds-the-io/index.html
index f2c299c634..18d7eae880 100644
--- a/apps/website/public/learn/who-holds-the-io/index.html
+++ b/apps/website/public/learn/who-holds-the-io/index.html
@@ -109,5 +109,6 @@ <h1>Who holds the IO</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a bus is</h3><p>Why a bus exists at all: a conversation on wires, framed and addressed, and who is allowed to talk.</p><ol><li><a href="/learn/wires-that-talk/">01 Wires that talk</a></li><li><a href="/learn/framing-a-conversation/">02 Framing a conversation</a></li><li><a href="/learn/master-and-servant/">03 Master and servant</a></li></ol></li><li><h3>Module 2 · UART</h3><p>The two-wire bus with no clock: the frame, the divisor that sets the rate, and the status a driver polls.</p><ol><li><a href="/learn/the-frame/">04 The frame: start, data, stop</a></li><li><a href="/learn/baud-and-the-divisor/">05 Baud and the divisor</a></li><li><a href="/learn/status-and-fifos/">06 Status and FIFOs</a></li></ol></li><li><h3>Module 3 · SPI</h3><p>The clocked conversation: four modes, a prescaler ladder, and a chip select per servant.</p><ol><li><a href="/learn/modes-cpol-cpha/">07 Modes: CPOL and CPHA</a></li><li><a href="/learn/the-prescaler-ladder/">08 The prescaler ladder</a></li><li><a href="/learn/chip-select-and-widths/">09 Chip select and widths</a></li></ol></li><li><h3>Module 4 · APB</h3><p>The register bus: PSEL and PENABLE, strobes and wait states, and how many address bits a peripheral count costs.</p><ol><li><a href="/learn/psel-and-penable/">10 PSEL and PENABLE</a></li><li><a href="/learn/strobes-and-waits/">11 Strobes and waits</a></li><li><a href="/learn/address-decode/">12 Address decode</a></li></ol></li><li><h3>Module 5 · AXI4</h3><p>The five channels: address, data and response in both directions, lite or full, bursts and IDs.</p><ol><li><a href="/learn/five-channels/">13 Five channels</a></li><li><a href="/learn/lite-or-full/">14 Lite or full</a></li><li><a href="/learn/bursts-and-ids/">15 Bursts and IDs</a></li></ol></li><li><h3>Module 6 · Memory</h3><p>What sits on the far side of every bus: memory maps, port kinds, and latency that a wait state must cover.</p><ol><li><a href="/learn/memory-maps/">16 Memory maps</a></li><li><a href="/learn/ports-and-kinds/">17 Ports and kinds</a></li><li><a href="/learn/latency/">18 Latency</a></li></ol></li><li><h3>Module 7 · Bridges</h3><p>Why designs grow more than one bus, and the packet bridge that moves work between them.</p><ol><li><a href="/learn/why-bridges/">19 Why bridges</a></li><li><a href="/learn/the-packet-bridge/">20 The packet bridge</a></li><li><a href="/learn/everything-in-one-bench/">21 Everything in one bench</a></li></ol></li><li><h3>Module 8 · Ethernet</h3><p>Frames, the frame check sequence, RGMII timing, and the pre-registered steps of a real bring-up.</p><ol><li><a href="/learn/frames-and-the-fcs/">22 Frames and the FCS</a></li><li><a href="/learn/rgmii-timing/">23 RGMII timing</a></li><li><a href="/learn/bring-up-on-the-bench/">24 Bring-up on the bench</a></li></ol></li><li><h3>Module 9 · The bench</h3><p>The discipline that guards real hardware: who holds the IO, taking and giving the lock, and what runs next.</p><ol><li class="on" aria-current="page"><a href="/learn/who-holds-the-io/">25 Who holds the IO</a></li><li><a href="/learn/claim-and-release/">26 Claim and release</a></li><li><a href="/learn/capstone-what-runs-next/">27 Capstone: what runs next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/whole-numbers/index.html b/apps/website/public/learn/whole-numbers/index.html
index 4ab6859d85..72499a70b7 100644
--- a/apps/website/public/learn/whole-numbers/index.html
+++ b/apps/website/public/learn/whole-numbers/index.html
@@ -85,5 +85,6 @@ <h1>Whole numbers and their widths</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · A spec file</h3><p>What a .t27 file holds: a module line, prose lines and comments.</p><ol><li><a href="/learn/what-a-spec-is/">01 What a spec is</a></li><li><a href="/learn/the-module-line/">02 The module line</a></li><li><a href="/learn/comments-and-prose/">03 Comments and prose lines</a></li></ol></li><li><h3>Module 2 · Values and types</h3><p>Constants, the width of a whole number, and true, false and text.</p><ol><li><a href="/learn/constants/">04 Constants</a></li><li class="on" aria-current="page"><a href="/learn/whole-numbers/">05 Whole numbers and their widths</a></li><li><a href="/learn/true-false-and-text/">06 True, false and text</a></li></ol></li><li><h3>Module 3 · Arrays, trits and expressions</h3><p>Lists of values, the three values of a trit, and what operators compute.</p><ol><li><a href="/learn/arrays/">07 Arrays</a></li><li><a href="/learn/trits/">08 Trits: minus one, zero, one</a></li><li><a href="/learn/expressions/">09 Expressions</a></li></ol></li><li><h3>Module 4 · Tests and functions</h3><p>A test block, several small tests in one spec, and a function with typed inputs.</p><ol><li><a href="/learn/a-test-block/">10 A test block</a></li><li><a href="/learn/many-tests/">11 Many tests in one spec</a></li><li><a href="/learn/functions/">12 Functions</a></li></ol></li><li><h3>Module 5 · Inside a function</h3><p>Names with let and var, choices with if and switch, and loops.</p><ol><li><a href="/learn/local-names/">13 Local names: let and var</a></li><li><a href="/learn/if-and-else/">14 Choices: if, else and switch</a></li><li><a href="/learn/loops/">15 Loops: while and for</a></li></ol></li><li><h3>Module 6 · Visibility and your own types</h3><p>What pub shares with other modules, and the structs and enums you declare.</p><ol><li><a href="/learn/pub-or-private/">16 pub or private</a></li><li><a href="/learn/structs/">17 Structs</a></li><li><a href="/learn/enums/">18 Enums</a></li></ol></li><li><h3>Module 7 · Modules, rules and gen-ts</h3><p>Reading another module, stating a rule that always holds, and a spec turned into TypeScript.</p><ol><li><a href="/learn/use-other-modules/">19 Using other modules</a></li><li><a href="/learn/invariants/">20 Invariants and benches</a></li><li><a href="/learn/gen-ts/">21 gen-ts: a spec becomes TypeScript</a></li></ol></li><li><h3>Module 8 · The compiler and tri</h3><p>The backends t27c writes, the stages it reads a spec in, and the tri command.</p><ol><li><a href="/learn/seven-backends/">22 Seven backends</a></li><li><a href="/learn/stages-of-t27c/">23 The stages of t27c</a></li><li><a href="/learn/tri-commands/">24 tri: the command line</a></li></ol></li><li><h3>Module 9 · Errors, a program, what next</h3><p>Reading a compiler error, one small complete program, and the course to take next.</p><ol><li><a href="/learn/reading-errors/">25 Reading compiler errors</a></li><li><a href="/learn/a-small-program/">26 A small complete program</a></li><li><a href="/learn/where-next/">27 Where to go next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/why-base-three/index.html b/apps/website/public/learn/why-base-three/index.html
index 80cbf1a3c0..5dd0a28a6b 100644
--- a/apps/website/public/learn/why-base-three/index.html
+++ b/apps/website/public/learn/why-base-three/index.html
@@ -414,5 +414,6 @@ <h1>Why base three</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The rule and its numbers</h3><p>One rule splits every width, the ratio it aims at, and the Lucas numbers behind the 3.</p><ol><li><a href="/learn/a-float-cut-by-phi/">01 A float cut by phi</a></li><li><a href="/learn/phi-as-a-ratio/">02 Phi as a ratio</a></li><li><a href="/learn/lucas-numbers-stay-whole/">03 Lucas numbers stay whole</a></li></ol></li><li><h3>Module 2 · Why phi, why three</h3><p>Why the split is phi, why base three, and how a spec checks GF16 keeps phi.</p><ol><li><a href="/learn/why-the-split-is-phi/">04 Why the split is phi</a></li><li class="on" aria-current="page"><a href="/learn/why-base-three/">05 Why base three</a></li><li><a href="/learn/phi-in-sixteen-bits/">06 Phi in sixteen bits</a></li></ol></li><li><h3>Module 3 · The small rungs: GF4 to GF8</h3><p>GF4, GF6 and GF8, the fewest bits, where rounding to whole bits costs the most.</p><ol><li><a href="/learn/gf4-four-bits/">07 GF4: four bits</a></li><li><a href="/learn/gf6-six-bits/">08 GF6: six bits</a></li><li><a href="/learn/gf8-one-byte/">09 GF8: one byte</a></li></ol></li><li><h3>Module 4 · Ten to fourteen bits</h3><p>GF10, GF12 and GF14, and how the distance from 1 / phi moves as the word grows.</p><ol><li><a href="/learn/gf10-ten-bits/">10 GF10: ten bits</a></li><li><a href="/learn/gf12-twelve-bits/">11 GF12: twelve bits</a></li><li><a href="/learn/gf14-fourteen-bits/">12 GF14: fourteen bits</a></li></ol></li><li><h3>Module 5 · GF16 at work</h3><p>The primary 16-bit format, a two-term dot product in GF-T16, then GF20 and GF24.</p><ol><li><a href="/learn/gf16-the-primary/">13 GF16: the primary format</a></li><li><a href="/learn/two-products-one-sum/">14 Two products, one sum</a></li><li><a href="/learn/gf24-twenty-four-bits/">15 GF20 and GF24</a></li></ol></li><li><h3>Module 6 · GF32 to GF64</h3><p>GF32 beside IEEE single, GF48 with no IEEE twin, GF64 beside IEEE double.</p><ol><li><a href="/learn/gf32-a-single/">16 GF32: a single</a></li><li><a href="/learn/gf48-forty-eight-bits/">17 GF48: forty-eight bits</a></li><li><a href="/learn/gf64-a-double/">18 GF64: a double</a></li></ol></li><li><h3>Module 7 · GF96 to GF256</h3><p>GF96, GF128 and GF256, where the specs hold the layout with invariants.</p><ol><li><a href="/learn/gf96-ninety-six-bits/">19 GF96: ninety-six bits</a></li><li><a href="/learn/gf128-a-quad/">20 GF128: a quad</a></li><li><a href="/learn/gf256-two-five-six/">21 GF256: two hundred fifty-six bits</a></li></ol></li><li><h3>Module 8 · The widest rungs, then trits</h3><p>GF512 and GF1024, the two widest rungs, then GF-T8, where the exponent moves to trits.</p><ol><li><a href="/learn/gf512-five-twelve/">22 GF512: five hundred twelve bits</a></li><li><a href="/learn/gf1024-one-kilobit/">23 GF1024: one kilobit</a></li><li><a href="/learn/gft8-exponent-in-trits/">24 GF-T8: an exponent in trits</a></li></ol></li><li><h3>Module 9 · More trits, then the decode</h3><p>GF-T16 and GF-T32, then why fixed fields decode in parallel and a posit does not.</p><ol><li><a href="/learn/gft16-six-trits/">25 GF-T16: six trits</a></li><li><a href="/learn/gft32-twelve-trits/">26 GF-T32: twelve trits</a></li><li><a href="/learn/decode-in-one-step/">27 Decode in one step</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/why-binary-fails/index.html b/apps/website/public/learn/why-binary-fails/index.html
index b1d2780d8a..8af2b79f83 100644
--- a/apps/website/public/learn/why-binary-fails/index.html
+++ b/apps/website/public/learn/why-binary-fails/index.html
@@ -434,5 +434,6 @@ <h1>Why binary fails</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a clock is</h3><p>One edge, one world: what shares a clock edge shares a world, the period and the jitter of a real edge, and where the clock enters a board.</p><ol><li><a href="/learn/one-edge-one-world/">01 One edge, one world</a></li><li><a href="/learn/period-and-jitter/">02 A period, an edge, and jitter</a></li><li><a href="/learn/where-the-clock-enters/">03 Where the clock enters</a></li></ol></li><li><h3>Module 2 · Clock trees</h3><p>Skew and insertion delay, the global buffer network, and the trap of gating a clock with logic.</p><ol><li><a href="/learn/skew-and-insertion/">04 Skew and insertion delay</a></li><li><a href="/learn/global-buffers/">05 The global buffer</a></li><li><a href="/learn/gated-clock-trap/">06 The gated-clock trap</a></li></ol></li><li><h3>Module 3 · PLL and MMCM</h3><p>Multiply and divide one clock into another, move its phase in steps of the VCO, and which clocks the analyzer treats as related.</p><ol><li><a href="/learn/multiply-and-divide/">07 Multiply and divide</a></li><li><a href="/learn/phase-and-lock/">08 Phase and lock</a></li><li><a href="/learn/related-or-not/">09 Related or not</a></li></ol></li><li><h3>Module 4 · Resets</h3><p>Assert asynchronously, release synchronously: the three reset kinds, the release pipe, and the tree a reset grows.</p><ol><li><a href="/learn/sync-or-async/">10 Sync or async</a></li><li><a href="/learn/reset-release/">11 The release pipe</a></li><li><a href="/learn/reset-fanout/">12 Reset fanout</a></li></ol></li><li><h3>Module 5 · Metastability</h3><p>The setup-hold window, the mean time between failures in integer arithmetic, and the two flops that fix it.</p><ol><li><a href="/learn/setup-hold-window/">13 The setup-hold window</a></li><li><a href="/learn/mtbf/">14 Mean time between failures</a></li><li><a href="/learn/two-flop-sync/">15 The two-flop synchronizer</a></li></ol></li><li><h3>Module 6 · Crossing many bits</h3><p>Why a binary bus tears, why Gray code does not, and the handshake that moves a pulse between worlds.</p><ol><li class="on" aria-current="page"><a href="/learn/why-binary-fails/">16 Why binary fails</a></li><li><a href="/learn/gray-code/">17 Gray code</a></li><li><a href="/learn/handshakes-and-pulses/">18 Handshakes and pulses</a></li></ol></li><li><h3>Module 7 · The asynchronous FIFO</h3><p>Pointers, flags and depth: the buffer that moves a stream between two clocks.</p><ol><li><a href="/learn/pointers-and-flags/">19 Pointers and flags</a></li><li><a href="/learn/depth-sizing/">20 Sizing the depth</a></li><li><a href="/learn/fifo-testbench/">21 The FIFO testbench</a></li></ol></li><li><h3>Module 8 · Constraints</h3><p>The lines that tell the analyzer what a clock is, which paths not to check, and what the pins must meet.</p><ol><li><a href="/learn/create-clock/">22 create_clock</a></li><li><a href="/learn/false-path-and-max-delay/">23 False paths and max delay</a></li><li><a href="/learn/io-timing/">24 IO timing</a></li></ol></li><li><h3>Module 9 · On the board</h3><p>A CDC report, one crossing captured at the flip-flops, and the bitstream diff that closes the course.</p><ol><li><a href="/learn/cdc-report/">25 The CDC report</a></li><li><a href="/learn/capture-a-crossing/">26 Capture a crossing</a></li><li><a href="/learn/capstone-diff/">27 Capstone: change it and diff it</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/why-bridges/index.html b/apps/website/public/learn/why-bridges/index.html
index 8d84cd5db5..37d243d879 100644
--- a/apps/website/public/learn/why-bridges/index.html
+++ b/apps/website/public/learn/why-bridges/index.html
@@ -593,5 +593,6 @@ <h1>Why bridges</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a bus is</h3><p>Why a bus exists at all: a conversation on wires, framed and addressed, and who is allowed to talk.</p><ol><li><a href="/learn/wires-that-talk/">01 Wires that talk</a></li><li><a href="/learn/framing-a-conversation/">02 Framing a conversation</a></li><li><a href="/learn/master-and-servant/">03 Master and servant</a></li></ol></li><li><h3>Module 2 · UART</h3><p>The two-wire bus with no clock: the frame, the divisor that sets the rate, and the status a driver polls.</p><ol><li><a href="/learn/the-frame/">04 The frame: start, data, stop</a></li><li><a href="/learn/baud-and-the-divisor/">05 Baud and the divisor</a></li><li><a href="/learn/status-and-fifos/">06 Status and FIFOs</a></li></ol></li><li><h3>Module 3 · SPI</h3><p>The clocked conversation: four modes, a prescaler ladder, and a chip select per servant.</p><ol><li><a href="/learn/modes-cpol-cpha/">07 Modes: CPOL and CPHA</a></li><li><a href="/learn/the-prescaler-ladder/">08 The prescaler ladder</a></li><li><a href="/learn/chip-select-and-widths/">09 Chip select and widths</a></li></ol></li><li><h3>Module 4 · APB</h3><p>The register bus: PSEL and PENABLE, strobes and wait states, and how many address bits a peripheral count costs.</p><ol><li><a href="/learn/psel-and-penable/">10 PSEL and PENABLE</a></li><li><a href="/learn/strobes-and-waits/">11 Strobes and waits</a></li><li><a href="/learn/address-decode/">12 Address decode</a></li></ol></li><li><h3>Module 5 · AXI4</h3><p>The five channels: address, data and response in both directions, lite or full, bursts and IDs.</p><ol><li><a href="/learn/five-channels/">13 Five channels</a></li><li><a href="/learn/lite-or-full/">14 Lite or full</a></li><li><a href="/learn/bursts-and-ids/">15 Bursts and IDs</a></li></ol></li><li><h3>Module 6 · Memory</h3><p>What sits on the far side of every bus: memory maps, port kinds, and latency that a wait state must cover.</p><ol><li><a href="/learn/memory-maps/">16 Memory maps</a></li><li><a href="/learn/ports-and-kinds/">17 Ports and kinds</a></li><li><a href="/learn/latency/">18 Latency</a></li></ol></li><li><h3>Module 7 · Bridges</h3><p>Why designs grow more than one bus, and the packet bridge that moves work between them.</p><ol><li class="on" aria-current="page"><a href="/learn/why-bridges/">19 Why bridges</a></li><li><a href="/learn/the-packet-bridge/">20 The packet bridge</a></li><li><a href="/learn/everything-in-one-bench/">21 Everything in one bench</a></li></ol></li><li><h3>Module 8 · Ethernet</h3><p>Frames, the frame check sequence, RGMII timing, and the pre-registered steps of a real bring-up.</p><ol><li><a href="/learn/frames-and-the-fcs/">22 Frames and the FCS</a></li><li><a href="/learn/rgmii-timing/">23 RGMII timing</a></li><li><a href="/learn/bring-up-on-the-bench/">24 Bring-up on the bench</a></li></ol></li><li><h3>Module 9 · The bench</h3><p>The discipline that guards real hardware: who holds the IO, taking and giving the lock, and what runs next.</p><ol><li><a href="/learn/who-holds-the-io/">25 Who holds the IO</a></li><li><a href="/learn/claim-and-release/">26 Claim and release</a></li><li><a href="/learn/capstone-what-runs-next/">27 Capstone: what runs next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/why-the-split-is-phi/index.html b/apps/website/public/learn/why-the-split-is-phi/index.html
index ede096fe9c..9247044491 100644
--- a/apps/website/public/learn/why-the-split-is-phi/index.html
+++ b/apps/website/public/learn/why-the-split-is-phi/index.html
@@ -428,5 +428,6 @@ <h1>Why the split is phi</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · The rule and its numbers</h3><p>One rule splits every width, the ratio it aims at, and the Lucas numbers behind the 3.</p><ol><li><a href="/learn/a-float-cut-by-phi/">01 A float cut by phi</a></li><li><a href="/learn/phi-as-a-ratio/">02 Phi as a ratio</a></li><li><a href="/learn/lucas-numbers-stay-whole/">03 Lucas numbers stay whole</a></li></ol></li><li><h3>Module 2 · Why phi, why three</h3><p>Why the split is phi, why base three, and how a spec checks GF16 keeps phi.</p><ol><li class="on" aria-current="page"><a href="/learn/why-the-split-is-phi/">04 Why the split is phi</a></li><li><a href="/learn/why-base-three/">05 Why base three</a></li><li><a href="/learn/phi-in-sixteen-bits/">06 Phi in sixteen bits</a></li></ol></li><li><h3>Module 3 · The small rungs: GF4 to GF8</h3><p>GF4, GF6 and GF8, the fewest bits, where rounding to whole bits costs the most.</p><ol><li><a href="/learn/gf4-four-bits/">07 GF4: four bits</a></li><li><a href="/learn/gf6-six-bits/">08 GF6: six bits</a></li><li><a href="/learn/gf8-one-byte/">09 GF8: one byte</a></li></ol></li><li><h3>Module 4 · Ten to fourteen bits</h3><p>GF10, GF12 and GF14, and how the distance from 1 / phi moves as the word grows.</p><ol><li><a href="/learn/gf10-ten-bits/">10 GF10: ten bits</a></li><li><a href="/learn/gf12-twelve-bits/">11 GF12: twelve bits</a></li><li><a href="/learn/gf14-fourteen-bits/">12 GF14: fourteen bits</a></li></ol></li><li><h3>Module 5 · GF16 at work</h3><p>The primary 16-bit format, a two-term dot product in GF-T16, then GF20 and GF24.</p><ol><li><a href="/learn/gf16-the-primary/">13 GF16: the primary format</a></li><li><a href="/learn/two-products-one-sum/">14 Two products, one sum</a></li><li><a href="/learn/gf24-twenty-four-bits/">15 GF20 and GF24</a></li></ol></li><li><h3>Module 6 · GF32 to GF64</h3><p>GF32 beside IEEE single, GF48 with no IEEE twin, GF64 beside IEEE double.</p><ol><li><a href="/learn/gf32-a-single/">16 GF32: a single</a></li><li><a href="/learn/gf48-forty-eight-bits/">17 GF48: forty-eight bits</a></li><li><a href="/learn/gf64-a-double/">18 GF64: a double</a></li></ol></li><li><h3>Module 7 · GF96 to GF256</h3><p>GF96, GF128 and GF256, where the specs hold the layout with invariants.</p><ol><li><a href="/learn/gf96-ninety-six-bits/">19 GF96: ninety-six bits</a></li><li><a href="/learn/gf128-a-quad/">20 GF128: a quad</a></li><li><a href="/learn/gf256-two-five-six/">21 GF256: two hundred fifty-six bits</a></li></ol></li><li><h3>Module 8 · The widest rungs, then trits</h3><p>GF512 and GF1024, the two widest rungs, then GF-T8, where the exponent moves to trits.</p><ol><li><a href="/learn/gf512-five-twelve/">22 GF512: five hundred twelve bits</a></li><li><a href="/learn/gf1024-one-kilobit/">23 GF1024: one kilobit</a></li><li><a href="/learn/gft8-exponent-in-trits/">24 GF-T8: an exponent in trits</a></li></ol></li><li><h3>Module 9 · More trits, then the decode</h3><p>GF-T16 and GF-T32, then why fixed fields decode in parallel and a posit does not.</p><ol><li><a href="/learn/gft16-six-trits/">25 GF-T16: six trits</a></li><li><a href="/learn/gft32-twelve-trits/">26 GF-T32: twelve trits</a></li><li><a href="/learn/decode-in-one-step/">27 Decode in one step</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/why-three/index.html b/apps/website/public/learn/why-three/index.html
index d2b3060826..5cde9673e9 100644
--- a/apps/website/public/learn/why-three/index.html
+++ b/apps/website/public/learn/why-three/index.html
@@ -184,5 +184,6 @@ <h1>Why three</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Three values</h3><p>Why three, how balanced ternary writes every number without a sign, and what flipping and cutting trits do.</p><ol><li class="on" aria-current="page"><a href="/learn/why-three/">01 Why three</a></li><li><a href="/learn/balanced-ternary/">02 Balanced ternary</a></li><li><a href="/learn/flip-and-cut/">03 Flip to negate, cut to round</a></li></ol></li><li><h3>Module 2 · Logic with unknown</h3><p>Kleene's three-valued gates, two gates binary has no twin for, and addition as a pair of tables.</p><ol><li><a href="/learn/kleene-logic/">04 Three-valued logic</a></li><li><a href="/learn/consensus-and-any/">05 Consensus and any</a></li><li><a href="/learn/sum-and-carry/">06 Sum and carry</a></li></ol></li><li><h3>Module 3 · Adding trits</h3><p>The half adder, the full adder and a carry that ripples left, one place at a time.</p><ol><li><a href="/learn/half-adder/">07 The half adder</a></li><li><a href="/learn/full-adder/">08 The full adder</a></li><li><a href="/learn/ripple-carry/">09 Ripple carry</a></li></ol></li><li><h3>Module 4 · Multiplying without a multiplier</h3><p>Copy, drop or flip: a product by one trit, long multiplication, and a MAC that only adds.</p><ol><li><a href="/learn/times-a-trit/">10 Times one trit</a></li><li><a href="/learn/long-multiplication/">11 Long multiplication</a></li><li><a href="/learn/ternary-mac/">12 A ternary MAC</a></li></ol></li><li><h3>Module 5 · Trits in binary memory</h3><p>Two bits per trit, five trits per byte, and the bits a 27-trit word needs.</p><ol><li><a href="/learn/two-bits-per-trit/">13 Two bits per trit</a></li><li><a href="/learn/five-trits-a-byte/">14 Five trits in a byte</a></li><li><a href="/learn/a-word-of-27/">15 A word of 27 trits</a></li></ol></li><li><h3>Module 6 · The TRI-27 instruction word</h3><p>The 32-bit word the Trinity emulator decodes: its fields, its 47 opcodes and its 15-bit immediate.</p><ol><li><a href="/learn/the-instruction-word/">16 The instruction word</a></li><li><a href="/learn/the-opcode-map/">17 The opcode map</a></li><li><a href="/learn/the-immediate/">18 The immediate</a></li></ol></li><li><h3>Module 7 · A ternary machine</h3><p>An ALU built from this course's adders, a three-way jump, and a program you can step.</p><ol><li><a href="/learn/a-ternary-alu/">19 A ternary ALU</a></li><li><a href="/learn/three-way-jump/">20 A three-way jump</a></li><li><a href="/learn/run-a-program/">21 Run a program</a></li></ol></li><li><h3>Module 8 · Three answers</h3><p>Compare at the highest differing trit, find a number in thirds, sort with three-way compares.</p><ol><li><a href="/learn/compare-at-the-top/">22 Compare at the top</a></li><li><a href="/learn/questions-in-thirds/">23 Questions in thirds</a></li><li><a href="/learn/sort-by-three/">24 Sort with three-way compares</a></li></ol></li><li><h3>Module 9 · Ternary neurons</h3><p>Weights of -1, 0, +1: one neuron, a detector and a small layer, with no multiplier anywhere.</p><ol><li><a href="/learn/one-ternary-neuron/">25 One ternary neuron</a></li><li><a href="/learn/a-bar-detector/">26 A bar detector</a></li><li><a href="/learn/a-tiny-layer/">27 A tiny layer</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/wires-that-talk/index.html b/apps/website/public/learn/wires-that-talk/index.html
index d391098270..4d6bc4736a 100644
--- a/apps/website/public/learn/wires-that-talk/index.html
+++ b/apps/website/public/learn/wires-that-talk/index.html
@@ -299,5 +299,6 @@ <h1>Wires that talk</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · What a bus is</h3><p>Why a bus exists at all: a conversation on wires, framed and addressed, and who is allowed to talk.</p><ol><li class="on" aria-current="page"><a href="/learn/wires-that-talk/">01 Wires that talk</a></li><li><a href="/learn/framing-a-conversation/">02 Framing a conversation</a></li><li><a href="/learn/master-and-servant/">03 Master and servant</a></li></ol></li><li><h3>Module 2 · UART</h3><p>The two-wire bus with no clock: the frame, the divisor that sets the rate, and the status a driver polls.</p><ol><li><a href="/learn/the-frame/">04 The frame: start, data, stop</a></li><li><a href="/learn/baud-and-the-divisor/">05 Baud and the divisor</a></li><li><a href="/learn/status-and-fifos/">06 Status and FIFOs</a></li></ol></li><li><h3>Module 3 · SPI</h3><p>The clocked conversation: four modes, a prescaler ladder, and a chip select per servant.</p><ol><li><a href="/learn/modes-cpol-cpha/">07 Modes: CPOL and CPHA</a></li><li><a href="/learn/the-prescaler-ladder/">08 The prescaler ladder</a></li><li><a href="/learn/chip-select-and-widths/">09 Chip select and widths</a></li></ol></li><li><h3>Module 4 · APB</h3><p>The register bus: PSEL and PENABLE, strobes and wait states, and how many address bits a peripheral count costs.</p><ol><li><a href="/learn/psel-and-penable/">10 PSEL and PENABLE</a></li><li><a href="/learn/strobes-and-waits/">11 Strobes and waits</a></li><li><a href="/learn/address-decode/">12 Address decode</a></li></ol></li><li><h3>Module 5 · AXI4</h3><p>The five channels: address, data and response in both directions, lite or full, bursts and IDs.</p><ol><li><a href="/learn/five-channels/">13 Five channels</a></li><li><a href="/learn/lite-or-full/">14 Lite or full</a></li><li><a href="/learn/bursts-and-ids/">15 Bursts and IDs</a></li></ol></li><li><h3>Module 6 · Memory</h3><p>What sits on the far side of every bus: memory maps, port kinds, and latency that a wait state must cover.</p><ol><li><a href="/learn/memory-maps/">16 Memory maps</a></li><li><a href="/learn/ports-and-kinds/">17 Ports and kinds</a></li><li><a href="/learn/latency/">18 Latency</a></li></ol></li><li><h3>Module 7 · Bridges</h3><p>Why designs grow more than one bus, and the packet bridge that moves work between them.</p><ol><li><a href="/learn/why-bridges/">19 Why bridges</a></li><li><a href="/learn/the-packet-bridge/">20 The packet bridge</a></li><li><a href="/learn/everything-in-one-bench/">21 Everything in one bench</a></li></ol></li><li><h3>Module 8 · Ethernet</h3><p>Frames, the frame check sequence, RGMII timing, and the pre-registered steps of a real bring-up.</p><ol><li><a href="/learn/frames-and-the-fcs/">22 Frames and the FCS</a></li><li><a href="/learn/rgmii-timing/">23 RGMII timing</a></li><li><a href="/learn/bring-up-on-the-bench/">24 Bring-up on the bench</a></li></ol></li><li><h3>Module 9 · The bench</h3><p>The discipline that guards real hardware: who holds the IO, taking and giving the lock, and what runs next.</p><ol><li><a href="/learn/who-holds-the-io/">25 Who holds the IO</a></li><li><a href="/learn/claim-and-release/">26 Claim and release</a></li><li><a href="/learn/capstone-what-runs-next/">27 Capstone: what runs next</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/learn/xor-needs-a-bend/index.html b/apps/website/public/learn/xor-needs-a-bend/index.html
index 6203d60b7f..026133b96a 100644
--- a/apps/website/public/learn/xor-needs-a-bend/index.html
+++ b/apps/website/public/learn/xor-needs-a-bend/index.html
@@ -215,5 +215,6 @@ <h1>XOR needs a bend</h1>
 <section><h2>All lessons</h2><ul class="outline"><li><h3>Module 1 · Lab: our own research</h3><p>A number format of our own, an honest scoreboard, and a model's tables multiplied on the board.</p><ol><li><a href="/learn/t27-number-format/">01 T27: a number format of our own</a></li><li><a href="/learn/an-honest-scoreboard/">02 An honest scoreboard</a></li><li><a href="/learn/a-model-on-the-board/">03 A model's tables on the board</a></li></ol></li><li><h3>Module 2 · AI numbers: the MX block</h3><p>How AI chips keep weights in a few bits: one shared scale per block, the scale byte itself, and what one outlier does to its neighbours.</p><ol><li><a href="/learn/one-scale-per-block/">04 One scale for a block of weights</a></li><li><a href="/learn/scale-byte-native/">05 The scale byte on a real machine</a></li><li><a href="/learn/one-outlier/">06 One outlier, many zeros</a></li></ol></li><li><h3>Module 3 · Ternary weights</h3><p>Weights that are only minus, zero or plus a scale, the five rules a ternary alphabet must pass, and a test pass that checked nothing.</p><ol><li><a href="/learn/phi-scaled-ternary/">07 Weights of minus phi, zero and plus phi</a></li><li><a href="/learn/a-pass-that-checked-nothing/">08 A pass that checked nothing</a></li><li><a href="/learn/five-rules-for-weights/">09 Five rules for a weight alphabet</a></li></ol></li><li><h3>Module 4 · The Ternary Network Float</h3><p>A rule the compiler enforces before any test runs, and a 17-bit float whose exponent is four balanced trits.</p><ol><li><a href="/learn/blocked-is-not-failed/">10 Blocked is not failed</a></li><li><a href="/learn/four-trit-exponent/">11 A float whose exponent is four trits</a></li><li><a href="/learn/negate-twice/">12 Negate twice, get the number back</a></li></ol></li><li><h3>Module 5 · Arithmetic on signed numbers</h3><p>Multiply two signed numbers, add them when their signs differ, and do both at once in a multiply-accumulate.</p><ol><li><a href="/learn/a-sign-is-one-bit/">13 A sign is one bit</a></li><li><a href="/learn/opposite-signs-subtract/">14 Opposite signs subtract</a></li><li><a href="/learn/multiply-then-add/">15 Multiply, then add</a></li></ol></li><li><h3>Module 6 · Parts of a neuron</h3><p>A ReLU that bends at zero, a power of two for softmax, and an argmax that names the answer.</p><ol><li><a href="/learn/a-bend-at-zero/">16 A bend at zero</a></li><li><a href="/learn/two-to-the-x/">17 Two to the x</a></li><li><a href="/learn/pick-the-largest/">18 Pick the largest</a></li></ol></li><li><h3>Module 7 · Learning from a mistake</h3><p>A loss that prices a wrong guess in bits, one step that moves a weight against its gradient, and the hidden layer that XOR needs.</p><ol><li><a href="/learn/a-loss-in-bits/">19 A loss in bits</a></li><li><a href="/learn/one-step-downhill/">20 One step downhill</a></li><li class="on" aria-current="page"><a href="/learn/xor-needs-a-bend/">21 XOR needs a bend</a></li></ol></li><li><h3>Module 8 · BitNet: ternary networks</h3><p>A threshold that squeezes a sum back to three values, one neuron that becomes a different function when its weights change, and a neuron that reads its inputs 27 trits at a time.</p><ol><li><a href="/learn/back-to-three-values/">22 Back to three values</a></li><li><a href="/learn/same-neuron-new-weights/">23 Same neuron, new weights</a></li><li><a href="/learn/a-neuron-in-chunks/">24 A neuron in chunks</a></li></ol></li><li><h3>Module 9 · The ternary MAC as a chip</h3><p>The 27-trit dot product as wires with no register, the same sum added into a register on every clock, and a small whole network to close the course.</p><ol><li><a href="/learn/a-dot-product-in-wires/">25 A dot product made of wires</a></li><li><a href="/learn/add-it-up-every-clock/">26 Add it up on every clock</a></li><li><a href="/learn/a-whole-network/">27 A whole network</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">This course is itself a t27 spec: read it</a></p><p>Widgets keep their own English words: each one lives in its own t27 spec.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/a-bar-detector/index.html b/apps/website/public/ru/learn/a-bar-detector/index.html
index 0fe9846a1e..a966f16fdb 100644
--- a/apps/website/public/ru/learn/a-bar-detector/index.html
+++ b/apps/website/public/ru/learn/a-bar-detector/index.html
@@ -123,5 +123,6 @@ <h1>Детектор полосы</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Три значения</h3><p>Почему три, как сбалансированная троичная система пишет любое число без знака и что дают переворот и отсечение тритов.</p><ol><li><a href="/ru/learn/why-three/">01 Почему три</a></li><li><a href="/ru/learn/balanced-ternary/">02 Сбалансированная троичная</a></li><li><a href="/ru/learn/flip-and-cut/">03 Переворот вместо минуса, отсечение вместо округления</a></li></ol></li><li><h3>Модуль 2 · Логика с «не знаю»</h3><p>Трёхзначные вентили Клини, два вентиля без двоичного двойника и сложение как пара таблиц.</p><ol><li><a href="/ru/learn/kleene-logic/">04 Трёхзначная логика</a></li><li><a href="/ru/learn/consensus-and-any/">05 Консенсус и «любой»</a></li><li><a href="/ru/learn/sum-and-carry/">06 Сумма и перенос</a></li></ol></li><li><h3>Модуль 3 · Складываем триты</h3><p>Полусумматор, полный сумматор и перенос, бегущий влево разряд за разрядом.</p><ol><li><a href="/ru/learn/half-adder/">07 Полусумматор</a></li><li><a href="/ru/learn/full-adder/">08 Полный сумматор</a></li><li><a href="/ru/learn/ripple-carry/">09 Сквозной перенос</a></li></ol></li><li><h3>Модуль 4 · Умножение без умножителя</h3><p>Копия, ноль или переворот: произведение на один трит, умножение столбиком и MAC, который только складывает.</p><ol><li><a href="/ru/learn/times-a-trit/">10 Умножить на один трит</a></li><li><a href="/ru/learn/long-multiplication/">11 Умножение столбиком</a></li><li><a href="/ru/learn/ternary-mac/">12 Троичный MAC</a></li></ol></li><li><h3>Модуль 5 · Триты в двоичной памяти</h3><p>Два бита на трит, пять тритов на байт и сколько бит нужно слову из 27 тритов.</p><ol><li><a href="/ru/learn/two-bits-per-trit/">13 Два бита на трит</a></li><li><a href="/ru/learn/five-trits-a-byte/">14 Пять тритов в байте</a></li><li><a href="/ru/learn/a-word-of-27/">15 Слово из 27 тритов</a></li></ol></li><li><h3>Модуль 6 · Командное слово TRI-27</h3><p>32-битное слово, которое декодирует эмулятор Trinity: его поля, 47 опкодов и 15-битное непосредственное значение.</p><ol><li><a href="/ru/learn/the-instruction-word/">16 Командное слово</a></li><li><a href="/ru/learn/the-opcode-map/">17 Карта опкодов</a></li><li><a href="/ru/learn/the-immediate/">18 Непосредственное значение</a></li></ol></li><li><h3>Модуль 7 · Троичная машина</h3><p>АЛУ из сумматоров этого курса, переход на три стороны и программа, которую можно пройти по шагам.</p><ol><li><a href="/ru/learn/a-ternary-alu/">19 Троичное АЛУ</a></li><li><a href="/ru/learn/three-way-jump/">20 Переход на три стороны</a></li><li><a href="/ru/learn/run-a-program/">21 Запустить программу</a></li></ol></li><li><h3>Модуль 8 · Три ответа</h3><p>Сравнение по старшему различному триту, поиск числа по третям и сортировка трёхзначными сравнениями.</p><ol><li><a href="/ru/learn/compare-at-the-top/">22 Сравнение сверху</a></li><li><a href="/ru/learn/questions-in-thirds/">23 Вопросы по третям</a></li><li><a href="/ru/learn/sort-by-three/">24 Сортировка трёхзначными сравнениями</a></li></ol></li><li><h3>Модуль 9 · Троичные нейроны</h3><p>Веса -1, 0, +1: один нейрон, детектор и маленький слой, и нигде нет умножителя.</p><ol><li><a href="/ru/learn/one-ternary-neuron/">25 Один троичный нейрон</a></li><li class="on" aria-current="page"><a href="/ru/learn/a-bar-detector/">26 Детектор полосы</a></li><li><a href="/ru/learn/a-tiny-layer/">27 Маленький слой</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/a-bend-at-zero/index.html b/apps/website/public/ru/learn/a-bend-at-zero/index.html
index b3bd55e25a..79e1d1c27e 100644
--- a/apps/website/public/ru/learn/a-bend-at-zero/index.html
+++ b/apps/website/public/ru/learn/a-bend-at-zero/index.html
@@ -94,5 +94,6 @@ <h1>Излом в нуле</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Лаборатория: наши исследования</h3><p>Свой формат чисел, честная таблица результатов и таблицы модели, перемноженные на плате.</p><ol><li><a href="/ru/learn/t27-number-format/">01 T27: наш собственный формат чисел</a></li><li><a href="/ru/learn/an-honest-scoreboard/">02 Честная таблица результатов</a></li><li><a href="/ru/learn/a-model-on-the-board/">03 Таблицы модели на плате</a></li></ol></li><li><h3>Модуль 2 · ИИ-числа: блок MX</h3><p>Как ИИ-чипы хранят веса в нескольких битах: один общий масштаб на блок, сам байт масштаба и что один выброс делает с соседями.</p><ol><li><a href="/ru/learn/one-scale-per-block/">04 Один масштаб на блок весов</a></li><li><a href="/ru/learn/scale-byte-native/">05 Байт масштаба на настоящей машине</a></li><li><a href="/ru/learn/one-outlier/">06 Один выброс, много нулей</a></li></ol></li><li><h3>Модуль 3 · Тернарные веса</h3><p>Веса, которые бывают только минус масштаб, ноль или плюс масштаб, пять правил, которые должен пройти тернарный алфавит, и прогон тестов, который ничего не проверил.</p><ol><li><a href="/ru/learn/phi-scaled-ternary/">07 Веса минус фи, ноль и плюс фи</a></li><li><a href="/ru/learn/a-pass-that-checked-nothing/">08 Прогон, который ничего не проверил</a></li><li><a href="/ru/learn/five-rules-for-weights/">09 Пять правил для алфавита весов</a></li></ol></li><li><h3>Модуль 4 · Тернарный формат Ternary Network Float</h3><p>Правило, которое компилятор проверяет до запуска любого теста, и 17-битное число с плавающей точкой, порядок которого — четыре сбалансированных трита.</p><ol><li><a href="/ru/learn/blocked-is-not-failed/">10 Заблокирован — не значит провален</a></li><li><a href="/ru/learn/four-trit-exponent/">11 Число, у которого порядок — четыре трита</a></li><li><a href="/ru/learn/negate-twice/">12 Дважды сменить знак — получить число обратно</a></li></ol></li><li><h3>Модуль 5 · Арифметика чисел со знаком</h3><p>Умножить два числа со знаком, сложить их, когда знаки разные, и сделать то и другое сразу в умножении с накоплением.</p><ol><li><a href="/ru/learn/a-sign-is-one-bit/">13 Знак — это один бит</a></li><li><a href="/ru/learn/opposite-signs-subtract/">14 Разные знаки — вычитание</a></li><li><a href="/ru/learn/multiply-then-add/">15 Умножить, потом сложить</a></li></ol></li><li><h3>Модуль 6 · Части нейрона</h3><p>ReLU с изломом в нуле, степень двойки для softmax и argmax, который называет ответ.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/a-bend-at-zero/">16 Излом в нуле</a></li><li><a href="/ru/learn/two-to-the-x/">17 Два в степени x</a></li><li><a href="/ru/learn/pick-the-largest/">18 Выбрать наибольшее</a></li></ol></li><li><h3>Модуль 7 · Обучение на ошибке</h3><p>Потеря, которая оценивает неверную догадку в битах, один шаг, который сдвигает вес против градиента, и скрытый слой, который нужен XOR.</p><ol><li><a href="/ru/learn/a-loss-in-bits/">19 Потеря в битах</a></li><li><a href="/ru/learn/one-step-downhill/">20 Один шаг вниз по склону</a></li><li><a href="/ru/learn/xor-needs-a-bend/">21 XOR нужен излом</a></li></ol></li><li><h3>Модуль 8 · BitNet: тернарные сети</h3><p>Порог, который сжимает сумму обратно до трёх значений, один нейрон, который становится другой функцией при смене весов, и нейрон, который читает входы по 27 тритов за раз.</p><ol><li><a href="/ru/learn/back-to-three-values/">22 Обратно в три значения</a></li><li><a href="/ru/learn/same-neuron-new-weights/">23 Тот же нейрон, другие веса</a></li><li><a href="/ru/learn/a-neuron-in-chunks/">24 Нейрон по кускам</a></li></ol></li><li><h3>Модуль 9 · Тернарный MAC как чип</h3><p>Скалярное произведение 27 тритов как провода без регистра, та же сумма, которую регистр копит на каждом такте, и небольшая целая сеть в конце курса.</p><ol><li><a href="/ru/learn/a-dot-product-in-wires/">25 Скалярное произведение из проводов</a></li><li><a href="/ru/learn/add-it-up-every-clock/">26 Складывать на каждом такте</a></li><li><a href="/ru/learn/a-whole-network/">27 Целая сеть</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/a-build-receipt/index.html b/apps/website/public/ru/learn/a-build-receipt/index.html
index 7d34d73b3e..b007789823 100644
--- a/apps/website/public/ru/learn/a-build-receipt/index.html
+++ b/apps/website/public/ru/learn/a-build-receipt/index.html
@@ -170,5 +170,6 @@ <h1>Сборка по пунктам</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Чип</h3><p>Что такое FPGA, какой чип у нас и как его выводы встречаются с платой.</p><ol><li><a href="/ru/learn/fpga-is-a-grid/">01 FPGA — это сетка, которую вы настраиваете</a></li><li><a href="/ru/learn/which-chip/">02 Какой это чип?</a></li><li><a href="/ru/learn/pins-and-package/">03 Выводы, шарики и корпус</a></li></ol></li><li><h3>Модуль 2 · Числа в железе</h3><p>Биты, триты и форматы чисел, и сколько стоит арифметика в логике.</p><ol><li><a href="/ru/learn/bits-and-trits/">04 Биты, триты и GF16</a></li><li><a href="/ru/learn/what-arithmetic-costs/">05 Сколько стоит арифметика</a></li><li><a href="/ru/learn/how-much-fits/">06 Сколько помещается на чип</a></li></ol></li><li><h3>Модуль 3 · Ваша программа на t27</h3><p>Напишите спеку, проверьте её и увидьте, почему «компилируется» не значит «верно».</p><ol><li><a href="/ru/learn/hello-t27/">07 Привет, t27</a></li><li><a href="/ru/learn/tests-are-the-spec/">08 Тесты — часть спеки</a></li><li><a href="/ru/learn/compiles-is-not-correct/">09 Компилируется — не значит верно</a></li></ol></li><li><h3>Модуль 4 · Внутри t27c</h3><p>Как компилятор читает спеку и что он пишет, включая нативный t27b.</p><ol><li><a href="/ru/learn/compiler-stages/">10 Стадии t27c</a></li><li><a href="/ru/learn/t27c-on-a-real-machine/">11 t27c на настоящей машине</a></li><li><a href="/ru/learn/seven-backends-and-t27b/">12 Семь бэкендов и t27b</a></li></ol></li><li><h3>Модуль 5 · От спеки к железу</h3><p>Verilog, который пишет t27c, ячейки, в которые он превращается, и что делает одна LUT.</p><ol><li><a href="/ru/learn/spec-to-verilog/">13 От спеки к Verilog</a></li><li><a href="/ru/learn/count-the-cells/">14 Посчитайте ячейки</a></li><li><a href="/ru/learn/one-lut-up-close/">15 Одна LUT вблизи</a></li></ol></li><li><h3>Модуль 6 · Читаем синтез</h3><p>Что yosys сообщает о вашем дизайне и какие предупреждения важны.</p><ol><li><a href="/ru/learn/synthesis-warnings/">16 Важные предупреждения синтеза</a></li><li><a href="/ru/learn/state-machines/">17 Конечные автоматы</a></li><li><a href="/ru/learn/where-the-luts-go/">18 Куда уходят LUT</a></li></ol></li><li><h3>Модуль 7 · Размещение, трассировка, тайминг</h3><p>Где ячейки оказываются на кристалле и успевает ли тактовая частота.</p><ol><li><a href="/ru/learn/placement/">19 Размещение: где оказываются ячейки</a></li><li><a href="/ru/learn/logic-depth/">20 Глубина логики</a></li><li><a href="/ru/learn/timing-slack/">21 Тайминг и запас</a></li></ol></li><li><h3>Модуль 8 · Битстрим</h3><p>Как оттрассированный дизайн становится битами, которые загружает чип, без Vivado.</p><ol><li><a href="/ru/learn/fasm-to-frames/">22 От FASM к фреймам</a></li><li class="on" aria-current="page"><a href="/ru/learn/a-build-receipt/">23 Сборка по пунктам</a></li><li><a href="/ru/learn/read-a-bitstream/">24 Читаем битстрим</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Спросите чип, кто он, загрузите биты и проверьте их.</p><ol><li><a href="/ru/learn/jtag-asks-the-chip/">25 JTAG спрашивает чип</a></li><li><a href="/ru/learn/same-bits-on-the-board/">26 Те же биты — на плате</a></li><li><a href="/ru/learn/capstone/">27 Итог: переугадайте yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/a-correlator-finds-the-frame/index.html b/apps/website/public/ru/learn/a-correlator-finds-the-frame/index.html
index 04027feab0..cfbda55e5d 100644
--- a/apps/website/public/ru/learn/a-correlator-finds-the-frame/index.html
+++ b/apps/website/public/ru/learn/a-correlator-finds-the-frame/index.html
@@ -284,5 +284,6 @@ <h1>Коррелятор находит кадр</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Сложение</h3><p>Почему сложение медленное: перенос, который идёт вверх по слову, выделенная цепь в каждом слайсе, префиксные сети и деревья сохранения переноса.</p><ol><li><a href="/ru/learn/the-carry-chain/">01 Цепь переноса</a></li><li><a href="/ru/learn/prefix-adders/">02 Взгляд вперёд: префиксные сумматоры</a></li><li><a href="/ru/learn/carry-save/">03 Три числа сразу</a></li></ol></li><li><h3>Модуль 2 · Умножение</h3><p>Произведение — сумма сдвинутых копий; перекодирование Бута вдвое сокращает их, а слайс DSP48E1 делает остальное одним блоком.</p><ol><li><a href="/ru/learn/shift-and-add/">04 Сдвиг и сложение</a></li><li><a href="/ru/learn/booths-trick/">05 Приём Бута</a></li><li><a href="/ru/learn/the-dsp48e1-slice/">06 Слайс DSP48E1</a></li></ol></li><li><h3>Модуль 3 · Фиксированная точка</h3><p>Где стоит двоичная точка, что округление делает со значением и с его средним и что даёт каждый бит квантователя.</p><ol><li><a href="/ru/learn/q-formats/">07 Форматы Q</a></li><li><a href="/ru/learn/rounding-and-saturation/">08 Округление и насыщение</a></li><li><a href="/ru/learn/quantization-noise/">09 Шум квантования</a></li></ol></li><li><h3>Модуль 4 · Функции в железе</h3><p>Синус, косинус, угол и длина из сдвигов и сложений — и когда таблица оказывается выгоднее.</p><ol><li><a href="/ru/learn/rotating-by-shifts/">10 Поворот сдвигами</a></li><li><a href="/ru/learn/angle-and-magnitude/">11 Угол и модуль</a></li><li><a href="/ru/learn/tables-and-interpolation/">12 Таблицы и интерполяция</a></li></ol></li><li><h3>Модуль 5 · Сигналы и дискретизация</h3><p>Что дискретизация делает с частотой, как генератор строится из сумматора и что измеряет бин ДПФ.</p><ol><li><a href="/ru/learn/sampling-and-aliasing/">13 Дискретизация и наложение спектров</a></li><li><a href="/ru/learn/the-nco/">14 Генератор с числовым управлением</a></li><li><a href="/ru/learn/bins-of-a-spectrum/">15 Бины спектра</a></li></ol></li><li><h3>Модуль 6 · Фильтры FIR</h3><p>Скользящее среднее, расчёт окном sinc с целыми коэффициентами и раскладка отводов по слайсам DSP.</p><ol><li><a href="/ru/learn/the-moving-average/">16 Скользящее среднее</a></li><li><a href="/ru/learn/designing-taps/">17 Расчёт коэффициентов</a></li><li><a href="/ru/learn/folding-taps-onto-slices/">18 Раскладка отводов по слайсам</a></li></ol></li><li><h3>Модуль 7 · Многоскоростная обработка</h3><p>Понижение частоты дискретизации без заворота шума: децимация, фильтр CIC и полифазная форма.</p><ol><li><a href="/ru/learn/lowering-the-rate/">19 Понижение частоты</a></li><li><a href="/ru/learn/the-cic-filter/">20 Фильтр CIC</a></li><li><a href="/ru/learn/polyphase/">21 Полифазная форма</a></li></ol></li><li><h3>Модуль 8 · БПФ</h3><p>N log N вместо N^2: бабочки, бит-реверсный порядок входа и биты, которые добавляет каждый этап.</p><ol><li><a href="/ru/learn/the-butterfly/">22 Бабочка</a></li><li><a href="/ru/learn/twiddles-and-bit-reversal/">23 Поворачивающие множители и бит-реверс</a></li><li><a href="/ru/learn/growth-through-the-stages/">24 Рост через этапы</a></li></ol></li><li><h3>Модуль 9 · На стенде</h3><p>От арифметики к плате: коррелятор из спеки модема, бюджет слайсов и тайминга и итоговый фильтр.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/a-correlator-finds-the-frame/">25 Коррелятор находит кадр</a></li><li><a href="/ru/learn/does-it-fit-and-close-timing/">26 Помещается ли и сходится ли тайминг?</a></li><li><a href="/ru/learn/capstone-a-filter-on-the-board/">27 Итог: фильтр на плате</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/a-dot-product-in-wires/index.html b/apps/website/public/ru/learn/a-dot-product-in-wires/index.html
index b0c495501f..3cb990f455 100644
--- a/apps/website/public/ru/learn/a-dot-product-in-wires/index.html
+++ b/apps/website/public/ru/learn/a-dot-product-in-wires/index.html
@@ -113,5 +113,6 @@ <h1>Скалярное произведение из проводов</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Лаборатория: наши исследования</h3><p>Свой формат чисел, честная таблица результатов и таблицы модели, перемноженные на плате.</p><ol><li><a href="/ru/learn/t27-number-format/">01 T27: наш собственный формат чисел</a></li><li><a href="/ru/learn/an-honest-scoreboard/">02 Честная таблица результатов</a></li><li><a href="/ru/learn/a-model-on-the-board/">03 Таблицы модели на плате</a></li></ol></li><li><h3>Модуль 2 · ИИ-числа: блок MX</h3><p>Как ИИ-чипы хранят веса в нескольких битах: один общий масштаб на блок, сам байт масштаба и что один выброс делает с соседями.</p><ol><li><a href="/ru/learn/one-scale-per-block/">04 Один масштаб на блок весов</a></li><li><a href="/ru/learn/scale-byte-native/">05 Байт масштаба на настоящей машине</a></li><li><a href="/ru/learn/one-outlier/">06 Один выброс, много нулей</a></li></ol></li><li><h3>Модуль 3 · Тернарные веса</h3><p>Веса, которые бывают только минус масштаб, ноль или плюс масштаб, пять правил, которые должен пройти тернарный алфавит, и прогон тестов, который ничего не проверил.</p><ol><li><a href="/ru/learn/phi-scaled-ternary/">07 Веса минус фи, ноль и плюс фи</a></li><li><a href="/ru/learn/a-pass-that-checked-nothing/">08 Прогон, который ничего не проверил</a></li><li><a href="/ru/learn/five-rules-for-weights/">09 Пять правил для алфавита весов</a></li></ol></li><li><h3>Модуль 4 · Тернарный формат Ternary Network Float</h3><p>Правило, которое компилятор проверяет до запуска любого теста, и 17-битное число с плавающей точкой, порядок которого — четыре сбалансированных трита.</p><ol><li><a href="/ru/learn/blocked-is-not-failed/">10 Заблокирован — не значит провален</a></li><li><a href="/ru/learn/four-trit-exponent/">11 Число, у которого порядок — четыре трита</a></li><li><a href="/ru/learn/negate-twice/">12 Дважды сменить знак — получить число обратно</a></li></ol></li><li><h3>Модуль 5 · Арифметика чисел со знаком</h3><p>Умножить два числа со знаком, сложить их, когда знаки разные, и сделать то и другое сразу в умножении с накоплением.</p><ol><li><a href="/ru/learn/a-sign-is-one-bit/">13 Знак — это один бит</a></li><li><a href="/ru/learn/opposite-signs-subtract/">14 Разные знаки — вычитание</a></li><li><a href="/ru/learn/multiply-then-add/">15 Умножить, потом сложить</a></li></ol></li><li><h3>Модуль 6 · Части нейрона</h3><p>ReLU с изломом в нуле, степень двойки для softmax и argmax, который называет ответ.</p><ol><li><a href="/ru/learn/a-bend-at-zero/">16 Излом в нуле</a></li><li><a href="/ru/learn/two-to-the-x/">17 Два в степени x</a></li><li><a href="/ru/learn/pick-the-largest/">18 Выбрать наибольшее</a></li></ol></li><li><h3>Модуль 7 · Обучение на ошибке</h3><p>Потеря, которая оценивает неверную догадку в битах, один шаг, который сдвигает вес против градиента, и скрытый слой, который нужен XOR.</p><ol><li><a href="/ru/learn/a-loss-in-bits/">19 Потеря в битах</a></li><li><a href="/ru/learn/one-step-downhill/">20 Один шаг вниз по склону</a></li><li><a href="/ru/learn/xor-needs-a-bend/">21 XOR нужен излом</a></li></ol></li><li><h3>Модуль 8 · BitNet: тернарные сети</h3><p>Порог, который сжимает сумму обратно до трёх значений, один нейрон, который становится другой функцией при смене весов, и нейрон, который читает входы по 27 тритов за раз.</p><ol><li><a href="/ru/learn/back-to-three-values/">22 Обратно в три значения</a></li><li><a href="/ru/learn/same-neuron-new-weights/">23 Тот же нейрон, другие веса</a></li><li><a href="/ru/learn/a-neuron-in-chunks/">24 Нейрон по кускам</a></li></ol></li><li><h3>Модуль 9 · Тернарный MAC как чип</h3><p>Скалярное произведение 27 тритов как провода без регистра, та же сумма, которую регистр копит на каждом такте, и небольшая целая сеть в конце курса.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/a-dot-product-in-wires/">25 Скалярное произведение из проводов</a></li><li><a href="/ru/learn/add-it-up-every-clock/">26 Складывать на каждом такте</a></li><li><a href="/ru/learn/a-whole-network/">27 Целая сеть</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/a-float-cut-by-phi/index.html b/apps/website/public/ru/learn/a-float-cut-by-phi/index.html
index 231f230b29..2ca8198082 100644
--- a/apps/website/public/ru/learn/a-float-cut-by-phi/index.html
+++ b/apps/website/public/ru/learn/a-float-cut-by-phi/index.html
@@ -629,5 +629,6 @@ <h1>Число с плавающей точкой, разрезанное по p
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Правило и его числа</h3><p>Одно правило делит каждую ширину, отношение, к которому оно стремится, и числа Люка за тройкой 3.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/a-float-cut-by-phi/">01 Число с плавающей точкой, разрезанное по phi</a></li><li><a href="/ru/learn/phi-as-a-ratio/">02 Phi как отношение</a></li><li><a href="/ru/learn/lucas-numbers-stay-whole/">03 Числа Люка остаются целыми</a></li></ol></li><li><h3>Модуль 2 · Почему phi, почему три</h3><p>Почему деление идёт по phi, почему основание три и как спека проверяет, что GF16 хранит phi.</p><ol><li><a href="/ru/learn/why-the-split-is-phi/">04 Почему деление по phi</a></li><li><a href="/ru/learn/why-base-three/">05 Почему основание три</a></li><li><a href="/ru/learn/phi-in-sixteen-bits/">06 Phi в шестнадцати битах</a></li></ol></li><li><h3>Модуль 3 · Малые ступени: от GF4 до GF8</h3><p>GF4, GF6 и GF8 — меньше всего битов, и округление до целых битов стоит здесь дороже всего.</p><ol><li><a href="/ru/learn/gf4-four-bits/">07 GF4: четыре бита</a></li><li><a href="/ru/learn/gf6-six-bits/">08 GF6: шесть битов</a></li><li><a href="/ru/learn/gf8-one-byte/">09 GF8: один байт</a></li></ol></li><li><h3>Модуль 4 · От десяти до четырнадцати битов</h3><p>GF10, GF12 и GF14 и то, как расстояние до 1 / phi меняется с ростом слова.</p><ol><li><a href="/ru/learn/gf10-ten-bits/">10 GF10: десять битов</a></li><li><a href="/ru/learn/gf12-twelve-bits/">11 GF12: двенадцать битов</a></li><li><a href="/ru/learn/gf14-fourteen-bits/">12 GF14: четырнадцать битов</a></li></ol></li><li><h3>Модуль 5 · GF16 в работе</h3><p>Основной 16-битный формат, скалярное произведение из двух слагаемых в GF-T16, затем GF20 и GF24.</p><ol><li><a href="/ru/learn/gf16-the-primary/">13 GF16: основной формат</a></li><li><a href="/ru/learn/two-products-one-sum/">14 Два произведения, одна сумма</a></li><li><a href="/ru/learn/gf24-twenty-four-bits/">15 GF20 и GF24</a></li></ol></li><li><h3>Модуль 6 · От GF32 до GF64</h3><p>GF32 рядом с IEEE single, GF48 без пары в IEEE, GF64 рядом с IEEE double.</p><ol><li><a href="/ru/learn/gf32-a-single/">16 GF32: одинарная точность</a></li><li><a href="/ru/learn/gf48-forty-eight-bits/">17 GF48: сорок восемь битов</a></li><li><a href="/ru/learn/gf64-a-double/">18 GF64: двойная точность</a></li></ol></li><li><h3>Модуль 7 · От GF96 до GF256</h3><p>GF96, GF128 и GF256, где спеки держат раскладку инвариантами.</p><ol><li><a href="/ru/learn/gf96-ninety-six-bits/">19 GF96: девяносто шесть битов</a></li><li><a href="/ru/learn/gf128-a-quad/">20 GF128: четверная точность</a></li><li><a href="/ru/learn/gf256-two-five-six/">21 GF256: двести пятьдесят шесть битов</a></li></ol></li><li><h3>Модуль 8 · Самые широкие ступени, затем триты</h3><p>GF512 и GF1024, две самые широкие ступени, затем GF-T8, где порядок уходит в триты.</p><ol><li><a href="/ru/learn/gf512-five-twelve/">22 GF512: пятьсот двенадцать битов</a></li><li><a href="/ru/learn/gf1024-one-kilobit/">23 GF1024: один килобит</a></li><li><a href="/ru/learn/gft8-exponent-in-trits/">24 GF-T8: порядок в тритах</a></li></ol></li><li><h3>Модуль 9 · Ещё триты, затем декодирование</h3><p>GF-T16 и GF-T32, затем почему фиксированные поля декодируются параллельно, а posit — нет.</p><ol><li><a href="/ru/learn/gft16-six-trits/">25 GF-T16: шесть тритов</a></li><li><a href="/ru/learn/gft32-twelve-trits/">26 GF-T32: двенадцать тритов</a></li><li><a href="/ru/learn/decode-in-one-step/">27 Раскодировать за один шаг</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/a-loss-in-bits/index.html b/apps/website/public/ru/learn/a-loss-in-bits/index.html
index eaef1d4bf0..f2b5b94cd0 100644
--- a/apps/website/public/ru/learn/a-loss-in-bits/index.html
+++ b/apps/website/public/ru/learn/a-loss-in-bits/index.html
@@ -172,5 +172,6 @@ <h1>Потеря в битах</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Лаборатория: наши исследования</h3><p>Свой формат чисел, честная таблица результатов и таблицы модели, перемноженные на плате.</p><ol><li><a href="/ru/learn/t27-number-format/">01 T27: наш собственный формат чисел</a></li><li><a href="/ru/learn/an-honest-scoreboard/">02 Честная таблица результатов</a></li><li><a href="/ru/learn/a-model-on-the-board/">03 Таблицы модели на плате</a></li></ol></li><li><h3>Модуль 2 · ИИ-числа: блок MX</h3><p>Как ИИ-чипы хранят веса в нескольких битах: один общий масштаб на блок, сам байт масштаба и что один выброс делает с соседями.</p><ol><li><a href="/ru/learn/one-scale-per-block/">04 Один масштаб на блок весов</a></li><li><a href="/ru/learn/scale-byte-native/">05 Байт масштаба на настоящей машине</a></li><li><a href="/ru/learn/one-outlier/">06 Один выброс, много нулей</a></li></ol></li><li><h3>Модуль 3 · Тернарные веса</h3><p>Веса, которые бывают только минус масштаб, ноль или плюс масштаб, пять правил, которые должен пройти тернарный алфавит, и прогон тестов, который ничего не проверил.</p><ol><li><a href="/ru/learn/phi-scaled-ternary/">07 Веса минус фи, ноль и плюс фи</a></li><li><a href="/ru/learn/a-pass-that-checked-nothing/">08 Прогон, который ничего не проверил</a></li><li><a href="/ru/learn/five-rules-for-weights/">09 Пять правил для алфавита весов</a></li></ol></li><li><h3>Модуль 4 · Тернарный формат Ternary Network Float</h3><p>Правило, которое компилятор проверяет до запуска любого теста, и 17-битное число с плавающей точкой, порядок которого — четыре сбалансированных трита.</p><ol><li><a href="/ru/learn/blocked-is-not-failed/">10 Заблокирован — не значит провален</a></li><li><a href="/ru/learn/four-trit-exponent/">11 Число, у которого порядок — четыре трита</a></li><li><a href="/ru/learn/negate-twice/">12 Дважды сменить знак — получить число обратно</a></li></ol></li><li><h3>Модуль 5 · Арифметика чисел со знаком</h3><p>Умножить два числа со знаком, сложить их, когда знаки разные, и сделать то и другое сразу в умножении с накоплением.</p><ol><li><a href="/ru/learn/a-sign-is-one-bit/">13 Знак — это один бит</a></li><li><a href="/ru/learn/opposite-signs-subtract/">14 Разные знаки — вычитание</a></li><li><a href="/ru/learn/multiply-then-add/">15 Умножить, потом сложить</a></li></ol></li><li><h3>Модуль 6 · Части нейрона</h3><p>ReLU с изломом в нуле, степень двойки для softmax и argmax, который называет ответ.</p><ol><li><a href="/ru/learn/a-bend-at-zero/">16 Излом в нуле</a></li><li><a href="/ru/learn/two-to-the-x/">17 Два в степени x</a></li><li><a href="/ru/learn/pick-the-largest/">18 Выбрать наибольшее</a></li></ol></li><li><h3>Модуль 7 · Обучение на ошибке</h3><p>Потеря, которая оценивает неверную догадку в битах, один шаг, который сдвигает вес против градиента, и скрытый слой, который нужен XOR.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/a-loss-in-bits/">19 Потеря в битах</a></li><li><a href="/ru/learn/one-step-downhill/">20 Один шаг вниз по склону</a></li><li><a href="/ru/learn/xor-needs-a-bend/">21 XOR нужен излом</a></li></ol></li><li><h3>Модуль 8 · BitNet: тернарные сети</h3><p>Порог, который сжимает сумму обратно до трёх значений, один нейрон, который становится другой функцией при смене весов, и нейрон, который читает входы по 27 тритов за раз.</p><ol><li><a href="/ru/learn/back-to-three-values/">22 Обратно в три значения</a></li><li><a href="/ru/learn/same-neuron-new-weights/">23 Тот же нейрон, другие веса</a></li><li><a href="/ru/learn/a-neuron-in-chunks/">24 Нейрон по кускам</a></li></ol></li><li><h3>Модуль 9 · Тернарный MAC как чип</h3><p>Скалярное произведение 27 тритов как провода без регистра, та же сумма, которую регистр копит на каждом такте, и небольшая целая сеть в конце курса.</p><ol><li><a href="/ru/learn/a-dot-product-in-wires/">25 Скалярное произведение из проводов</a></li><li><a href="/ru/learn/add-it-up-every-clock/">26 Складывать на каждом такте</a></li><li><a href="/ru/learn/a-whole-network/">27 Целая сеть</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/a-model-on-the-board/index.html b/apps/website/public/ru/learn/a-model-on-the-board/index.html
index 6ef2cf008d..3b72e42765 100644
--- a/apps/website/public/ru/learn/a-model-on-the-board/index.html
+++ b/apps/website/public/ru/learn/a-model-on-the-board/index.html
@@ -316,5 +316,6 @@ <h1>Таблицы модели на плате</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Лаборатория: наши исследования</h3><p>Свой формат чисел, честная таблица результатов и таблицы модели, перемноженные на плате.</p><ol><li><a href="/ru/learn/t27-number-format/">01 T27: наш собственный формат чисел</a></li><li><a href="/ru/learn/an-honest-scoreboard/">02 Честная таблица результатов</a></li><li class="on" aria-current="page"><a href="/ru/learn/a-model-on-the-board/">03 Таблицы модели на плате</a></li></ol></li><li><h3>Модуль 2 · ИИ-числа: блок MX</h3><p>Как ИИ-чипы хранят веса в нескольких битах: один общий масштаб на блок, сам байт масштаба и что один выброс делает с соседями.</p><ol><li><a href="/ru/learn/one-scale-per-block/">04 Один масштаб на блок весов</a></li><li><a href="/ru/learn/scale-byte-native/">05 Байт масштаба на настоящей машине</a></li><li><a href="/ru/learn/one-outlier/">06 Один выброс, много нулей</a></li></ol></li><li><h3>Модуль 3 · Тернарные веса</h3><p>Веса, которые бывают только минус масштаб, ноль или плюс масштаб, пять правил, которые должен пройти тернарный алфавит, и прогон тестов, который ничего не проверил.</p><ol><li><a href="/ru/learn/phi-scaled-ternary/">07 Веса минус фи, ноль и плюс фи</a></li><li><a href="/ru/learn/a-pass-that-checked-nothing/">08 Прогон, который ничего не проверил</a></li><li><a href="/ru/learn/five-rules-for-weights/">09 Пять правил для алфавита весов</a></li></ol></li><li><h3>Модуль 4 · Тернарный формат Ternary Network Float</h3><p>Правило, которое компилятор проверяет до запуска любого теста, и 17-битное число с плавающей точкой, порядок которого — четыре сбалансированных трита.</p><ol><li><a href="/ru/learn/blocked-is-not-failed/">10 Заблокирован — не значит провален</a></li><li><a href="/ru/learn/four-trit-exponent/">11 Число, у которого порядок — четыре трита</a></li><li><a href="/ru/learn/negate-twice/">12 Дважды сменить знак — получить число обратно</a></li></ol></li><li><h3>Модуль 5 · Арифметика чисел со знаком</h3><p>Умножить два числа со знаком, сложить их, когда знаки разные, и сделать то и другое сразу в умножении с накоплением.</p><ol><li><a href="/ru/learn/a-sign-is-one-bit/">13 Знак — это один бит</a></li><li><a href="/ru/learn/opposite-signs-subtract/">14 Разные знаки — вычитание</a></li><li><a href="/ru/learn/multiply-then-add/">15 Умножить, потом сложить</a></li></ol></li><li><h3>Модуль 6 · Части нейрона</h3><p>ReLU с изломом в нуле, степень двойки для softmax и argmax, который называет ответ.</p><ol><li><a href="/ru/learn/a-bend-at-zero/">16 Излом в нуле</a></li><li><a href="/ru/learn/two-to-the-x/">17 Два в степени x</a></li><li><a href="/ru/learn/pick-the-largest/">18 Выбрать наибольшее</a></li></ol></li><li><h3>Модуль 7 · Обучение на ошибке</h3><p>Потеря, которая оценивает неверную догадку в битах, один шаг, который сдвигает вес против градиента, и скрытый слой, который нужен XOR.</p><ol><li><a href="/ru/learn/a-loss-in-bits/">19 Потеря в битах</a></li><li><a href="/ru/learn/one-step-downhill/">20 Один шаг вниз по склону</a></li><li><a href="/ru/learn/xor-needs-a-bend/">21 XOR нужен излом</a></li></ol></li><li><h3>Модуль 8 · BitNet: тернарные сети</h3><p>Порог, который сжимает сумму обратно до трёх значений, один нейрон, который становится другой функцией при смене весов, и нейрон, который читает входы по 27 тритов за раз.</p><ol><li><a href="/ru/learn/back-to-three-values/">22 Обратно в три значения</a></li><li><a href="/ru/learn/same-neuron-new-weights/">23 Тот же нейрон, другие веса</a></li><li><a href="/ru/learn/a-neuron-in-chunks/">24 Нейрон по кускам</a></li></ol></li><li><h3>Модуль 9 · Тернарный MAC как чип</h3><p>Скалярное произведение 27 тритов как провода без регистра, та же сумма, которую регистр копит на каждом такте, и небольшая целая сеть в конце курса.</p><ol><li><a href="/ru/learn/a-dot-product-in-wires/">25 Скалярное произведение из проводов</a></li><li><a href="/ru/learn/add-it-up-every-clock/">26 Складывать на каждом такте</a></li><li><a href="/ru/learn/a-whole-network/">27 Целая сеть</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/a-neuron-in-chunks/index.html b/apps/website/public/ru/learn/a-neuron-in-chunks/index.html
index 98a5e6c297..03fcb2b59f 100644
--- a/apps/website/public/ru/learn/a-neuron-in-chunks/index.html
+++ b/apps/website/public/ru/learn/a-neuron-in-chunks/index.html
@@ -141,5 +141,6 @@ <h1>Нейрон по кускам</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Лаборатория: наши исследования</h3><p>Свой формат чисел, честная таблица результатов и таблицы модели, перемноженные на плате.</p><ol><li><a href="/ru/learn/t27-number-format/">01 T27: наш собственный формат чисел</a></li><li><a href="/ru/learn/an-honest-scoreboard/">02 Честная таблица результатов</a></li><li><a href="/ru/learn/a-model-on-the-board/">03 Таблицы модели на плате</a></li></ol></li><li><h3>Модуль 2 · ИИ-числа: блок MX</h3><p>Как ИИ-чипы хранят веса в нескольких битах: один общий масштаб на блок, сам байт масштаба и что один выброс делает с соседями.</p><ol><li><a href="/ru/learn/one-scale-per-block/">04 Один масштаб на блок весов</a></li><li><a href="/ru/learn/scale-byte-native/">05 Байт масштаба на настоящей машине</a></li><li><a href="/ru/learn/one-outlier/">06 Один выброс, много нулей</a></li></ol></li><li><h3>Модуль 3 · Тернарные веса</h3><p>Веса, которые бывают только минус масштаб, ноль или плюс масштаб, пять правил, которые должен пройти тернарный алфавит, и прогон тестов, который ничего не проверил.</p><ol><li><a href="/ru/learn/phi-scaled-ternary/">07 Веса минус фи, ноль и плюс фи</a></li><li><a href="/ru/learn/a-pass-that-checked-nothing/">08 Прогон, который ничего не проверил</a></li><li><a href="/ru/learn/five-rules-for-weights/">09 Пять правил для алфавита весов</a></li></ol></li><li><h3>Модуль 4 · Тернарный формат Ternary Network Float</h3><p>Правило, которое компилятор проверяет до запуска любого теста, и 17-битное число с плавающей точкой, порядок которого — четыре сбалансированных трита.</p><ol><li><a href="/ru/learn/blocked-is-not-failed/">10 Заблокирован — не значит провален</a></li><li><a href="/ru/learn/four-trit-exponent/">11 Число, у которого порядок — четыре трита</a></li><li><a href="/ru/learn/negate-twice/">12 Дважды сменить знак — получить число обратно</a></li></ol></li><li><h3>Модуль 5 · Арифметика чисел со знаком</h3><p>Умножить два числа со знаком, сложить их, когда знаки разные, и сделать то и другое сразу в умножении с накоплением.</p><ol><li><a href="/ru/learn/a-sign-is-one-bit/">13 Знак — это один бит</a></li><li><a href="/ru/learn/opposite-signs-subtract/">14 Разные знаки — вычитание</a></li><li><a href="/ru/learn/multiply-then-add/">15 Умножить, потом сложить</a></li></ol></li><li><h3>Модуль 6 · Части нейрона</h3><p>ReLU с изломом в нуле, степень двойки для softmax и argmax, который называет ответ.</p><ol><li><a href="/ru/learn/a-bend-at-zero/">16 Излом в нуле</a></li><li><a href="/ru/learn/two-to-the-x/">17 Два в степени x</a></li><li><a href="/ru/learn/pick-the-largest/">18 Выбрать наибольшее</a></li></ol></li><li><h3>Модуль 7 · Обучение на ошибке</h3><p>Потеря, которая оценивает неверную догадку в битах, один шаг, который сдвигает вес против градиента, и скрытый слой, который нужен XOR.</p><ol><li><a href="/ru/learn/a-loss-in-bits/">19 Потеря в битах</a></li><li><a href="/ru/learn/one-step-downhill/">20 Один шаг вниз по склону</a></li><li><a href="/ru/learn/xor-needs-a-bend/">21 XOR нужен излом</a></li></ol></li><li><h3>Модуль 8 · BitNet: тернарные сети</h3><p>Порог, который сжимает сумму обратно до трёх значений, один нейрон, который становится другой функцией при смене весов, и нейрон, который читает входы по 27 тритов за раз.</p><ol><li><a href="/ru/learn/back-to-three-values/">22 Обратно в три значения</a></li><li><a href="/ru/learn/same-neuron-new-weights/">23 Тот же нейрон, другие веса</a></li><li class="on" aria-current="page"><a href="/ru/learn/a-neuron-in-chunks/">24 Нейрон по кускам</a></li></ol></li><li><h3>Модуль 9 · Тернарный MAC как чип</h3><p>Скалярное произведение 27 тритов как провода без регистра, та же сумма, которую регистр копит на каждом такте, и небольшая целая сеть в конце курса.</p><ol><li><a href="/ru/learn/a-dot-product-in-wires/">25 Скалярное произведение из проводов</a></li><li><a href="/ru/learn/add-it-up-every-clock/">26 Складывать на каждом такте</a></li><li><a href="/ru/learn/a-whole-network/">27 Целая сеть</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/a-pass-that-checked-nothing/index.html b/apps/website/public/ru/learn/a-pass-that-checked-nothing/index.html
index 603f86c22e..aa5cc7b450 100644
--- a/apps/website/public/ru/learn/a-pass-that-checked-nothing/index.html
+++ b/apps/website/public/ru/learn/a-pass-that-checked-nothing/index.html
@@ -454,5 +454,6 @@ <h1>Прогон, который ничего не проверил</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Лаборатория: наши исследования</h3><p>Свой формат чисел, честная таблица результатов и таблицы модели, перемноженные на плате.</p><ol><li><a href="/ru/learn/t27-number-format/">01 T27: наш собственный формат чисел</a></li><li><a href="/ru/learn/an-honest-scoreboard/">02 Честная таблица результатов</a></li><li><a href="/ru/learn/a-model-on-the-board/">03 Таблицы модели на плате</a></li></ol></li><li><h3>Модуль 2 · ИИ-числа: блок MX</h3><p>Как ИИ-чипы хранят веса в нескольких битах: один общий масштаб на блок, сам байт масштаба и что один выброс делает с соседями.</p><ol><li><a href="/ru/learn/one-scale-per-block/">04 Один масштаб на блок весов</a></li><li><a href="/ru/learn/scale-byte-native/">05 Байт масштаба на настоящей машине</a></li><li><a href="/ru/learn/one-outlier/">06 Один выброс, много нулей</a></li></ol></li><li><h3>Модуль 3 · Тернарные веса</h3><p>Веса, которые бывают только минус масштаб, ноль или плюс масштаб, пять правил, которые должен пройти тернарный алфавит, и прогон тестов, который ничего не проверил.</p><ol><li><a href="/ru/learn/phi-scaled-ternary/">07 Веса минус фи, ноль и плюс фи</a></li><li class="on" aria-current="page"><a href="/ru/learn/a-pass-that-checked-nothing/">08 Прогон, который ничего не проверил</a></li><li><a href="/ru/learn/five-rules-for-weights/">09 Пять правил для алфавита весов</a></li></ol></li><li><h3>Модуль 4 · Тернарный формат Ternary Network Float</h3><p>Правило, которое компилятор проверяет до запуска любого теста, и 17-битное число с плавающей точкой, порядок которого — четыре сбалансированных трита.</p><ol><li><a href="/ru/learn/blocked-is-not-failed/">10 Заблокирован — не значит провален</a></li><li><a href="/ru/learn/four-trit-exponent/">11 Число, у которого порядок — четыре трита</a></li><li><a href="/ru/learn/negate-twice/">12 Дважды сменить знак — получить число обратно</a></li></ol></li><li><h3>Модуль 5 · Арифметика чисел со знаком</h3><p>Умножить два числа со знаком, сложить их, когда знаки разные, и сделать то и другое сразу в умножении с накоплением.</p><ol><li><a href="/ru/learn/a-sign-is-one-bit/">13 Знак — это один бит</a></li><li><a href="/ru/learn/opposite-signs-subtract/">14 Разные знаки — вычитание</a></li><li><a href="/ru/learn/multiply-then-add/">15 Умножить, потом сложить</a></li></ol></li><li><h3>Модуль 6 · Части нейрона</h3><p>ReLU с изломом в нуле, степень двойки для softmax и argmax, который называет ответ.</p><ol><li><a href="/ru/learn/a-bend-at-zero/">16 Излом в нуле</a></li><li><a href="/ru/learn/two-to-the-x/">17 Два в степени x</a></li><li><a href="/ru/learn/pick-the-largest/">18 Выбрать наибольшее</a></li></ol></li><li><h3>Модуль 7 · Обучение на ошибке</h3><p>Потеря, которая оценивает неверную догадку в битах, один шаг, который сдвигает вес против градиента, и скрытый слой, который нужен XOR.</p><ol><li><a href="/ru/learn/a-loss-in-bits/">19 Потеря в битах</a></li><li><a href="/ru/learn/one-step-downhill/">20 Один шаг вниз по склону</a></li><li><a href="/ru/learn/xor-needs-a-bend/">21 XOR нужен излом</a></li></ol></li><li><h3>Модуль 8 · BitNet: тернарные сети</h3><p>Порог, который сжимает сумму обратно до трёх значений, один нейрон, который становится другой функцией при смене весов, и нейрон, который читает входы по 27 тритов за раз.</p><ol><li><a href="/ru/learn/back-to-three-values/">22 Обратно в три значения</a></li><li><a href="/ru/learn/same-neuron-new-weights/">23 Тот же нейрон, другие веса</a></li><li><a href="/ru/learn/a-neuron-in-chunks/">24 Нейрон по кускам</a></li></ol></li><li><h3>Модуль 9 · Тернарный MAC как чип</h3><p>Скалярное произведение 27 тритов как провода без регистра, та же сумма, которую регистр копит на каждом такте, и небольшая целая сеть в конце курса.</p><ol><li><a href="/ru/learn/a-dot-product-in-wires/">25 Скалярное произведение из проводов</a></li><li><a href="/ru/learn/add-it-up-every-clock/">26 Складывать на каждом такте</a></li><li><a href="/ru/learn/a-whole-network/">27 Целая сеть</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/a-sign-is-one-bit/index.html b/apps/website/public/ru/learn/a-sign-is-one-bit/index.html
index 5cfba88efc..16bd969fac 100644
--- a/apps/website/public/ru/learn/a-sign-is-one-bit/index.html
+++ b/apps/website/public/ru/learn/a-sign-is-one-bit/index.html
@@ -192,5 +192,6 @@ <h1>Знак — это один бит</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Лаборатория: наши исследования</h3><p>Свой формат чисел, честная таблица результатов и таблицы модели, перемноженные на плате.</p><ol><li><a href="/ru/learn/t27-number-format/">01 T27: наш собственный формат чисел</a></li><li><a href="/ru/learn/an-honest-scoreboard/">02 Честная таблица результатов</a></li><li><a href="/ru/learn/a-model-on-the-board/">03 Таблицы модели на плате</a></li></ol></li><li><h3>Модуль 2 · ИИ-числа: блок MX</h3><p>Как ИИ-чипы хранят веса в нескольких битах: один общий масштаб на блок, сам байт масштаба и что один выброс делает с соседями.</p><ol><li><a href="/ru/learn/one-scale-per-block/">04 Один масштаб на блок весов</a></li><li><a href="/ru/learn/scale-byte-native/">05 Байт масштаба на настоящей машине</a></li><li><a href="/ru/learn/one-outlier/">06 Один выброс, много нулей</a></li></ol></li><li><h3>Модуль 3 · Тернарные веса</h3><p>Веса, которые бывают только минус масштаб, ноль или плюс масштаб, пять правил, которые должен пройти тернарный алфавит, и прогон тестов, который ничего не проверил.</p><ol><li><a href="/ru/learn/phi-scaled-ternary/">07 Веса минус фи, ноль и плюс фи</a></li><li><a href="/ru/learn/a-pass-that-checked-nothing/">08 Прогон, который ничего не проверил</a></li><li><a href="/ru/learn/five-rules-for-weights/">09 Пять правил для алфавита весов</a></li></ol></li><li><h3>Модуль 4 · Тернарный формат Ternary Network Float</h3><p>Правило, которое компилятор проверяет до запуска любого теста, и 17-битное число с плавающей точкой, порядок которого — четыре сбалансированных трита.</p><ol><li><a href="/ru/learn/blocked-is-not-failed/">10 Заблокирован — не значит провален</a></li><li><a href="/ru/learn/four-trit-exponent/">11 Число, у которого порядок — четыре трита</a></li><li><a href="/ru/learn/negate-twice/">12 Дважды сменить знак — получить число обратно</a></li></ol></li><li><h3>Модуль 5 · Арифметика чисел со знаком</h3><p>Умножить два числа со знаком, сложить их, когда знаки разные, и сделать то и другое сразу в умножении с накоплением.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/a-sign-is-one-bit/">13 Знак — это один бит</a></li><li><a href="/ru/learn/opposite-signs-subtract/">14 Разные знаки — вычитание</a></li><li><a href="/ru/learn/multiply-then-add/">15 Умножить, потом сложить</a></li></ol></li><li><h3>Модуль 6 · Части нейрона</h3><p>ReLU с изломом в нуле, степень двойки для softmax и argmax, который называет ответ.</p><ol><li><a href="/ru/learn/a-bend-at-zero/">16 Излом в нуле</a></li><li><a href="/ru/learn/two-to-the-x/">17 Два в степени x</a></li><li><a href="/ru/learn/pick-the-largest/">18 Выбрать наибольшее</a></li></ol></li><li><h3>Модуль 7 · Обучение на ошибке</h3><p>Потеря, которая оценивает неверную догадку в битах, один шаг, который сдвигает вес против градиента, и скрытый слой, который нужен XOR.</p><ol><li><a href="/ru/learn/a-loss-in-bits/">19 Потеря в битах</a></li><li><a href="/ru/learn/one-step-downhill/">20 Один шаг вниз по склону</a></li><li><a href="/ru/learn/xor-needs-a-bend/">21 XOR нужен излом</a></li></ol></li><li><h3>Модуль 8 · BitNet: тернарные сети</h3><p>Порог, который сжимает сумму обратно до трёх значений, один нейрон, который становится другой функцией при смене весов, и нейрон, который читает входы по 27 тритов за раз.</p><ol><li><a href="/ru/learn/back-to-three-values/">22 Обратно в три значения</a></li><li><a href="/ru/learn/same-neuron-new-weights/">23 Тот же нейрон, другие веса</a></li><li><a href="/ru/learn/a-neuron-in-chunks/">24 Нейрон по кускам</a></li></ol></li><li><h3>Модуль 9 · Тернарный MAC как чип</h3><p>Скалярное произведение 27 тритов как провода без регистра, та же сумма, которую регистр копит на каждом такте, и небольшая целая сеть в конце курса.</p><ol><li><a href="/ru/learn/a-dot-product-in-wires/">25 Скалярное произведение из проводов</a></li><li><a href="/ru/learn/add-it-up-every-clock/">26 Складывать на каждом такте</a></li><li><a href="/ru/learn/a-whole-network/">27 Целая сеть</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/a-small-program/index.html b/apps/website/public/ru/learn/a-small-program/index.html
index e52c87fff9..c6ba0be915 100644
--- a/apps/website/public/ru/learn/a-small-program/index.html
+++ b/apps/website/public/ru/learn/a-small-program/index.html
@@ -89,5 +89,6 @@ <h1>Маленькая законченная программа</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Файл спеки</h3><p>Что лежит в файле .t27: строка module, строки прозы и комментарии.</p><ol><li><a href="/ru/learn/what-a-spec-is/">01 Что такое спека</a></li><li><a href="/ru/learn/the-module-line/">02 Строка module</a></li><li><a href="/ru/learn/comments-and-prose/">03 Комментарии и строки прозы</a></li></ol></li><li><h3>Модуль 2 · Значения и типы</h3><p>Константы, ширина целого числа, а также true, false и текст.</p><ol><li><a href="/ru/learn/constants/">04 Константы</a></li><li><a href="/ru/learn/whole-numbers/">05 Целые числа и их ширина</a></li><li><a href="/ru/learn/true-false-and-text/">06 Истина, ложь и текст</a></li></ol></li><li><h3>Модуль 3 · Массивы, триты и выражения</h3><p>Списки значений, три значения трита и что вычисляют операторы.</p><ol><li><a href="/ru/learn/arrays/">07 Массивы</a></li><li><a href="/ru/learn/trits/">08 Триты: минус один, ноль, один</a></li><li><a href="/ru/learn/expressions/">09 Выражения</a></li></ol></li><li><h3>Модуль 4 · Тесты и функции</h3><p>Блок test, несколько маленьких тестов в одной спеке и функция с типизированными входами.</p><ol><li><a href="/ru/learn/a-test-block/">10 Блок test</a></li><li><a href="/ru/learn/many-tests/">11 Много тестов в одной спеке</a></li><li><a href="/ru/learn/functions/">12 Функции</a></li></ol></li><li><h3>Модуль 5 · Внутри функции</h3><p>Имена через let и var, выбор через if и switch и циклы.</p><ol><li><a href="/ru/learn/local-names/">13 Локальные имена: let и var</a></li><li><a href="/ru/learn/if-and-else/">14 Выбор: if, else и switch</a></li><li><a href="/ru/learn/loops/">15 Циклы: while и for</a></li></ol></li><li><h3>Модуль 6 · Видимость и свои типы</h3><p>Что pub открывает другим модулям, и структуры и перечисления, которые вы объявляете.</p><ol><li><a href="/ru/learn/pub-or-private/">16 pub или закрытое имя</a></li><li><a href="/ru/learn/structs/">17 Структуры</a></li><li><a href="/ru/learn/enums/">18 Перечисления</a></li></ol></li><li><h3>Модуль 7 · Модули, правила и gen-ts</h3><p>Чтение другого модуля, правило, которое выполняется всегда, и спека, превращённая в TypeScript.</p><ol><li><a href="/ru/learn/use-other-modules/">19 Другие модули</a></li><li><a href="/ru/learn/invariants/">20 Инварианты и замеры</a></li><li><a href="/ru/learn/gen-ts/">21 gen-ts: спека превращается в TypeScript</a></li></ol></li><li><h3>Модуль 8 · Компилятор и tri</h3><p>Бэкенды, которые пишет t27c, стадии, в которых он читает спеку, и команда tri.</p><ol><li><a href="/ru/learn/seven-backends/">22 Семь бэкендов</a></li><li><a href="/ru/learn/stages-of-t27c/">23 Стадии t27c</a></li><li><a href="/ru/learn/tri-commands/">24 tri: командная строка</a></li></ol></li><li><h3>Модуль 9 · Ошибки, программа и что дальше</h3><p>Как читать ошибку компилятора, одна маленькая законченная программа и какой курс пройти дальше.</p><ol><li><a href="/ru/learn/reading-errors/">25 Как читать ошибки компилятора</a></li><li class="on" aria-current="page"><a href="/ru/learn/a-small-program/">26 Маленькая законченная программа</a></li><li><a href="/ru/learn/where-next/">27 Куда дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/a-ternary-alu/index.html b/apps/website/public/ru/learn/a-ternary-alu/index.html
index 47c5826451..29385c46a6 100644
--- a/apps/website/public/ru/learn/a-ternary-alu/index.html
+++ b/apps/website/public/ru/learn/a-ternary-alu/index.html
@@ -535,5 +535,6 @@ <h1>Троичное АЛУ</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Три значения</h3><p>Почему три, как сбалансированная троичная система пишет любое число без знака и что дают переворот и отсечение тритов.</p><ol><li><a href="/ru/learn/why-three/">01 Почему три</a></li><li><a href="/ru/learn/balanced-ternary/">02 Сбалансированная троичная</a></li><li><a href="/ru/learn/flip-and-cut/">03 Переворот вместо минуса, отсечение вместо округления</a></li></ol></li><li><h3>Модуль 2 · Логика с «не знаю»</h3><p>Трёхзначные вентили Клини, два вентиля без двоичного двойника и сложение как пара таблиц.</p><ol><li><a href="/ru/learn/kleene-logic/">04 Трёхзначная логика</a></li><li><a href="/ru/learn/consensus-and-any/">05 Консенсус и «любой»</a></li><li><a href="/ru/learn/sum-and-carry/">06 Сумма и перенос</a></li></ol></li><li><h3>Модуль 3 · Складываем триты</h3><p>Полусумматор, полный сумматор и перенос, бегущий влево разряд за разрядом.</p><ol><li><a href="/ru/learn/half-adder/">07 Полусумматор</a></li><li><a href="/ru/learn/full-adder/">08 Полный сумматор</a></li><li><a href="/ru/learn/ripple-carry/">09 Сквозной перенос</a></li></ol></li><li><h3>Модуль 4 · Умножение без умножителя</h3><p>Копия, ноль или переворот: произведение на один трит, умножение столбиком и MAC, который только складывает.</p><ol><li><a href="/ru/learn/times-a-trit/">10 Умножить на один трит</a></li><li><a href="/ru/learn/long-multiplication/">11 Умножение столбиком</a></li><li><a href="/ru/learn/ternary-mac/">12 Троичный MAC</a></li></ol></li><li><h3>Модуль 5 · Триты в двоичной памяти</h3><p>Два бита на трит, пять тритов на байт и сколько бит нужно слову из 27 тритов.</p><ol><li><a href="/ru/learn/two-bits-per-trit/">13 Два бита на трит</a></li><li><a href="/ru/learn/five-trits-a-byte/">14 Пять тритов в байте</a></li><li><a href="/ru/learn/a-word-of-27/">15 Слово из 27 тритов</a></li></ol></li><li><h3>Модуль 6 · Командное слово TRI-27</h3><p>32-битное слово, которое декодирует эмулятор Trinity: его поля, 47 опкодов и 15-битное непосредственное значение.</p><ol><li><a href="/ru/learn/the-instruction-word/">16 Командное слово</a></li><li><a href="/ru/learn/the-opcode-map/">17 Карта опкодов</a></li><li><a href="/ru/learn/the-immediate/">18 Непосредственное значение</a></li></ol></li><li><h3>Модуль 7 · Троичная машина</h3><p>АЛУ из сумматоров этого курса, переход на три стороны и программа, которую можно пройти по шагам.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/a-ternary-alu/">19 Троичное АЛУ</a></li><li><a href="/ru/learn/three-way-jump/">20 Переход на три стороны</a></li><li><a href="/ru/learn/run-a-program/">21 Запустить программу</a></li></ol></li><li><h3>Модуль 8 · Три ответа</h3><p>Сравнение по старшему различному триту, поиск числа по третям и сортировка трёхзначными сравнениями.</p><ol><li><a href="/ru/learn/compare-at-the-top/">22 Сравнение сверху</a></li><li><a href="/ru/learn/questions-in-thirds/">23 Вопросы по третям</a></li><li><a href="/ru/learn/sort-by-three/">24 Сортировка трёхзначными сравнениями</a></li></ol></li><li><h3>Модуль 9 · Троичные нейроны</h3><p>Веса -1, 0, +1: один нейрон, детектор и маленький слой, и нигде нет умножителя.</p><ol><li><a href="/ru/learn/one-ternary-neuron/">25 Один троичный нейрон</a></li><li><a href="/ru/learn/a-bar-detector/">26 Детектор полосы</a></li><li><a href="/ru/learn/a-tiny-layer/">27 Маленький слой</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/a-test-block/index.html b/apps/website/public/ru/learn/a-test-block/index.html
index 70a31d7b42..bddbf44a05 100644
--- a/apps/website/public/ru/learn/a-test-block/index.html
+++ b/apps/website/public/ru/learn/a-test-block/index.html
@@ -82,5 +82,6 @@ <h1>Блок test</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Файл спеки</h3><p>Что лежит в файле .t27: строка module, строки прозы и комментарии.</p><ol><li><a href="/ru/learn/what-a-spec-is/">01 Что такое спека</a></li><li><a href="/ru/learn/the-module-line/">02 Строка module</a></li><li><a href="/ru/learn/comments-and-prose/">03 Комментарии и строки прозы</a></li></ol></li><li><h3>Модуль 2 · Значения и типы</h3><p>Константы, ширина целого числа, а также true, false и текст.</p><ol><li><a href="/ru/learn/constants/">04 Константы</a></li><li><a href="/ru/learn/whole-numbers/">05 Целые числа и их ширина</a></li><li><a href="/ru/learn/true-false-and-text/">06 Истина, ложь и текст</a></li></ol></li><li><h3>Модуль 3 · Массивы, триты и выражения</h3><p>Списки значений, три значения трита и что вычисляют операторы.</p><ol><li><a href="/ru/learn/arrays/">07 Массивы</a></li><li><a href="/ru/learn/trits/">08 Триты: минус один, ноль, один</a></li><li><a href="/ru/learn/expressions/">09 Выражения</a></li></ol></li><li><h3>Модуль 4 · Тесты и функции</h3><p>Блок test, несколько маленьких тестов в одной спеке и функция с типизированными входами.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/a-test-block/">10 Блок test</a></li><li><a href="/ru/learn/many-tests/">11 Много тестов в одной спеке</a></li><li><a href="/ru/learn/functions/">12 Функции</a></li></ol></li><li><h3>Модуль 5 · Внутри функции</h3><p>Имена через let и var, выбор через if и switch и циклы.</p><ol><li><a href="/ru/learn/local-names/">13 Локальные имена: let и var</a></li><li><a href="/ru/learn/if-and-else/">14 Выбор: if, else и switch</a></li><li><a href="/ru/learn/loops/">15 Циклы: while и for</a></li></ol></li><li><h3>Модуль 6 · Видимость и свои типы</h3><p>Что pub открывает другим модулям, и структуры и перечисления, которые вы объявляете.</p><ol><li><a href="/ru/learn/pub-or-private/">16 pub или закрытое имя</a></li><li><a href="/ru/learn/structs/">17 Структуры</a></li><li><a href="/ru/learn/enums/">18 Перечисления</a></li></ol></li><li><h3>Модуль 7 · Модули, правила и gen-ts</h3><p>Чтение другого модуля, правило, которое выполняется всегда, и спека, превращённая в TypeScript.</p><ol><li><a href="/ru/learn/use-other-modules/">19 Другие модули</a></li><li><a href="/ru/learn/invariants/">20 Инварианты и замеры</a></li><li><a href="/ru/learn/gen-ts/">21 gen-ts: спека превращается в TypeScript</a></li></ol></li><li><h3>Модуль 8 · Компилятор и tri</h3><p>Бэкенды, которые пишет t27c, стадии, в которых он читает спеку, и команда tri.</p><ol><li><a href="/ru/learn/seven-backends/">22 Семь бэкендов</a></li><li><a href="/ru/learn/stages-of-t27c/">23 Стадии t27c</a></li><li><a href="/ru/learn/tri-commands/">24 tri: командная строка</a></li></ol></li><li><h3>Модуль 9 · Ошибки, программа и что дальше</h3><p>Как читать ошибку компилятора, одна маленькая законченная программа и какой курс пройти дальше.</p><ol><li><a href="/ru/learn/reading-errors/">25 Как читать ошибки компилятора</a></li><li><a href="/ru/learn/a-small-program/">26 Маленькая законченная программа</a></li><li><a href="/ru/learn/where-next/">27 Куда дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/a-tiny-layer/index.html b/apps/website/public/ru/learn/a-tiny-layer/index.html
index dba876968d..81eae673bf 100644
--- a/apps/website/public/ru/learn/a-tiny-layer/index.html
+++ b/apps/website/public/ru/learn/a-tiny-layer/index.html
@@ -132,5 +132,6 @@ <h1>Маленький слой</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Три значения</h3><p>Почему три, как сбалансированная троичная система пишет любое число без знака и что дают переворот и отсечение тритов.</p><ol><li><a href="/ru/learn/why-three/">01 Почему три</a></li><li><a href="/ru/learn/balanced-ternary/">02 Сбалансированная троичная</a></li><li><a href="/ru/learn/flip-and-cut/">03 Переворот вместо минуса, отсечение вместо округления</a></li></ol></li><li><h3>Модуль 2 · Логика с «не знаю»</h3><p>Трёхзначные вентили Клини, два вентиля без двоичного двойника и сложение как пара таблиц.</p><ol><li><a href="/ru/learn/kleene-logic/">04 Трёхзначная логика</a></li><li><a href="/ru/learn/consensus-and-any/">05 Консенсус и «любой»</a></li><li><a href="/ru/learn/sum-and-carry/">06 Сумма и перенос</a></li></ol></li><li><h3>Модуль 3 · Складываем триты</h3><p>Полусумматор, полный сумматор и перенос, бегущий влево разряд за разрядом.</p><ol><li><a href="/ru/learn/half-adder/">07 Полусумматор</a></li><li><a href="/ru/learn/full-adder/">08 Полный сумматор</a></li><li><a href="/ru/learn/ripple-carry/">09 Сквозной перенос</a></li></ol></li><li><h3>Модуль 4 · Умножение без умножителя</h3><p>Копия, ноль или переворот: произведение на один трит, умножение столбиком и MAC, который только складывает.</p><ol><li><a href="/ru/learn/times-a-trit/">10 Умножить на один трит</a></li><li><a href="/ru/learn/long-multiplication/">11 Умножение столбиком</a></li><li><a href="/ru/learn/ternary-mac/">12 Троичный MAC</a></li></ol></li><li><h3>Модуль 5 · Триты в двоичной памяти</h3><p>Два бита на трит, пять тритов на байт и сколько бит нужно слову из 27 тритов.</p><ol><li><a href="/ru/learn/two-bits-per-trit/">13 Два бита на трит</a></li><li><a href="/ru/learn/five-trits-a-byte/">14 Пять тритов в байте</a></li><li><a href="/ru/learn/a-word-of-27/">15 Слово из 27 тритов</a></li></ol></li><li><h3>Модуль 6 · Командное слово TRI-27</h3><p>32-битное слово, которое декодирует эмулятор Trinity: его поля, 47 опкодов и 15-битное непосредственное значение.</p><ol><li><a href="/ru/learn/the-instruction-word/">16 Командное слово</a></li><li><a href="/ru/learn/the-opcode-map/">17 Карта опкодов</a></li><li><a href="/ru/learn/the-immediate/">18 Непосредственное значение</a></li></ol></li><li><h3>Модуль 7 · Троичная машина</h3><p>АЛУ из сумматоров этого курса, переход на три стороны и программа, которую можно пройти по шагам.</p><ol><li><a href="/ru/learn/a-ternary-alu/">19 Троичное АЛУ</a></li><li><a href="/ru/learn/three-way-jump/">20 Переход на три стороны</a></li><li><a href="/ru/learn/run-a-program/">21 Запустить программу</a></li></ol></li><li><h3>Модуль 8 · Три ответа</h3><p>Сравнение по старшему различному триту, поиск числа по третям и сортировка трёхзначными сравнениями.</p><ol><li><a href="/ru/learn/compare-at-the-top/">22 Сравнение сверху</a></li><li><a href="/ru/learn/questions-in-thirds/">23 Вопросы по третям</a></li><li><a href="/ru/learn/sort-by-three/">24 Сортировка трёхзначными сравнениями</a></li></ol></li><li><h3>Модуль 9 · Троичные нейроны</h3><p>Веса -1, 0, +1: один нейрон, детектор и маленький слой, и нигде нет умножителя.</p><ol><li><a href="/ru/learn/one-ternary-neuron/">25 Один троичный нейрон</a></li><li><a href="/ru/learn/a-bar-detector/">26 Детектор полосы</a></li><li class="on" aria-current="page"><a href="/ru/learn/a-tiny-layer/">27 Маленький слой</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/a-whole-network/index.html b/apps/website/public/ru/learn/a-whole-network/index.html
index ac5dc49554..3416283d6d 100644
--- a/apps/website/public/ru/learn/a-whole-network/index.html
+++ b/apps/website/public/ru/learn/a-whole-network/index.html
@@ -142,5 +142,6 @@ <h1>Целая сеть</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Лаборатория: наши исследования</h3><p>Свой формат чисел, честная таблица результатов и таблицы модели, перемноженные на плате.</p><ol><li><a href="/ru/learn/t27-number-format/">01 T27: наш собственный формат чисел</a></li><li><a href="/ru/learn/an-honest-scoreboard/">02 Честная таблица результатов</a></li><li><a href="/ru/learn/a-model-on-the-board/">03 Таблицы модели на плате</a></li></ol></li><li><h3>Модуль 2 · ИИ-числа: блок MX</h3><p>Как ИИ-чипы хранят веса в нескольких битах: один общий масштаб на блок, сам байт масштаба и что один выброс делает с соседями.</p><ol><li><a href="/ru/learn/one-scale-per-block/">04 Один масштаб на блок весов</a></li><li><a href="/ru/learn/scale-byte-native/">05 Байт масштаба на настоящей машине</a></li><li><a href="/ru/learn/one-outlier/">06 Один выброс, много нулей</a></li></ol></li><li><h3>Модуль 3 · Тернарные веса</h3><p>Веса, которые бывают только минус масштаб, ноль или плюс масштаб, пять правил, которые должен пройти тернарный алфавит, и прогон тестов, который ничего не проверил.</p><ol><li><a href="/ru/learn/phi-scaled-ternary/">07 Веса минус фи, ноль и плюс фи</a></li><li><a href="/ru/learn/a-pass-that-checked-nothing/">08 Прогон, который ничего не проверил</a></li><li><a href="/ru/learn/five-rules-for-weights/">09 Пять правил для алфавита весов</a></li></ol></li><li><h3>Модуль 4 · Тернарный формат Ternary Network Float</h3><p>Правило, которое компилятор проверяет до запуска любого теста, и 17-битное число с плавающей точкой, порядок которого — четыре сбалансированных трита.</p><ol><li><a href="/ru/learn/blocked-is-not-failed/">10 Заблокирован — не значит провален</a></li><li><a href="/ru/learn/four-trit-exponent/">11 Число, у которого порядок — четыре трита</a></li><li><a href="/ru/learn/negate-twice/">12 Дважды сменить знак — получить число обратно</a></li></ol></li><li><h3>Модуль 5 · Арифметика чисел со знаком</h3><p>Умножить два числа со знаком, сложить их, когда знаки разные, и сделать то и другое сразу в умножении с накоплением.</p><ol><li><a href="/ru/learn/a-sign-is-one-bit/">13 Знак — это один бит</a></li><li><a href="/ru/learn/opposite-signs-subtract/">14 Разные знаки — вычитание</a></li><li><a href="/ru/learn/multiply-then-add/">15 Умножить, потом сложить</a></li></ol></li><li><h3>Модуль 6 · Части нейрона</h3><p>ReLU с изломом в нуле, степень двойки для softmax и argmax, который называет ответ.</p><ol><li><a href="/ru/learn/a-bend-at-zero/">16 Излом в нуле</a></li><li><a href="/ru/learn/two-to-the-x/">17 Два в степени x</a></li><li><a href="/ru/learn/pick-the-largest/">18 Выбрать наибольшее</a></li></ol></li><li><h3>Модуль 7 · Обучение на ошибке</h3><p>Потеря, которая оценивает неверную догадку в битах, один шаг, который сдвигает вес против градиента, и скрытый слой, который нужен XOR.</p><ol><li><a href="/ru/learn/a-loss-in-bits/">19 Потеря в битах</a></li><li><a href="/ru/learn/one-step-downhill/">20 Один шаг вниз по склону</a></li><li><a href="/ru/learn/xor-needs-a-bend/">21 XOR нужен излом</a></li></ol></li><li><h3>Модуль 8 · BitNet: тернарные сети</h3><p>Порог, который сжимает сумму обратно до трёх значений, один нейрон, который становится другой функцией при смене весов, и нейрон, который читает входы по 27 тритов за раз.</p><ol><li><a href="/ru/learn/back-to-three-values/">22 Обратно в три значения</a></li><li><a href="/ru/learn/same-neuron-new-weights/">23 Тот же нейрон, другие веса</a></li><li><a href="/ru/learn/a-neuron-in-chunks/">24 Нейрон по кускам</a></li></ol></li><li><h3>Модуль 9 · Тернарный MAC как чип</h3><p>Скалярное произведение 27 тритов как провода без регистра, та же сумма, которую регистр копит на каждом такте, и небольшая целая сеть в конце курса.</p><ol><li><a href="/ru/learn/a-dot-product-in-wires/">25 Скалярное произведение из проводов</a></li><li><a href="/ru/learn/add-it-up-every-clock/">26 Складывать на каждом такте</a></li><li class="on" aria-current="page"><a href="/ru/learn/a-whole-network/">27 Целая сеть</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/a-word-of-27/index.html b/apps/website/public/ru/learn/a-word-of-27/index.html
index 3bfcaa9782..1e5cc103b2 100644
--- a/apps/website/public/ru/learn/a-word-of-27/index.html
+++ b/apps/website/public/ru/learn/a-word-of-27/index.html
@@ -769,5 +769,6 @@ <h1>Слово из 27 тритов</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Три значения</h3><p>Почему три, как сбалансированная троичная система пишет любое число без знака и что дают переворот и отсечение тритов.</p><ol><li><a href="/ru/learn/why-three/">01 Почему три</a></li><li><a href="/ru/learn/balanced-ternary/">02 Сбалансированная троичная</a></li><li><a href="/ru/learn/flip-and-cut/">03 Переворот вместо минуса, отсечение вместо округления</a></li></ol></li><li><h3>Модуль 2 · Логика с «не знаю»</h3><p>Трёхзначные вентили Клини, два вентиля без двоичного двойника и сложение как пара таблиц.</p><ol><li><a href="/ru/learn/kleene-logic/">04 Трёхзначная логика</a></li><li><a href="/ru/learn/consensus-and-any/">05 Консенсус и «любой»</a></li><li><a href="/ru/learn/sum-and-carry/">06 Сумма и перенос</a></li></ol></li><li><h3>Модуль 3 · Складываем триты</h3><p>Полусумматор, полный сумматор и перенос, бегущий влево разряд за разрядом.</p><ol><li><a href="/ru/learn/half-adder/">07 Полусумматор</a></li><li><a href="/ru/learn/full-adder/">08 Полный сумматор</a></li><li><a href="/ru/learn/ripple-carry/">09 Сквозной перенос</a></li></ol></li><li><h3>Модуль 4 · Умножение без умножителя</h3><p>Копия, ноль или переворот: произведение на один трит, умножение столбиком и MAC, который только складывает.</p><ol><li><a href="/ru/learn/times-a-trit/">10 Умножить на один трит</a></li><li><a href="/ru/learn/long-multiplication/">11 Умножение столбиком</a></li><li><a href="/ru/learn/ternary-mac/">12 Троичный MAC</a></li></ol></li><li><h3>Модуль 5 · Триты в двоичной памяти</h3><p>Два бита на трит, пять тритов на байт и сколько бит нужно слову из 27 тритов.</p><ol><li><a href="/ru/learn/two-bits-per-trit/">13 Два бита на трит</a></li><li><a href="/ru/learn/five-trits-a-byte/">14 Пять тритов в байте</a></li><li class="on" aria-current="page"><a href="/ru/learn/a-word-of-27/">15 Слово из 27 тритов</a></li></ol></li><li><h3>Модуль 6 · Командное слово TRI-27</h3><p>32-битное слово, которое декодирует эмулятор Trinity: его поля, 47 опкодов и 15-битное непосредственное значение.</p><ol><li><a href="/ru/learn/the-instruction-word/">16 Командное слово</a></li><li><a href="/ru/learn/the-opcode-map/">17 Карта опкодов</a></li><li><a href="/ru/learn/the-immediate/">18 Непосредственное значение</a></li></ol></li><li><h3>Модуль 7 · Троичная машина</h3><p>АЛУ из сумматоров этого курса, переход на три стороны и программа, которую можно пройти по шагам.</p><ol><li><a href="/ru/learn/a-ternary-alu/">19 Троичное АЛУ</a></li><li><a href="/ru/learn/three-way-jump/">20 Переход на три стороны</a></li><li><a href="/ru/learn/run-a-program/">21 Запустить программу</a></li></ol></li><li><h3>Модуль 8 · Три ответа</h3><p>Сравнение по старшему различному триту, поиск числа по третям и сортировка трёхзначными сравнениями.</p><ol><li><a href="/ru/learn/compare-at-the-top/">22 Сравнение сверху</a></li><li><a href="/ru/learn/questions-in-thirds/">23 Вопросы по третям</a></li><li><a href="/ru/learn/sort-by-three/">24 Сортировка трёхзначными сравнениями</a></li></ol></li><li><h3>Модуль 9 · Троичные нейроны</h3><p>Веса -1, 0, +1: один нейрон, детектор и маленький слой, и нигде нет умножителя.</p><ol><li><a href="/ru/learn/one-ternary-neuron/">25 Один троичный нейрон</a></li><li><a href="/ru/learn/a-bar-detector/">26 Детектор полосы</a></li><li><a href="/ru/learn/a-tiny-layer/">27 Маленький слой</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/add-it-up-every-clock/index.html b/apps/website/public/ru/learn/add-it-up-every-clock/index.html
index 8dd338dae9..a86ad21888 100644
--- a/apps/website/public/ru/learn/add-it-up-every-clock/index.html
+++ b/apps/website/public/ru/learn/add-it-up-every-clock/index.html
@@ -122,5 +122,6 @@ <h1>Складывать на каждом такте</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Лаборатория: наши исследования</h3><p>Свой формат чисел, честная таблица результатов и таблицы модели, перемноженные на плате.</p><ol><li><a href="/ru/learn/t27-number-format/">01 T27: наш собственный формат чисел</a></li><li><a href="/ru/learn/an-honest-scoreboard/">02 Честная таблица результатов</a></li><li><a href="/ru/learn/a-model-on-the-board/">03 Таблицы модели на плате</a></li></ol></li><li><h3>Модуль 2 · ИИ-числа: блок MX</h3><p>Как ИИ-чипы хранят веса в нескольких битах: один общий масштаб на блок, сам байт масштаба и что один выброс делает с соседями.</p><ol><li><a href="/ru/learn/one-scale-per-block/">04 Один масштаб на блок весов</a></li><li><a href="/ru/learn/scale-byte-native/">05 Байт масштаба на настоящей машине</a></li><li><a href="/ru/learn/one-outlier/">06 Один выброс, много нулей</a></li></ol></li><li><h3>Модуль 3 · Тернарные веса</h3><p>Веса, которые бывают только минус масштаб, ноль или плюс масштаб, пять правил, которые должен пройти тернарный алфавит, и прогон тестов, который ничего не проверил.</p><ol><li><a href="/ru/learn/phi-scaled-ternary/">07 Веса минус фи, ноль и плюс фи</a></li><li><a href="/ru/learn/a-pass-that-checked-nothing/">08 Прогон, который ничего не проверил</a></li><li><a href="/ru/learn/five-rules-for-weights/">09 Пять правил для алфавита весов</a></li></ol></li><li><h3>Модуль 4 · Тернарный формат Ternary Network Float</h3><p>Правило, которое компилятор проверяет до запуска любого теста, и 17-битное число с плавающей точкой, порядок которого — четыре сбалансированных трита.</p><ol><li><a href="/ru/learn/blocked-is-not-failed/">10 Заблокирован — не значит провален</a></li><li><a href="/ru/learn/four-trit-exponent/">11 Число, у которого порядок — четыре трита</a></li><li><a href="/ru/learn/negate-twice/">12 Дважды сменить знак — получить число обратно</a></li></ol></li><li><h3>Модуль 5 · Арифметика чисел со знаком</h3><p>Умножить два числа со знаком, сложить их, когда знаки разные, и сделать то и другое сразу в умножении с накоплением.</p><ol><li><a href="/ru/learn/a-sign-is-one-bit/">13 Знак — это один бит</a></li><li><a href="/ru/learn/opposite-signs-subtract/">14 Разные знаки — вычитание</a></li><li><a href="/ru/learn/multiply-then-add/">15 Умножить, потом сложить</a></li></ol></li><li><h3>Модуль 6 · Части нейрона</h3><p>ReLU с изломом в нуле, степень двойки для softmax и argmax, который называет ответ.</p><ol><li><a href="/ru/learn/a-bend-at-zero/">16 Излом в нуле</a></li><li><a href="/ru/learn/two-to-the-x/">17 Два в степени x</a></li><li><a href="/ru/learn/pick-the-largest/">18 Выбрать наибольшее</a></li></ol></li><li><h3>Модуль 7 · Обучение на ошибке</h3><p>Потеря, которая оценивает неверную догадку в битах, один шаг, который сдвигает вес против градиента, и скрытый слой, который нужен XOR.</p><ol><li><a href="/ru/learn/a-loss-in-bits/">19 Потеря в битах</a></li><li><a href="/ru/learn/one-step-downhill/">20 Один шаг вниз по склону</a></li><li><a href="/ru/learn/xor-needs-a-bend/">21 XOR нужен излом</a></li></ol></li><li><h3>Модуль 8 · BitNet: тернарные сети</h3><p>Порог, который сжимает сумму обратно до трёх значений, один нейрон, который становится другой функцией при смене весов, и нейрон, который читает входы по 27 тритов за раз.</p><ol><li><a href="/ru/learn/back-to-three-values/">22 Обратно в три значения</a></li><li><a href="/ru/learn/same-neuron-new-weights/">23 Тот же нейрон, другие веса</a></li><li><a href="/ru/learn/a-neuron-in-chunks/">24 Нейрон по кускам</a></li></ol></li><li><h3>Модуль 9 · Тернарный MAC как чип</h3><p>Скалярное произведение 27 тритов как провода без регистра, та же сумма, которую регистр копит на каждом такте, и небольшая целая сеть в конце курса.</p><ol><li><a href="/ru/learn/a-dot-product-in-wires/">25 Скалярное произведение из проводов</a></li><li class="on" aria-current="page"><a href="/ru/learn/add-it-up-every-clock/">26 Складывать на каждом такте</a></li><li><a href="/ru/learn/a-whole-network/">27 Целая сеть</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/address-decode/index.html b/apps/website/public/ru/learn/address-decode/index.html
index 84f0cac487..43fe821a11 100644
--- a/apps/website/public/ru/learn/address-decode/index.html
+++ b/apps/website/public/ru/learn/address-decode/index.html
@@ -228,5 +228,6 @@ <h1>Дешифрация адреса</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое шина</h3><p>Зачем вообще существует шина: разговор по проводам, с кадрами и адресами, и кому позволено говорить.</p><ol><li><a href="/ru/learn/wires-that-talk/">01 Провода, которые говорят</a></li><li><a href="/ru/learn/framing-a-conversation/">02 Кадрирование разговора</a></li><li><a href="/ru/learn/master-and-servant/">03 Хозяин и слуга</a></li></ol></li><li><h3>Модуль 2 · Разговор по UART</h3><p>Двухпроводная шина без такта: кадр, делитель, задающий скорость, и статус, который опрашивает драйвер.</p><ol><li><a href="/ru/learn/the-frame/">04 Кадр: старт, данные, стоп</a></li><li><a href="/ru/learn/baud-and-the-divisor/">05 Бод и делитель</a></li><li><a href="/ru/learn/status-and-fifos/">06 Статус и FIFO</a></li></ol></li><li><h3>Модуль 3 · SPI под тактом</h3><p>Разговор под тактом: четыре режима, лестница предделителей и выбор кристалла на каждого подчинённого.</p><ol><li><a href="/ru/learn/modes-cpol-cpha/">07 Режимы: CPOL и CPHA</a></li><li><a href="/ru/learn/the-prescaler-ladder/">08 Лестница предделителей</a></li><li><a href="/ru/learn/chip-select-and-widths/">09 Выбор кристалла и ширины</a></li></ol></li><li><h3>Модуль 4 · Регистровая шина APB</h3><p>Шина регистров: PSEL и PENABLE, строобы и ожидания, и сколько адресных бит стоит количество периферии.</p><ol><li><a href="/ru/learn/psel-and-penable/">10 PSEL и PENABLE</a></li><li><a href="/ru/learn/strobes-and-waits/">11 Строобы и ожидания</a></li><li class="on" aria-current="page"><a href="/ru/learn/address-decode/">12 Дешифрация адреса</a></li></ol></li><li><h3>Модуль 5 · Пять каналов AXI4</h3><p>Пять каналов: адрес, данные и отклик в обе стороны, lite или full, пакеты и идентификаторы.</p><ol><li><a href="/ru/learn/five-channels/">13 Пять каналов</a></li><li><a href="/ru/learn/lite-or-full/">14 Lite или full</a></li><li><a href="/ru/learn/bursts-and-ids/">15 Пакеты и идентификаторы</a></li></ol></li><li><h3>Модуль 6 · Память</h3><p>Что находится на дальнем конце каждой шины: карты памяти, типы портов и задержка, которую должно покрывать ожидание.</p><ol><li><a href="/ru/learn/memory-maps/">16 Карты памяти</a></li><li><a href="/ru/learn/ports-and-kinds/">17 Порты и типы</a></li><li><a href="/ru/learn/latency/">18 Задержка</a></li></ol></li><li><h3>Модуль 7 · Мосты</h3><p>Почему в дизайне появляется больше одной шины и пакетный мост, переносящий работу между ними.</p><ol><li><a href="/ru/learn/why-bridges/">19 Зачем мосты</a></li><li><a href="/ru/learn/the-packet-bridge/">20 Пакетный мост</a></li><li><a href="/ru/learn/everything-in-one-bench/">21 Всё в одном стенде</a></li></ol></li><li><h3>Модуль 8 · Ethernet: кадры и тайминг</h3><p>Кадры, контрольная сумма кадра, тайминг RGMII и заранее зарегистрированные шаги настоящего запуска.</p><ol><li><a href="/ru/learn/frames-and-the-fcs/">22 Кадры и контрольная сумма</a></li><li><a href="/ru/learn/rgmii-timing/">23 Тайминг RGMII</a></li><li><a href="/ru/learn/bring-up-on-the-bench/">24 Запуск на стенде</a></li></ol></li><li><h3>Модуль 9 · Стенд</h3><p>Дисциплина, оберегающая настоящее железо: кто держит IO, как взять и вернуть захват, и что запускается дальше.</p><ol><li><a href="/ru/learn/who-holds-the-io/">25 Кто держит IO</a></li><li><a href="/ru/learn/claim-and-release/">26 Захват и возврат</a></li><li><a href="/ru/learn/capstone-what-runs-next/">27 Капстоун: что запускается дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/ai-numbers/index.html b/apps/website/public/ru/learn/ai-numbers/index.html
index edf8895971..ba9889bfbe 100644
--- a/apps/website/public/ru/learn/ai-numbers/index.html
+++ b/apps/website/public/ru/learn/ai-numbers/index.html
@@ -67,5 +67,6 @@ <h1>ИИ-числа на t27: от форматов чисел до тернар
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Лаборатория: наши исследования</h3><p>Свой формат чисел, честная таблица результатов и таблицы модели, перемноженные на плате.</p><ol><li><a href="/ru/learn/t27-number-format/">01 T27: наш собственный формат чисел</a></li><li><a href="/ru/learn/an-honest-scoreboard/">02 Честная таблица результатов</a></li><li><a href="/ru/learn/a-model-on-the-board/">03 Таблицы модели на плате</a></li></ol></li><li><h3>Модуль 2 · ИИ-числа: блок MX</h3><p>Как ИИ-чипы хранят веса в нескольких битах: один общий масштаб на блок, сам байт масштаба и что один выброс делает с соседями.</p><ol><li><a href="/ru/learn/one-scale-per-block/">04 Один масштаб на блок весов</a></li><li><a href="/ru/learn/scale-byte-native/">05 Байт масштаба на настоящей машине</a></li><li><a href="/ru/learn/one-outlier/">06 Один выброс, много нулей</a></li></ol></li><li><h3>Модуль 3 · Тернарные веса</h3><p>Веса, которые бывают только минус масштаб, ноль или плюс масштаб, пять правил, которые должен пройти тернарный алфавит, и прогон тестов, который ничего не проверил.</p><ol><li><a href="/ru/learn/phi-scaled-ternary/">07 Веса минус фи, ноль и плюс фи</a></li><li><a href="/ru/learn/a-pass-that-checked-nothing/">08 Прогон, который ничего не проверил</a></li><li><a href="/ru/learn/five-rules-for-weights/">09 Пять правил для алфавита весов</a></li></ol></li><li><h3>Модуль 4 · Тернарный формат Ternary Network Float</h3><p>Правило, которое компилятор проверяет до запуска любого теста, и 17-битное число с плавающей точкой, порядок которого — четыре сбалансированных трита.</p><ol><li><a href="/ru/learn/blocked-is-not-failed/">10 Заблокирован — не значит провален</a></li><li><a href="/ru/learn/four-trit-exponent/">11 Число, у которого порядок — четыре трита</a></li><li><a href="/ru/learn/negate-twice/">12 Дважды сменить знак — получить число обратно</a></li></ol></li><li><h3>Модуль 5 · Арифметика чисел со знаком</h3><p>Умножить два числа со знаком, сложить их, когда знаки разные, и сделать то и другое сразу в умножении с накоплением.</p><ol><li><a href="/ru/learn/a-sign-is-one-bit/">13 Знак — это один бит</a></li><li><a href="/ru/learn/opposite-signs-subtract/">14 Разные знаки — вычитание</a></li><li><a href="/ru/learn/multiply-then-add/">15 Умножить, потом сложить</a></li></ol></li><li><h3>Модуль 6 · Части нейрона</h3><p>ReLU с изломом в нуле, степень двойки для softmax и argmax, который называет ответ.</p><ol><li><a href="/ru/learn/a-bend-at-zero/">16 Излом в нуле</a></li><li><a href="/ru/learn/two-to-the-x/">17 Два в степени x</a></li><li><a href="/ru/learn/pick-the-largest/">18 Выбрать наибольшее</a></li></ol></li><li><h3>Модуль 7 · Обучение на ошибке</h3><p>Потеря, которая оценивает неверную догадку в битах, один шаг, который сдвигает вес против градиента, и скрытый слой, который нужен XOR.</p><ol><li><a href="/ru/learn/a-loss-in-bits/">19 Потеря в битах</a></li><li><a href="/ru/learn/one-step-downhill/">20 Один шаг вниз по склону</a></li><li><a href="/ru/learn/xor-needs-a-bend/">21 XOR нужен излом</a></li></ol></li><li><h3>Модуль 8 · BitNet: тернарные сети</h3><p>Порог, который сжимает сумму обратно до трёх значений, один нейрон, который становится другой функцией при смене весов, и нейрон, который читает входы по 27 тритов за раз.</p><ol><li><a href="/ru/learn/back-to-three-values/">22 Обратно в три значения</a></li><li><a href="/ru/learn/same-neuron-new-weights/">23 Тот же нейрон, другие веса</a></li><li><a href="/ru/learn/a-neuron-in-chunks/">24 Нейрон по кускам</a></li></ol></li><li><h3>Модуль 9 · Тернарный MAC как чип</h3><p>Скалярное произведение 27 тритов как провода без регистра, та же сумма, которую регистр копит на каждом такте, и небольшая целая сеть в конце курса.</p><ol><li><a href="/ru/learn/a-dot-product-in-wires/">25 Скалярное произведение из проводов</a></li><li><a href="/ru/learn/add-it-up-every-clock/">26 Складывать на каждом такте</a></li><li><a href="/ru/learn/a-whole-network/">27 Целая сеть</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/an-honest-scoreboard/index.html b/apps/website/public/ru/learn/an-honest-scoreboard/index.html
index 20df445166..409ca925e9 100644
--- a/apps/website/public/ru/learn/an-honest-scoreboard/index.html
+++ b/apps/website/public/ru/learn/an-honest-scoreboard/index.html
@@ -210,5 +210,6 @@ <h1>Честная таблица результатов</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Лаборатория: наши исследования</h3><p>Свой формат чисел, честная таблица результатов и таблицы модели, перемноженные на плате.</p><ol><li><a href="/ru/learn/t27-number-format/">01 T27: наш собственный формат чисел</a></li><li class="on" aria-current="page"><a href="/ru/learn/an-honest-scoreboard/">02 Честная таблица результатов</a></li><li><a href="/ru/learn/a-model-on-the-board/">03 Таблицы модели на плате</a></li></ol></li><li><h3>Модуль 2 · ИИ-числа: блок MX</h3><p>Как ИИ-чипы хранят веса в нескольких битах: один общий масштаб на блок, сам байт масштаба и что один выброс делает с соседями.</p><ol><li><a href="/ru/learn/one-scale-per-block/">04 Один масштаб на блок весов</a></li><li><a href="/ru/learn/scale-byte-native/">05 Байт масштаба на настоящей машине</a></li><li><a href="/ru/learn/one-outlier/">06 Один выброс, много нулей</a></li></ol></li><li><h3>Модуль 3 · Тернарные веса</h3><p>Веса, которые бывают только минус масштаб, ноль или плюс масштаб, пять правил, которые должен пройти тернарный алфавит, и прогон тестов, который ничего не проверил.</p><ol><li><a href="/ru/learn/phi-scaled-ternary/">07 Веса минус фи, ноль и плюс фи</a></li><li><a href="/ru/learn/a-pass-that-checked-nothing/">08 Прогон, который ничего не проверил</a></li><li><a href="/ru/learn/five-rules-for-weights/">09 Пять правил для алфавита весов</a></li></ol></li><li><h3>Модуль 4 · Тернарный формат Ternary Network Float</h3><p>Правило, которое компилятор проверяет до запуска любого теста, и 17-битное число с плавающей точкой, порядок которого — четыре сбалансированных трита.</p><ol><li><a href="/ru/learn/blocked-is-not-failed/">10 Заблокирован — не значит провален</a></li><li><a href="/ru/learn/four-trit-exponent/">11 Число, у которого порядок — четыре трита</a></li><li><a href="/ru/learn/negate-twice/">12 Дважды сменить знак — получить число обратно</a></li></ol></li><li><h3>Модуль 5 · Арифметика чисел со знаком</h3><p>Умножить два числа со знаком, сложить их, когда знаки разные, и сделать то и другое сразу в умножении с накоплением.</p><ol><li><a href="/ru/learn/a-sign-is-one-bit/">13 Знак — это один бит</a></li><li><a href="/ru/learn/opposite-signs-subtract/">14 Разные знаки — вычитание</a></li><li><a href="/ru/learn/multiply-then-add/">15 Умножить, потом сложить</a></li></ol></li><li><h3>Модуль 6 · Части нейрона</h3><p>ReLU с изломом в нуле, степень двойки для softmax и argmax, который называет ответ.</p><ol><li><a href="/ru/learn/a-bend-at-zero/">16 Излом в нуле</a></li><li><a href="/ru/learn/two-to-the-x/">17 Два в степени x</a></li><li><a href="/ru/learn/pick-the-largest/">18 Выбрать наибольшее</a></li></ol></li><li><h3>Модуль 7 · Обучение на ошибке</h3><p>Потеря, которая оценивает неверную догадку в битах, один шаг, который сдвигает вес против градиента, и скрытый слой, который нужен XOR.</p><ol><li><a href="/ru/learn/a-loss-in-bits/">19 Потеря в битах</a></li><li><a href="/ru/learn/one-step-downhill/">20 Один шаг вниз по склону</a></li><li><a href="/ru/learn/xor-needs-a-bend/">21 XOR нужен излом</a></li></ol></li><li><h3>Модуль 8 · BitNet: тернарные сети</h3><p>Порог, который сжимает сумму обратно до трёх значений, один нейрон, который становится другой функцией при смене весов, и нейрон, который читает входы по 27 тритов за раз.</p><ol><li><a href="/ru/learn/back-to-three-values/">22 Обратно в три значения</a></li><li><a href="/ru/learn/same-neuron-new-weights/">23 Тот же нейрон, другие веса</a></li><li><a href="/ru/learn/a-neuron-in-chunks/">24 Нейрон по кускам</a></li></ol></li><li><h3>Модуль 9 · Тернарный MAC как чип</h3><p>Скалярное произведение 27 тритов как провода без регистра, та же сумма, которую регистр копит на каждом такте, и небольшая целая сеть в конце курса.</p><ol><li><a href="/ru/learn/a-dot-product-in-wires/">25 Скалярное произведение из проводов</a></li><li><a href="/ru/learn/add-it-up-every-clock/">26 Складывать на каждом такте</a></li><li><a href="/ru/learn/a-whole-network/">27 Целая сеть</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/anatomy-of-a-tb/index.html b/apps/website/public/ru/learn/anatomy-of-a-tb/index.html
index f033539e0e..2bb6a424fd 100644
--- a/apps/website/public/ru/learn/anatomy-of-a-tb/index.html
+++ b/apps/website/public/ru/learn/anatomy-of-a-tb/index.html
@@ -304,5 +304,6 @@ <h1>Анатомия тестбенча</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Зачем проверять</h3><p>Дизайны, которые компилируются и ошибаются; модель, которая выносит вердикт; план, записанный до кода.</p><ol><li><a href="/ru/learn/bugs-that-compile/">01 Баги, которые компилируются</a></li><li><a href="/ru/learn/golden-model/">02 Золотая модель</a></li><li><a href="/ru/learn/plan-before-code/">03 Сначала план</a></li></ol></li><li><h3>Модуль 2 · Тестбенчи</h3><p>Стимулы, проверки и вердикт, записанные как один spec рядом с дизайном, который они судят.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/anatomy-of-a-tb/">04 Анатомия тестбенча</a></li><li><a href="/ru/learn/self-checking/">05 Самопроверка вместо всматривания</a></li><li><a href="/ru/learn/directed-vs-random/">06 Направленные против случайных</a></li></ol></li><li><h3>Модуль 3 · Временные диаграммы</h3><p>Трасса каждого сигнала, прочитанная так, как её читает инженер по железу, и два прогона, сравнённые между собой.</p><ol><li><a href="/ru/learn/vcd-format/">07 Формат VCD</a></li><li><a href="/ru/learn/debug-with-waves/">08 Отладка по волнам</a></li><li><a href="/ru/learn/compare-traces/">09 Сравнение двух трасс</a></li></ol></li><li><h3>Модуль 4 · Векторы соответствия</h3><p>Случаи с ответом, записанным рядом, там, откуда их достанет компилятор.</p><ol><li><a href="/ru/learn/what-a-vector-is/">10 Что такое вектор</a></li><li><a href="/ru/learn/golden-ruler-vectors/">11 Векторы как эталонная линейка</a></li><li><a href="/ru/learn/end-to-end-scenarios/">12 Сквозные сценарии</a></li></ol></li><li><h3>Модуль 5 · Косимуляция</h3><p>Spec, симулятор и плата сходятся в одном ответе на стенде Artix-7 XC7A200T, и что делать, когда не сходятся.</p><ol><li><a href="/ru/learn/three-way-agreement/">13 Тройное согласие</a></li><li><a href="/ru/learn/simulator-internals/">14 Внутри симулятора</a></li><li><a href="/ru/learn/when-they-disagree/">15 Когда они не сходятся</a></li></ol></li><li><h3>Модуль 6 · Покрытие</h3><p>Чего коснулись тесты: строки, переключения, состояния — и что прячет это число.</p><ol><li><a href="/ru/learn/line-and-toggle/">16 Строки и переключения</a></li><li><a href="/ru/learn/fsm-coverage/">17 Покрытие автоматов</a></li><li><a href="/ru/learn/coverage-lies/">18 Покрытие лжёт</a></li></ol></li><li><h3>Модуль 7 · Формальные методы</h3><p>Ассерты, верные каждый такт; ограниченный поиск контрпримера; и почему доказательству нужна индукция.</p><ol><li><a href="/ru/learn/assertions/">19 Ассерты</a></li><li><a href="/ru/learn/bounded-model-checking/">20 Ограниченная проверка моделей</a></li><li><a href="/ru/learn/induction/">21 Индукция</a></li></ol></li><li><h3>Модуль 8 · Мутации</h3><p>Ломайте дизайн нарочно и считайте, что заметили тесты.</p><ol><li><a href="/ru/learn/kill-the-mutant/">22 Убить мутанта</a></li><li><a href="/ru/learn/surviving-mutants/">23 Выжившие мутанты</a></li><li><a href="/ru/learn/test-strength/">24 Сила тестов</a></li></ol></li><li><h3>Модуль 9 · Приёмка</h3><p>Одна команда, все квитанции, чистый вердикт, который можно показать.</p><ol><li><a href="/ru/learn/build-verify/">25 Сборка и проверка</a></li><li><a href="/ru/learn/board-selftest/">26 Стенд проверяет себя</a></li><li><a href="/ru/learn/verify-capstone/">27 Капстоун: проверить модуль целиком</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">Этот курс сам является spec t27: прочитайте его</a></p><p>Виджеты сохраняют свои английские слова: каждый живёт в собственном spec t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/angle-and-magnitude/index.html b/apps/website/public/ru/learn/angle-and-magnitude/index.html
index e04a67cdd9..791966af08 100644
--- a/apps/website/public/ru/learn/angle-and-magnitude/index.html
+++ b/apps/website/public/ru/learn/angle-and-magnitude/index.html
@@ -439,5 +439,6 @@ <h1>Угол и модуль</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Сложение</h3><p>Почему сложение медленное: перенос, который идёт вверх по слову, выделенная цепь в каждом слайсе, префиксные сети и деревья сохранения переноса.</p><ol><li><a href="/ru/learn/the-carry-chain/">01 Цепь переноса</a></li><li><a href="/ru/learn/prefix-adders/">02 Взгляд вперёд: префиксные сумматоры</a></li><li><a href="/ru/learn/carry-save/">03 Три числа сразу</a></li></ol></li><li><h3>Модуль 2 · Умножение</h3><p>Произведение — сумма сдвинутых копий; перекодирование Бута вдвое сокращает их, а слайс DSP48E1 делает остальное одним блоком.</p><ol><li><a href="/ru/learn/shift-and-add/">04 Сдвиг и сложение</a></li><li><a href="/ru/learn/booths-trick/">05 Приём Бута</a></li><li><a href="/ru/learn/the-dsp48e1-slice/">06 Слайс DSP48E1</a></li></ol></li><li><h3>Модуль 3 · Фиксированная точка</h3><p>Где стоит двоичная точка, что округление делает со значением и с его средним и что даёт каждый бит квантователя.</p><ol><li><a href="/ru/learn/q-formats/">07 Форматы Q</a></li><li><a href="/ru/learn/rounding-and-saturation/">08 Округление и насыщение</a></li><li><a href="/ru/learn/quantization-noise/">09 Шум квантования</a></li></ol></li><li><h3>Модуль 4 · Функции в железе</h3><p>Синус, косинус, угол и длина из сдвигов и сложений — и когда таблица оказывается выгоднее.</p><ol><li><a href="/ru/learn/rotating-by-shifts/">10 Поворот сдвигами</a></li><li class="on" aria-current="page"><a href="/ru/learn/angle-and-magnitude/">11 Угол и модуль</a></li><li><a href="/ru/learn/tables-and-interpolation/">12 Таблицы и интерполяция</a></li></ol></li><li><h3>Модуль 5 · Сигналы и дискретизация</h3><p>Что дискретизация делает с частотой, как генератор строится из сумматора и что измеряет бин ДПФ.</p><ol><li><a href="/ru/learn/sampling-and-aliasing/">13 Дискретизация и наложение спектров</a></li><li><a href="/ru/learn/the-nco/">14 Генератор с числовым управлением</a></li><li><a href="/ru/learn/bins-of-a-spectrum/">15 Бины спектра</a></li></ol></li><li><h3>Модуль 6 · Фильтры FIR</h3><p>Скользящее среднее, расчёт окном sinc с целыми коэффициентами и раскладка отводов по слайсам DSP.</p><ol><li><a href="/ru/learn/the-moving-average/">16 Скользящее среднее</a></li><li><a href="/ru/learn/designing-taps/">17 Расчёт коэффициентов</a></li><li><a href="/ru/learn/folding-taps-onto-slices/">18 Раскладка отводов по слайсам</a></li></ol></li><li><h3>Модуль 7 · Многоскоростная обработка</h3><p>Понижение частоты дискретизации без заворота шума: децимация, фильтр CIC и полифазная форма.</p><ol><li><a href="/ru/learn/lowering-the-rate/">19 Понижение частоты</a></li><li><a href="/ru/learn/the-cic-filter/">20 Фильтр CIC</a></li><li><a href="/ru/learn/polyphase/">21 Полифазная форма</a></li></ol></li><li><h3>Модуль 8 · БПФ</h3><p>N log N вместо N^2: бабочки, бит-реверсный порядок входа и биты, которые добавляет каждый этап.</p><ol><li><a href="/ru/learn/the-butterfly/">22 Бабочка</a></li><li><a href="/ru/learn/twiddles-and-bit-reversal/">23 Поворачивающие множители и бит-реверс</a></li><li><a href="/ru/learn/growth-through-the-stages/">24 Рост через этапы</a></li></ol></li><li><h3>Модуль 9 · На стенде</h3><p>От арифметики к плате: коррелятор из спеки модема, бюджет слайсов и тайминга и итоговый фильтр.</p><ol><li><a href="/ru/learn/a-correlator-finds-the-frame/">25 Коррелятор находит кадр</a></li><li><a href="/ru/learn/does-it-fit-and-close-timing/">26 Помещается ли и сходится ли тайминг?</a></li><li><a href="/ru/learn/capstone-a-filter-on-the-board/">27 Итог: фильтр на плате</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/arithmetic-and-dsp/index.html b/apps/website/public/ru/learn/arithmetic-and-dsp/index.html
index 74c51bcb1c..258fbc0d1d 100644
--- a/apps/website/public/ru/learn/arithmetic-and-dsp/index.html
+++ b/apps/website/public/ru/learn/arithmetic-and-dsp/index.html
@@ -67,5 +67,6 @@ <h1>Арифметика и DSP на FPGA: от цепи переноса до 
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Сложение</h3><p>Почему сложение медленное: перенос, который идёт вверх по слову, выделенная цепь в каждом слайсе, префиксные сети и деревья сохранения переноса.</p><ol><li><a href="/ru/learn/the-carry-chain/">01 Цепь переноса</a></li><li><a href="/ru/learn/prefix-adders/">02 Взгляд вперёд: префиксные сумматоры</a></li><li><a href="/ru/learn/carry-save/">03 Три числа сразу</a></li></ol></li><li><h3>Модуль 2 · Умножение</h3><p>Произведение — сумма сдвинутых копий; перекодирование Бута вдвое сокращает их, а слайс DSP48E1 делает остальное одним блоком.</p><ol><li><a href="/ru/learn/shift-and-add/">04 Сдвиг и сложение</a></li><li><a href="/ru/learn/booths-trick/">05 Приём Бута</a></li><li><a href="/ru/learn/the-dsp48e1-slice/">06 Слайс DSP48E1</a></li></ol></li><li><h3>Модуль 3 · Фиксированная точка</h3><p>Где стоит двоичная точка, что округление делает со значением и с его средним и что даёт каждый бит квантователя.</p><ol><li><a href="/ru/learn/q-formats/">07 Форматы Q</a></li><li><a href="/ru/learn/rounding-and-saturation/">08 Округление и насыщение</a></li><li><a href="/ru/learn/quantization-noise/">09 Шум квантования</a></li></ol></li><li><h3>Модуль 4 · Функции в железе</h3><p>Синус, косинус, угол и длина из сдвигов и сложений — и когда таблица оказывается выгоднее.</p><ol><li><a href="/ru/learn/rotating-by-shifts/">10 Поворот сдвигами</a></li><li><a href="/ru/learn/angle-and-magnitude/">11 Угол и модуль</a></li><li><a href="/ru/learn/tables-and-interpolation/">12 Таблицы и интерполяция</a></li></ol></li><li><h3>Модуль 5 · Сигналы и дискретизация</h3><p>Что дискретизация делает с частотой, как генератор строится из сумматора и что измеряет бин ДПФ.</p><ol><li><a href="/ru/learn/sampling-and-aliasing/">13 Дискретизация и наложение спектров</a></li><li><a href="/ru/learn/the-nco/">14 Генератор с числовым управлением</a></li><li><a href="/ru/learn/bins-of-a-spectrum/">15 Бины спектра</a></li></ol></li><li><h3>Модуль 6 · Фильтры FIR</h3><p>Скользящее среднее, расчёт окном sinc с целыми коэффициентами и раскладка отводов по слайсам DSP.</p><ol><li><a href="/ru/learn/the-moving-average/">16 Скользящее среднее</a></li><li><a href="/ru/learn/designing-taps/">17 Расчёт коэффициентов</a></li><li><a href="/ru/learn/folding-taps-onto-slices/">18 Раскладка отводов по слайсам</a></li></ol></li><li><h3>Модуль 7 · Многоскоростная обработка</h3><p>Понижение частоты дискретизации без заворота шума: децимация, фильтр CIC и полифазная форма.</p><ol><li><a href="/ru/learn/lowering-the-rate/">19 Понижение частоты</a></li><li><a href="/ru/learn/the-cic-filter/">20 Фильтр CIC</a></li><li><a href="/ru/learn/polyphase/">21 Полифазная форма</a></li></ol></li><li><h3>Модуль 8 · БПФ</h3><p>N log N вместо N^2: бабочки, бит-реверсный порядок входа и биты, которые добавляет каждый этап.</p><ol><li><a href="/ru/learn/the-butterfly/">22 Бабочка</a></li><li><a href="/ru/learn/twiddles-and-bit-reversal/">23 Поворачивающие множители и бит-реверс</a></li><li><a href="/ru/learn/growth-through-the-stages/">24 Рост через этапы</a></li></ol></li><li><h3>Модуль 9 · На стенде</h3><p>От арифметики к плате: коррелятор из спеки модема, бюджет слайсов и тайминга и итоговый фильтр.</p><ol><li><a href="/ru/learn/a-correlator-finds-the-frame/">25 Коррелятор находит кадр</a></li><li><a href="/ru/learn/does-it-fit-and-close-timing/">26 Помещается ли и сходится ли тайминг?</a></li><li><a href="/ru/learn/capstone-a-filter-on-the-board/">27 Итог: фильтр на плате</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/arrays/index.html b/apps/website/public/ru/learn/arrays/index.html
index 728cc68723..75d45a4c17 100644
--- a/apps/website/public/ru/learn/arrays/index.html
+++ b/apps/website/public/ru/learn/arrays/index.html
@@ -84,5 +84,6 @@ <h1>Массивы</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Файл спеки</h3><p>Что лежит в файле .t27: строка module, строки прозы и комментарии.</p><ol><li><a href="/ru/learn/what-a-spec-is/">01 Что такое спека</a></li><li><a href="/ru/learn/the-module-line/">02 Строка module</a></li><li><a href="/ru/learn/comments-and-prose/">03 Комментарии и строки прозы</a></li></ol></li><li><h3>Модуль 2 · Значения и типы</h3><p>Константы, ширина целого числа, а также true, false и текст.</p><ol><li><a href="/ru/learn/constants/">04 Константы</a></li><li><a href="/ru/learn/whole-numbers/">05 Целые числа и их ширина</a></li><li><a href="/ru/learn/true-false-and-text/">06 Истина, ложь и текст</a></li></ol></li><li><h3>Модуль 3 · Массивы, триты и выражения</h3><p>Списки значений, три значения трита и что вычисляют операторы.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/arrays/">07 Массивы</a></li><li><a href="/ru/learn/trits/">08 Триты: минус один, ноль, один</a></li><li><a href="/ru/learn/expressions/">09 Выражения</a></li></ol></li><li><h3>Модуль 4 · Тесты и функции</h3><p>Блок test, несколько маленьких тестов в одной спеке и функция с типизированными входами.</p><ol><li><a href="/ru/learn/a-test-block/">10 Блок test</a></li><li><a href="/ru/learn/many-tests/">11 Много тестов в одной спеке</a></li><li><a href="/ru/learn/functions/">12 Функции</a></li></ol></li><li><h3>Модуль 5 · Внутри функции</h3><p>Имена через let и var, выбор через if и switch и циклы.</p><ol><li><a href="/ru/learn/local-names/">13 Локальные имена: let и var</a></li><li><a href="/ru/learn/if-and-else/">14 Выбор: if, else и switch</a></li><li><a href="/ru/learn/loops/">15 Циклы: while и for</a></li></ol></li><li><h3>Модуль 6 · Видимость и свои типы</h3><p>Что pub открывает другим модулям, и структуры и перечисления, которые вы объявляете.</p><ol><li><a href="/ru/learn/pub-or-private/">16 pub или закрытое имя</a></li><li><a href="/ru/learn/structs/">17 Структуры</a></li><li><a href="/ru/learn/enums/">18 Перечисления</a></li></ol></li><li><h3>Модуль 7 · Модули, правила и gen-ts</h3><p>Чтение другого модуля, правило, которое выполняется всегда, и спека, превращённая в TypeScript.</p><ol><li><a href="/ru/learn/use-other-modules/">19 Другие модули</a></li><li><a href="/ru/learn/invariants/">20 Инварианты и замеры</a></li><li><a href="/ru/learn/gen-ts/">21 gen-ts: спека превращается в TypeScript</a></li></ol></li><li><h3>Модуль 8 · Компилятор и tri</h3><p>Бэкенды, которые пишет t27c, стадии, в которых он читает спеку, и команда tri.</p><ol><li><a href="/ru/learn/seven-backends/">22 Семь бэкендов</a></li><li><a href="/ru/learn/stages-of-t27c/">23 Стадии t27c</a></li><li><a href="/ru/learn/tri-commands/">24 tri: командная строка</a></li></ol></li><li><h3>Модуль 9 · Ошибки, программа и что дальше</h3><p>Как читать ошибку компилятора, одна маленькая законченная программа и какой курс пройти дальше.</p><ol><li><a href="/ru/learn/reading-errors/">25 Как читать ошибки компилятора</a></li><li><a href="/ru/learn/a-small-program/">26 Маленькая законченная программа</a></li><li><a href="/ru/learn/where-next/">27 Куда дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/assertions/index.html b/apps/website/public/ru/learn/assertions/index.html
index 8157126454..1a8b517ca6 100644
--- a/apps/website/public/ru/learn/assertions/index.html
+++ b/apps/website/public/ru/learn/assertions/index.html
@@ -424,5 +424,6 @@ <h1>Ассерты</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Зачем проверять</h3><p>Дизайны, которые компилируются и ошибаются; модель, которая выносит вердикт; план, записанный до кода.</p><ol><li><a href="/ru/learn/bugs-that-compile/">01 Баги, которые компилируются</a></li><li><a href="/ru/learn/golden-model/">02 Золотая модель</a></li><li><a href="/ru/learn/plan-before-code/">03 Сначала план</a></li></ol></li><li><h3>Модуль 2 · Тестбенчи</h3><p>Стимулы, проверки и вердикт, записанные как один spec рядом с дизайном, который они судят.</p><ol><li><a href="/ru/learn/anatomy-of-a-tb/">04 Анатомия тестбенча</a></li><li><a href="/ru/learn/self-checking/">05 Самопроверка вместо всматривания</a></li><li><a href="/ru/learn/directed-vs-random/">06 Направленные против случайных</a></li></ol></li><li><h3>Модуль 3 · Временные диаграммы</h3><p>Трасса каждого сигнала, прочитанная так, как её читает инженер по железу, и два прогона, сравнённые между собой.</p><ol><li><a href="/ru/learn/vcd-format/">07 Формат VCD</a></li><li><a href="/ru/learn/debug-with-waves/">08 Отладка по волнам</a></li><li><a href="/ru/learn/compare-traces/">09 Сравнение двух трасс</a></li></ol></li><li><h3>Модуль 4 · Векторы соответствия</h3><p>Случаи с ответом, записанным рядом, там, откуда их достанет компилятор.</p><ol><li><a href="/ru/learn/what-a-vector-is/">10 Что такое вектор</a></li><li><a href="/ru/learn/golden-ruler-vectors/">11 Векторы как эталонная линейка</a></li><li><a href="/ru/learn/end-to-end-scenarios/">12 Сквозные сценарии</a></li></ol></li><li><h3>Модуль 5 · Косимуляция</h3><p>Spec, симулятор и плата сходятся в одном ответе на стенде Artix-7 XC7A200T, и что делать, когда не сходятся.</p><ol><li><a href="/ru/learn/three-way-agreement/">13 Тройное согласие</a></li><li><a href="/ru/learn/simulator-internals/">14 Внутри симулятора</a></li><li><a href="/ru/learn/when-they-disagree/">15 Когда они не сходятся</a></li></ol></li><li><h3>Модуль 6 · Покрытие</h3><p>Чего коснулись тесты: строки, переключения, состояния — и что прячет это число.</p><ol><li><a href="/ru/learn/line-and-toggle/">16 Строки и переключения</a></li><li><a href="/ru/learn/fsm-coverage/">17 Покрытие автоматов</a></li><li><a href="/ru/learn/coverage-lies/">18 Покрытие лжёт</a></li></ol></li><li><h3>Модуль 7 · Формальные методы</h3><p>Ассерты, верные каждый такт; ограниченный поиск контрпримера; и почему доказательству нужна индукция.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/assertions/">19 Ассерты</a></li><li><a href="/ru/learn/bounded-model-checking/">20 Ограниченная проверка моделей</a></li><li><a href="/ru/learn/induction/">21 Индукция</a></li></ol></li><li><h3>Модуль 8 · Мутации</h3><p>Ломайте дизайн нарочно и считайте, что заметили тесты.</p><ol><li><a href="/ru/learn/kill-the-mutant/">22 Убить мутанта</a></li><li><a href="/ru/learn/surviving-mutants/">23 Выжившие мутанты</a></li><li><a href="/ru/learn/test-strength/">24 Сила тестов</a></li></ol></li><li><h3>Модуль 9 · Приёмка</h3><p>Одна команда, все квитанции, чистый вердикт, который можно показать.</p><ol><li><a href="/ru/learn/build-verify/">25 Сборка и проверка</a></li><li><a href="/ru/learn/board-selftest/">26 Стенд проверяет себя</a></li><li><a href="/ru/learn/verify-capstone/">27 Капстоун: проверить модуль целиком</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">Этот курс сам является spec t27: прочитайте его</a></p><p>Виджеты сохраняют свои английские слова: каждый живёт в собственном spec t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/back-to-three-values/index.html b/apps/website/public/ru/learn/back-to-three-values/index.html
index f66d0328a1..7a61531a02 100644
--- a/apps/website/public/ru/learn/back-to-three-values/index.html
+++ b/apps/website/public/ru/learn/back-to-three-values/index.html
@@ -103,5 +103,6 @@ <h1>Обратно в три значения</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Лаборатория: наши исследования</h3><p>Свой формат чисел, честная таблица результатов и таблицы модели, перемноженные на плате.</p><ol><li><a href="/ru/learn/t27-number-format/">01 T27: наш собственный формат чисел</a></li><li><a href="/ru/learn/an-honest-scoreboard/">02 Честная таблица результатов</a></li><li><a href="/ru/learn/a-model-on-the-board/">03 Таблицы модели на плате</a></li></ol></li><li><h3>Модуль 2 · ИИ-числа: блок MX</h3><p>Как ИИ-чипы хранят веса в нескольких битах: один общий масштаб на блок, сам байт масштаба и что один выброс делает с соседями.</p><ol><li><a href="/ru/learn/one-scale-per-block/">04 Один масштаб на блок весов</a></li><li><a href="/ru/learn/scale-byte-native/">05 Байт масштаба на настоящей машине</a></li><li><a href="/ru/learn/one-outlier/">06 Один выброс, много нулей</a></li></ol></li><li><h3>Модуль 3 · Тернарные веса</h3><p>Веса, которые бывают только минус масштаб, ноль или плюс масштаб, пять правил, которые должен пройти тернарный алфавит, и прогон тестов, который ничего не проверил.</p><ol><li><a href="/ru/learn/phi-scaled-ternary/">07 Веса минус фи, ноль и плюс фи</a></li><li><a href="/ru/learn/a-pass-that-checked-nothing/">08 Прогон, который ничего не проверил</a></li><li><a href="/ru/learn/five-rules-for-weights/">09 Пять правил для алфавита весов</a></li></ol></li><li><h3>Модуль 4 · Тернарный формат Ternary Network Float</h3><p>Правило, которое компилятор проверяет до запуска любого теста, и 17-битное число с плавающей точкой, порядок которого — четыре сбалансированных трита.</p><ol><li><a href="/ru/learn/blocked-is-not-failed/">10 Заблокирован — не значит провален</a></li><li><a href="/ru/learn/four-trit-exponent/">11 Число, у которого порядок — четыре трита</a></li><li><a href="/ru/learn/negate-twice/">12 Дважды сменить знак — получить число обратно</a></li></ol></li><li><h3>Модуль 5 · Арифметика чисел со знаком</h3><p>Умножить два числа со знаком, сложить их, когда знаки разные, и сделать то и другое сразу в умножении с накоплением.</p><ol><li><a href="/ru/learn/a-sign-is-one-bit/">13 Знак — это один бит</a></li><li><a href="/ru/learn/opposite-signs-subtract/">14 Разные знаки — вычитание</a></li><li><a href="/ru/learn/multiply-then-add/">15 Умножить, потом сложить</a></li></ol></li><li><h3>Модуль 6 · Части нейрона</h3><p>ReLU с изломом в нуле, степень двойки для softmax и argmax, который называет ответ.</p><ol><li><a href="/ru/learn/a-bend-at-zero/">16 Излом в нуле</a></li><li><a href="/ru/learn/two-to-the-x/">17 Два в степени x</a></li><li><a href="/ru/learn/pick-the-largest/">18 Выбрать наибольшее</a></li></ol></li><li><h3>Модуль 7 · Обучение на ошибке</h3><p>Потеря, которая оценивает неверную догадку в битах, один шаг, который сдвигает вес против градиента, и скрытый слой, который нужен XOR.</p><ol><li><a href="/ru/learn/a-loss-in-bits/">19 Потеря в битах</a></li><li><a href="/ru/learn/one-step-downhill/">20 Один шаг вниз по склону</a></li><li><a href="/ru/learn/xor-needs-a-bend/">21 XOR нужен излом</a></li></ol></li><li><h3>Модуль 8 · BitNet: тернарные сети</h3><p>Порог, который сжимает сумму обратно до трёх значений, один нейрон, который становится другой функцией при смене весов, и нейрон, который читает входы по 27 тритов за раз.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/back-to-three-values/">22 Обратно в три значения</a></li><li><a href="/ru/learn/same-neuron-new-weights/">23 Тот же нейрон, другие веса</a></li><li><a href="/ru/learn/a-neuron-in-chunks/">24 Нейрон по кускам</a></li></ol></li><li><h3>Модуль 9 · Тернарный MAC как чип</h3><p>Скалярное произведение 27 тритов как провода без регистра, та же сумма, которую регистр копит на каждом такте, и небольшая целая сеть в конце курса.</p><ol><li><a href="/ru/learn/a-dot-product-in-wires/">25 Скалярное произведение из проводов</a></li><li><a href="/ru/learn/add-it-up-every-clock/">26 Складывать на каждом такте</a></li><li><a href="/ru/learn/a-whole-network/">27 Целая сеть</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/balanced-ternary/index.html b/apps/website/public/ru/learn/balanced-ternary/index.html
index 1c00f2c6e1..db6fa05603 100644
--- a/apps/website/public/ru/learn/balanced-ternary/index.html
+++ b/apps/website/public/ru/learn/balanced-ternary/index.html
@@ -769,5 +769,6 @@ <h1>Сбалансированная троичная</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Три значения</h3><p>Почему три, как сбалансированная троичная система пишет любое число без знака и что дают переворот и отсечение тритов.</p><ol><li><a href="/ru/learn/why-three/">01 Почему три</a></li><li class="on" aria-current="page"><a href="/ru/learn/balanced-ternary/">02 Сбалансированная троичная</a></li><li><a href="/ru/learn/flip-and-cut/">03 Переворот вместо минуса, отсечение вместо округления</a></li></ol></li><li><h3>Модуль 2 · Логика с «не знаю»</h3><p>Трёхзначные вентили Клини, два вентиля без двоичного двойника и сложение как пара таблиц.</p><ol><li><a href="/ru/learn/kleene-logic/">04 Трёхзначная логика</a></li><li><a href="/ru/learn/consensus-and-any/">05 Консенсус и «любой»</a></li><li><a href="/ru/learn/sum-and-carry/">06 Сумма и перенос</a></li></ol></li><li><h3>Модуль 3 · Складываем триты</h3><p>Полусумматор, полный сумматор и перенос, бегущий влево разряд за разрядом.</p><ol><li><a href="/ru/learn/half-adder/">07 Полусумматор</a></li><li><a href="/ru/learn/full-adder/">08 Полный сумматор</a></li><li><a href="/ru/learn/ripple-carry/">09 Сквозной перенос</a></li></ol></li><li><h3>Модуль 4 · Умножение без умножителя</h3><p>Копия, ноль или переворот: произведение на один трит, умножение столбиком и MAC, который только складывает.</p><ol><li><a href="/ru/learn/times-a-trit/">10 Умножить на один трит</a></li><li><a href="/ru/learn/long-multiplication/">11 Умножение столбиком</a></li><li><a href="/ru/learn/ternary-mac/">12 Троичный MAC</a></li></ol></li><li><h3>Модуль 5 · Триты в двоичной памяти</h3><p>Два бита на трит, пять тритов на байт и сколько бит нужно слову из 27 тритов.</p><ol><li><a href="/ru/learn/two-bits-per-trit/">13 Два бита на трит</a></li><li><a href="/ru/learn/five-trits-a-byte/">14 Пять тритов в байте</a></li><li><a href="/ru/learn/a-word-of-27/">15 Слово из 27 тритов</a></li></ol></li><li><h3>Модуль 6 · Командное слово TRI-27</h3><p>32-битное слово, которое декодирует эмулятор Trinity: его поля, 47 опкодов и 15-битное непосредственное значение.</p><ol><li><a href="/ru/learn/the-instruction-word/">16 Командное слово</a></li><li><a href="/ru/learn/the-opcode-map/">17 Карта опкодов</a></li><li><a href="/ru/learn/the-immediate/">18 Непосредственное значение</a></li></ol></li><li><h3>Модуль 7 · Троичная машина</h3><p>АЛУ из сумматоров этого курса, переход на три стороны и программа, которую можно пройти по шагам.</p><ol><li><a href="/ru/learn/a-ternary-alu/">19 Троичное АЛУ</a></li><li><a href="/ru/learn/three-way-jump/">20 Переход на три стороны</a></li><li><a href="/ru/learn/run-a-program/">21 Запустить программу</a></li></ol></li><li><h3>Модуль 8 · Три ответа</h3><p>Сравнение по старшему различному триту, поиск числа по третям и сортировка трёхзначными сравнениями.</p><ol><li><a href="/ru/learn/compare-at-the-top/">22 Сравнение сверху</a></li><li><a href="/ru/learn/questions-in-thirds/">23 Вопросы по третям</a></li><li><a href="/ru/learn/sort-by-three/">24 Сортировка трёхзначными сравнениями</a></li></ol></li><li><h3>Модуль 9 · Троичные нейроны</h3><p>Веса -1, 0, +1: один нейрон, детектор и маленький слой, и нигде нет умножителя.</p><ol><li><a href="/ru/learn/one-ternary-neuron/">25 Один троичный нейрон</a></li><li><a href="/ru/learn/a-bar-detector/">26 Детектор полосы</a></li><li><a href="/ru/learn/a-tiny-layer/">27 Маленький слой</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/baud-and-the-divisor/index.html b/apps/website/public/ru/learn/baud-and-the-divisor/index.html
index 997976f263..9a976acd66 100644
--- a/apps/website/public/ru/learn/baud-and-the-divisor/index.html
+++ b/apps/website/public/ru/learn/baud-and-the-divisor/index.html
@@ -299,5 +299,6 @@ <h1>Бод и делитель</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое шина</h3><p>Зачем вообще существует шина: разговор по проводам, с кадрами и адресами, и кому позволено говорить.</p><ol><li><a href="/ru/learn/wires-that-talk/">01 Провода, которые говорят</a></li><li><a href="/ru/learn/framing-a-conversation/">02 Кадрирование разговора</a></li><li><a href="/ru/learn/master-and-servant/">03 Хозяин и слуга</a></li></ol></li><li><h3>Модуль 2 · Разговор по UART</h3><p>Двухпроводная шина без такта: кадр, делитель, задающий скорость, и статус, который опрашивает драйвер.</p><ol><li><a href="/ru/learn/the-frame/">04 Кадр: старт, данные, стоп</a></li><li class="on" aria-current="page"><a href="/ru/learn/baud-and-the-divisor/">05 Бод и делитель</a></li><li><a href="/ru/learn/status-and-fifos/">06 Статус и FIFO</a></li></ol></li><li><h3>Модуль 3 · SPI под тактом</h3><p>Разговор под тактом: четыре режима, лестница предделителей и выбор кристалла на каждого подчинённого.</p><ol><li><a href="/ru/learn/modes-cpol-cpha/">07 Режимы: CPOL и CPHA</a></li><li><a href="/ru/learn/the-prescaler-ladder/">08 Лестница предделителей</a></li><li><a href="/ru/learn/chip-select-and-widths/">09 Выбор кристалла и ширины</a></li></ol></li><li><h3>Модуль 4 · Регистровая шина APB</h3><p>Шина регистров: PSEL и PENABLE, строобы и ожидания, и сколько адресных бит стоит количество периферии.</p><ol><li><a href="/ru/learn/psel-and-penable/">10 PSEL и PENABLE</a></li><li><a href="/ru/learn/strobes-and-waits/">11 Строобы и ожидания</a></li><li><a href="/ru/learn/address-decode/">12 Дешифрация адреса</a></li></ol></li><li><h3>Модуль 5 · Пять каналов AXI4</h3><p>Пять каналов: адрес, данные и отклик в обе стороны, lite или full, пакеты и идентификаторы.</p><ol><li><a href="/ru/learn/five-channels/">13 Пять каналов</a></li><li><a href="/ru/learn/lite-or-full/">14 Lite или full</a></li><li><a href="/ru/learn/bursts-and-ids/">15 Пакеты и идентификаторы</a></li></ol></li><li><h3>Модуль 6 · Память</h3><p>Что находится на дальнем конце каждой шины: карты памяти, типы портов и задержка, которую должно покрывать ожидание.</p><ol><li><a href="/ru/learn/memory-maps/">16 Карты памяти</a></li><li><a href="/ru/learn/ports-and-kinds/">17 Порты и типы</a></li><li><a href="/ru/learn/latency/">18 Задержка</a></li></ol></li><li><h3>Модуль 7 · Мосты</h3><p>Почему в дизайне появляется больше одной шины и пакетный мост, переносящий работу между ними.</p><ol><li><a href="/ru/learn/why-bridges/">19 Зачем мосты</a></li><li><a href="/ru/learn/the-packet-bridge/">20 Пакетный мост</a></li><li><a href="/ru/learn/everything-in-one-bench/">21 Всё в одном стенде</a></li></ol></li><li><h3>Модуль 8 · Ethernet: кадры и тайминг</h3><p>Кадры, контрольная сумма кадра, тайминг RGMII и заранее зарегистрированные шаги настоящего запуска.</p><ol><li><a href="/ru/learn/frames-and-the-fcs/">22 Кадры и контрольная сумма</a></li><li><a href="/ru/learn/rgmii-timing/">23 Тайминг RGMII</a></li><li><a href="/ru/learn/bring-up-on-the-bench/">24 Запуск на стенде</a></li></ol></li><li><h3>Модуль 9 · Стенд</h3><p>Дисциплина, оберегающая настоящее железо: кто держит IO, как взять и вернуть захват, и что запускается дальше.</p><ol><li><a href="/ru/learn/who-holds-the-io/">25 Кто держит IO</a></li><li><a href="/ru/learn/claim-and-release/">26 Захват и возврат</a></li><li><a href="/ru/learn/capstone-what-runs-next/">27 Капстоун: что запускается дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/bins-of-a-spectrum/index.html b/apps/website/public/ru/learn/bins-of-a-spectrum/index.html
index 8d34b7f3b8..34bae7b3b8 100644
--- a/apps/website/public/ru/learn/bins-of-a-spectrum/index.html
+++ b/apps/website/public/ru/learn/bins-of-a-spectrum/index.html
@@ -1122,5 +1122,6 @@ <h1>Бины спектра</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Сложение</h3><p>Почему сложение медленное: перенос, который идёт вверх по слову, выделенная цепь в каждом слайсе, префиксные сети и деревья сохранения переноса.</p><ol><li><a href="/ru/learn/the-carry-chain/">01 Цепь переноса</a></li><li><a href="/ru/learn/prefix-adders/">02 Взгляд вперёд: префиксные сумматоры</a></li><li><a href="/ru/learn/carry-save/">03 Три числа сразу</a></li></ol></li><li><h3>Модуль 2 · Умножение</h3><p>Произведение — сумма сдвинутых копий; перекодирование Бута вдвое сокращает их, а слайс DSP48E1 делает остальное одним блоком.</p><ol><li><a href="/ru/learn/shift-and-add/">04 Сдвиг и сложение</a></li><li><a href="/ru/learn/booths-trick/">05 Приём Бута</a></li><li><a href="/ru/learn/the-dsp48e1-slice/">06 Слайс DSP48E1</a></li></ol></li><li><h3>Модуль 3 · Фиксированная точка</h3><p>Где стоит двоичная точка, что округление делает со значением и с его средним и что даёт каждый бит квантователя.</p><ol><li><a href="/ru/learn/q-formats/">07 Форматы Q</a></li><li><a href="/ru/learn/rounding-and-saturation/">08 Округление и насыщение</a></li><li><a href="/ru/learn/quantization-noise/">09 Шум квантования</a></li></ol></li><li><h3>Модуль 4 · Функции в железе</h3><p>Синус, косинус, угол и длина из сдвигов и сложений — и когда таблица оказывается выгоднее.</p><ol><li><a href="/ru/learn/rotating-by-shifts/">10 Поворот сдвигами</a></li><li><a href="/ru/learn/angle-and-magnitude/">11 Угол и модуль</a></li><li><a href="/ru/learn/tables-and-interpolation/">12 Таблицы и интерполяция</a></li></ol></li><li><h3>Модуль 5 · Сигналы и дискретизация</h3><p>Что дискретизация делает с частотой, как генератор строится из сумматора и что измеряет бин ДПФ.</p><ol><li><a href="/ru/learn/sampling-and-aliasing/">13 Дискретизация и наложение спектров</a></li><li><a href="/ru/learn/the-nco/">14 Генератор с числовым управлением</a></li><li class="on" aria-current="page"><a href="/ru/learn/bins-of-a-spectrum/">15 Бины спектра</a></li></ol></li><li><h3>Модуль 6 · Фильтры FIR</h3><p>Скользящее среднее, расчёт окном sinc с целыми коэффициентами и раскладка отводов по слайсам DSP.</p><ol><li><a href="/ru/learn/the-moving-average/">16 Скользящее среднее</a></li><li><a href="/ru/learn/designing-taps/">17 Расчёт коэффициентов</a></li><li><a href="/ru/learn/folding-taps-onto-slices/">18 Раскладка отводов по слайсам</a></li></ol></li><li><h3>Модуль 7 · Многоскоростная обработка</h3><p>Понижение частоты дискретизации без заворота шума: децимация, фильтр CIC и полифазная форма.</p><ol><li><a href="/ru/learn/lowering-the-rate/">19 Понижение частоты</a></li><li><a href="/ru/learn/the-cic-filter/">20 Фильтр CIC</a></li><li><a href="/ru/learn/polyphase/">21 Полифазная форма</a></li></ol></li><li><h3>Модуль 8 · БПФ</h3><p>N log N вместо N^2: бабочки, бит-реверсный порядок входа и биты, которые добавляет каждый этап.</p><ol><li><a href="/ru/learn/the-butterfly/">22 Бабочка</a></li><li><a href="/ru/learn/twiddles-and-bit-reversal/">23 Поворачивающие множители и бит-реверс</a></li><li><a href="/ru/learn/growth-through-the-stages/">24 Рост через этапы</a></li></ol></li><li><h3>Модуль 9 · На стенде</h3><p>От арифметики к плате: коррелятор из спеки модема, бюджет слайсов и тайминга и итоговый фильтр.</p><ol><li><a href="/ru/learn/a-correlator-finds-the-frame/">25 Коррелятор находит кадр</a></li><li><a href="/ru/learn/does-it-fit-and-close-timing/">26 Помещается ли и сходится ли тайминг?</a></li><li><a href="/ru/learn/capstone-a-filter-on-the-board/">27 Итог: фильтр на плате</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/bits-and-trits/index.html b/apps/website/public/ru/learn/bits-and-trits/index.html
index ab848b3ef8..638eee10a2 100644
--- a/apps/website/public/ru/learn/bits-and-trits/index.html
+++ b/apps/website/public/ru/learn/bits-and-trits/index.html
@@ -140,5 +140,6 @@ <h1>Биты, триты и GF16</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Чип</h3><p>Что такое FPGA, какой чип у нас и как его выводы встречаются с платой.</p><ol><li><a href="/ru/learn/fpga-is-a-grid/">01 FPGA — это сетка, которую вы настраиваете</a></li><li><a href="/ru/learn/which-chip/">02 Какой это чип?</a></li><li><a href="/ru/learn/pins-and-package/">03 Выводы, шарики и корпус</a></li></ol></li><li><h3>Модуль 2 · Числа в железе</h3><p>Биты, триты и форматы чисел, и сколько стоит арифметика в логике.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/bits-and-trits/">04 Биты, триты и GF16</a></li><li><a href="/ru/learn/what-arithmetic-costs/">05 Сколько стоит арифметика</a></li><li><a href="/ru/learn/how-much-fits/">06 Сколько помещается на чип</a></li></ol></li><li><h3>Модуль 3 · Ваша программа на t27</h3><p>Напишите спеку, проверьте её и увидьте, почему «компилируется» не значит «верно».</p><ol><li><a href="/ru/learn/hello-t27/">07 Привет, t27</a></li><li><a href="/ru/learn/tests-are-the-spec/">08 Тесты — часть спеки</a></li><li><a href="/ru/learn/compiles-is-not-correct/">09 Компилируется — не значит верно</a></li></ol></li><li><h3>Модуль 4 · Внутри t27c</h3><p>Как компилятор читает спеку и что он пишет, включая нативный t27b.</p><ol><li><a href="/ru/learn/compiler-stages/">10 Стадии t27c</a></li><li><a href="/ru/learn/t27c-on-a-real-machine/">11 t27c на настоящей машине</a></li><li><a href="/ru/learn/seven-backends-and-t27b/">12 Семь бэкендов и t27b</a></li></ol></li><li><h3>Модуль 5 · От спеки к железу</h3><p>Verilog, который пишет t27c, ячейки, в которые он превращается, и что делает одна LUT.</p><ol><li><a href="/ru/learn/spec-to-verilog/">13 От спеки к Verilog</a></li><li><a href="/ru/learn/count-the-cells/">14 Посчитайте ячейки</a></li><li><a href="/ru/learn/one-lut-up-close/">15 Одна LUT вблизи</a></li></ol></li><li><h3>Модуль 6 · Читаем синтез</h3><p>Что yosys сообщает о вашем дизайне и какие предупреждения важны.</p><ol><li><a href="/ru/learn/synthesis-warnings/">16 Важные предупреждения синтеза</a></li><li><a href="/ru/learn/state-machines/">17 Конечные автоматы</a></li><li><a href="/ru/learn/where-the-luts-go/">18 Куда уходят LUT</a></li></ol></li><li><h3>Модуль 7 · Размещение, трассировка, тайминг</h3><p>Где ячейки оказываются на кристалле и успевает ли тактовая частота.</p><ol><li><a href="/ru/learn/placement/">19 Размещение: где оказываются ячейки</a></li><li><a href="/ru/learn/logic-depth/">20 Глубина логики</a></li><li><a href="/ru/learn/timing-slack/">21 Тайминг и запас</a></li></ol></li><li><h3>Модуль 8 · Битстрим</h3><p>Как оттрассированный дизайн становится битами, которые загружает чип, без Vivado.</p><ol><li><a href="/ru/learn/fasm-to-frames/">22 От FASM к фреймам</a></li><li><a href="/ru/learn/a-build-receipt/">23 Сборка по пунктам</a></li><li><a href="/ru/learn/read-a-bitstream/">24 Читаем битстрим</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Спросите чип, кто он, загрузите биты и проверьте их.</p><ol><li><a href="/ru/learn/jtag-asks-the-chip/">25 JTAG спрашивает чип</a></li><li><a href="/ru/learn/same-bits-on-the-board/">26 Те же биты — на плате</a></li><li><a href="/ru/learn/capstone/">27 Итог: переугадайте yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/blocked-is-not-failed/index.html b/apps/website/public/ru/learn/blocked-is-not-failed/index.html
index de82ba71cd..2c9c25f855 100644
--- a/apps/website/public/ru/learn/blocked-is-not-failed/index.html
+++ b/apps/website/public/ru/learn/blocked-is-not-failed/index.html
@@ -616,5 +616,6 @@ <h1>Заблокирован — не значит провален</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Лаборатория: наши исследования</h3><p>Свой формат чисел, честная таблица результатов и таблицы модели, перемноженные на плате.</p><ol><li><a href="/ru/learn/t27-number-format/">01 T27: наш собственный формат чисел</a></li><li><a href="/ru/learn/an-honest-scoreboard/">02 Честная таблица результатов</a></li><li><a href="/ru/learn/a-model-on-the-board/">03 Таблицы модели на плате</a></li></ol></li><li><h3>Модуль 2 · ИИ-числа: блок MX</h3><p>Как ИИ-чипы хранят веса в нескольких битах: один общий масштаб на блок, сам байт масштаба и что один выброс делает с соседями.</p><ol><li><a href="/ru/learn/one-scale-per-block/">04 Один масштаб на блок весов</a></li><li><a href="/ru/learn/scale-byte-native/">05 Байт масштаба на настоящей машине</a></li><li><a href="/ru/learn/one-outlier/">06 Один выброс, много нулей</a></li></ol></li><li><h3>Модуль 3 · Тернарные веса</h3><p>Веса, которые бывают только минус масштаб, ноль или плюс масштаб, пять правил, которые должен пройти тернарный алфавит, и прогон тестов, который ничего не проверил.</p><ol><li><a href="/ru/learn/phi-scaled-ternary/">07 Веса минус фи, ноль и плюс фи</a></li><li><a href="/ru/learn/a-pass-that-checked-nothing/">08 Прогон, который ничего не проверил</a></li><li><a href="/ru/learn/five-rules-for-weights/">09 Пять правил для алфавита весов</a></li></ol></li><li><h3>Модуль 4 · Тернарный формат Ternary Network Float</h3><p>Правило, которое компилятор проверяет до запуска любого теста, и 17-битное число с плавающей точкой, порядок которого — четыре сбалансированных трита.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/blocked-is-not-failed/">10 Заблокирован — не значит провален</a></li><li><a href="/ru/learn/four-trit-exponent/">11 Число, у которого порядок — четыре трита</a></li><li><a href="/ru/learn/negate-twice/">12 Дважды сменить знак — получить число обратно</a></li></ol></li><li><h3>Модуль 5 · Арифметика чисел со знаком</h3><p>Умножить два числа со знаком, сложить их, когда знаки разные, и сделать то и другое сразу в умножении с накоплением.</p><ol><li><a href="/ru/learn/a-sign-is-one-bit/">13 Знак — это один бит</a></li><li><a href="/ru/learn/opposite-signs-subtract/">14 Разные знаки — вычитание</a></li><li><a href="/ru/learn/multiply-then-add/">15 Умножить, потом сложить</a></li></ol></li><li><h3>Модуль 6 · Части нейрона</h3><p>ReLU с изломом в нуле, степень двойки для softmax и argmax, который называет ответ.</p><ol><li><a href="/ru/learn/a-bend-at-zero/">16 Излом в нуле</a></li><li><a href="/ru/learn/two-to-the-x/">17 Два в степени x</a></li><li><a href="/ru/learn/pick-the-largest/">18 Выбрать наибольшее</a></li></ol></li><li><h3>Модуль 7 · Обучение на ошибке</h3><p>Потеря, которая оценивает неверную догадку в битах, один шаг, который сдвигает вес против градиента, и скрытый слой, который нужен XOR.</p><ol><li><a href="/ru/learn/a-loss-in-bits/">19 Потеря в битах</a></li><li><a href="/ru/learn/one-step-downhill/">20 Один шаг вниз по склону</a></li><li><a href="/ru/learn/xor-needs-a-bend/">21 XOR нужен излом</a></li></ol></li><li><h3>Модуль 8 · BitNet: тернарные сети</h3><p>Порог, который сжимает сумму обратно до трёх значений, один нейрон, который становится другой функцией при смене весов, и нейрон, который читает входы по 27 тритов за раз.</p><ol><li><a href="/ru/learn/back-to-three-values/">22 Обратно в три значения</a></li><li><a href="/ru/learn/same-neuron-new-weights/">23 Тот же нейрон, другие веса</a></li><li><a href="/ru/learn/a-neuron-in-chunks/">24 Нейрон по кускам</a></li></ol></li><li><h3>Модуль 9 · Тернарный MAC как чип</h3><p>Скалярное произведение 27 тритов как провода без регистра, та же сумма, которую регистр копит на каждом такте, и небольшая целая сеть в конце курса.</p><ol><li><a href="/ru/learn/a-dot-product-in-wires/">25 Скалярное произведение из проводов</a></li><li><a href="/ru/learn/add-it-up-every-clock/">26 Складывать на каждом такте</a></li><li><a href="/ru/learn/a-whole-network/">27 Целая сеть</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/board-selftest/index.html b/apps/website/public/ru/learn/board-selftest/index.html
index 705cf6a969..ec7948942b 100644
--- a/apps/website/public/ru/learn/board-selftest/index.html
+++ b/apps/website/public/ru/learn/board-selftest/index.html
@@ -110,5 +110,6 @@ <h1>Стенд проверяет себя</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Зачем проверять</h3><p>Дизайны, которые компилируются и ошибаются; модель, которая выносит вердикт; план, записанный до кода.</p><ol><li><a href="/ru/learn/bugs-that-compile/">01 Баги, которые компилируются</a></li><li><a href="/ru/learn/golden-model/">02 Золотая модель</a></li><li><a href="/ru/learn/plan-before-code/">03 Сначала план</a></li></ol></li><li><h3>Модуль 2 · Тестбенчи</h3><p>Стимулы, проверки и вердикт, записанные как один spec рядом с дизайном, который они судят.</p><ol><li><a href="/ru/learn/anatomy-of-a-tb/">04 Анатомия тестбенча</a></li><li><a href="/ru/learn/self-checking/">05 Самопроверка вместо всматривания</a></li><li><a href="/ru/learn/directed-vs-random/">06 Направленные против случайных</a></li></ol></li><li><h3>Модуль 3 · Временные диаграммы</h3><p>Трасса каждого сигнала, прочитанная так, как её читает инженер по железу, и два прогона, сравнённые между собой.</p><ol><li><a href="/ru/learn/vcd-format/">07 Формат VCD</a></li><li><a href="/ru/learn/debug-with-waves/">08 Отладка по волнам</a></li><li><a href="/ru/learn/compare-traces/">09 Сравнение двух трасс</a></li></ol></li><li><h3>Модуль 4 · Векторы соответствия</h3><p>Случаи с ответом, записанным рядом, там, откуда их достанет компилятор.</p><ol><li><a href="/ru/learn/what-a-vector-is/">10 Что такое вектор</a></li><li><a href="/ru/learn/golden-ruler-vectors/">11 Векторы как эталонная линейка</a></li><li><a href="/ru/learn/end-to-end-scenarios/">12 Сквозные сценарии</a></li></ol></li><li><h3>Модуль 5 · Косимуляция</h3><p>Spec, симулятор и плата сходятся в одном ответе на стенде Artix-7 XC7A200T, и что делать, когда не сходятся.</p><ol><li><a href="/ru/learn/three-way-agreement/">13 Тройное согласие</a></li><li><a href="/ru/learn/simulator-internals/">14 Внутри симулятора</a></li><li><a href="/ru/learn/when-they-disagree/">15 Когда они не сходятся</a></li></ol></li><li><h3>Модуль 6 · Покрытие</h3><p>Чего коснулись тесты: строки, переключения, состояния — и что прячет это число.</p><ol><li><a href="/ru/learn/line-and-toggle/">16 Строки и переключения</a></li><li><a href="/ru/learn/fsm-coverage/">17 Покрытие автоматов</a></li><li><a href="/ru/learn/coverage-lies/">18 Покрытие лжёт</a></li></ol></li><li><h3>Модуль 7 · Формальные методы</h3><p>Ассерты, верные каждый такт; ограниченный поиск контрпримера; и почему доказательству нужна индукция.</p><ol><li><a href="/ru/learn/assertions/">19 Ассерты</a></li><li><a href="/ru/learn/bounded-model-checking/">20 Ограниченная проверка моделей</a></li><li><a href="/ru/learn/induction/">21 Индукция</a></li></ol></li><li><h3>Модуль 8 · Мутации</h3><p>Ломайте дизайн нарочно и считайте, что заметили тесты.</p><ol><li><a href="/ru/learn/kill-the-mutant/">22 Убить мутанта</a></li><li><a href="/ru/learn/surviving-mutants/">23 Выжившие мутанты</a></li><li><a href="/ru/learn/test-strength/">24 Сила тестов</a></li></ol></li><li><h3>Модуль 9 · Приёмка</h3><p>Одна команда, все квитанции, чистый вердикт, который можно показать.</p><ol><li><a href="/ru/learn/build-verify/">25 Сборка и проверка</a></li><li class="on" aria-current="page"><a href="/ru/learn/board-selftest/">26 Стенд проверяет себя</a></li><li><a href="/ru/learn/verify-capstone/">27 Капстоун: проверить модуль целиком</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">Этот курс сам является spec t27: прочитайте его</a></p><p>Виджеты сохраняют свои английские слова: каждый живёт в собственном spec t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/booths-trick/index.html b/apps/website/public/ru/learn/booths-trick/index.html
index 12f6f49ebf..efb1049e2c 100644
--- a/apps/website/public/ru/learn/booths-trick/index.html
+++ b/apps/website/public/ru/learn/booths-trick/index.html
@@ -372,5 +372,6 @@ <h1>Приём Бута</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Сложение</h3><p>Почему сложение медленное: перенос, который идёт вверх по слову, выделенная цепь в каждом слайсе, префиксные сети и деревья сохранения переноса.</p><ol><li><a href="/ru/learn/the-carry-chain/">01 Цепь переноса</a></li><li><a href="/ru/learn/prefix-adders/">02 Взгляд вперёд: префиксные сумматоры</a></li><li><a href="/ru/learn/carry-save/">03 Три числа сразу</a></li></ol></li><li><h3>Модуль 2 · Умножение</h3><p>Произведение — сумма сдвинутых копий; перекодирование Бута вдвое сокращает их, а слайс DSP48E1 делает остальное одним блоком.</p><ol><li><a href="/ru/learn/shift-and-add/">04 Сдвиг и сложение</a></li><li class="on" aria-current="page"><a href="/ru/learn/booths-trick/">05 Приём Бута</a></li><li><a href="/ru/learn/the-dsp48e1-slice/">06 Слайс DSP48E1</a></li></ol></li><li><h3>Модуль 3 · Фиксированная точка</h3><p>Где стоит двоичная точка, что округление делает со значением и с его средним и что даёт каждый бит квантователя.</p><ol><li><a href="/ru/learn/q-formats/">07 Форматы Q</a></li><li><a href="/ru/learn/rounding-and-saturation/">08 Округление и насыщение</a></li><li><a href="/ru/learn/quantization-noise/">09 Шум квантования</a></li></ol></li><li><h3>Модуль 4 · Функции в железе</h3><p>Синус, косинус, угол и длина из сдвигов и сложений — и когда таблица оказывается выгоднее.</p><ol><li><a href="/ru/learn/rotating-by-shifts/">10 Поворот сдвигами</a></li><li><a href="/ru/learn/angle-and-magnitude/">11 Угол и модуль</a></li><li><a href="/ru/learn/tables-and-interpolation/">12 Таблицы и интерполяция</a></li></ol></li><li><h3>Модуль 5 · Сигналы и дискретизация</h3><p>Что дискретизация делает с частотой, как генератор строится из сумматора и что измеряет бин ДПФ.</p><ol><li><a href="/ru/learn/sampling-and-aliasing/">13 Дискретизация и наложение спектров</a></li><li><a href="/ru/learn/the-nco/">14 Генератор с числовым управлением</a></li><li><a href="/ru/learn/bins-of-a-spectrum/">15 Бины спектра</a></li></ol></li><li><h3>Модуль 6 · Фильтры FIR</h3><p>Скользящее среднее, расчёт окном sinc с целыми коэффициентами и раскладка отводов по слайсам DSP.</p><ol><li><a href="/ru/learn/the-moving-average/">16 Скользящее среднее</a></li><li><a href="/ru/learn/designing-taps/">17 Расчёт коэффициентов</a></li><li><a href="/ru/learn/folding-taps-onto-slices/">18 Раскладка отводов по слайсам</a></li></ol></li><li><h3>Модуль 7 · Многоскоростная обработка</h3><p>Понижение частоты дискретизации без заворота шума: децимация, фильтр CIC и полифазная форма.</p><ol><li><a href="/ru/learn/lowering-the-rate/">19 Понижение частоты</a></li><li><a href="/ru/learn/the-cic-filter/">20 Фильтр CIC</a></li><li><a href="/ru/learn/polyphase/">21 Полифазная форма</a></li></ol></li><li><h3>Модуль 8 · БПФ</h3><p>N log N вместо N^2: бабочки, бит-реверсный порядок входа и биты, которые добавляет каждый этап.</p><ol><li><a href="/ru/learn/the-butterfly/">22 Бабочка</a></li><li><a href="/ru/learn/twiddles-and-bit-reversal/">23 Поворачивающие множители и бит-реверс</a></li><li><a href="/ru/learn/growth-through-the-stages/">24 Рост через этапы</a></li></ol></li><li><h3>Модуль 9 · На стенде</h3><p>От арифметики к плате: коррелятор из спеки модема, бюджет слайсов и тайминга и итоговый фильтр.</p><ol><li><a href="/ru/learn/a-correlator-finds-the-frame/">25 Коррелятор находит кадр</a></li><li><a href="/ru/learn/does-it-fit-and-close-timing/">26 Помещается ли и сходится ли тайминг?</a></li><li><a href="/ru/learn/capstone-a-filter-on-the-board/">27 Итог: фильтр на плате</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/bounded-model-checking/index.html b/apps/website/public/ru/learn/bounded-model-checking/index.html
index 649610f471..083958020e 100644
--- a/apps/website/public/ru/learn/bounded-model-checking/index.html
+++ b/apps/website/public/ru/learn/bounded-model-checking/index.html
@@ -424,5 +424,6 @@ <h1>Ограниченная проверка моделей</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Зачем проверять</h3><p>Дизайны, которые компилируются и ошибаются; модель, которая выносит вердикт; план, записанный до кода.</p><ol><li><a href="/ru/learn/bugs-that-compile/">01 Баги, которые компилируются</a></li><li><a href="/ru/learn/golden-model/">02 Золотая модель</a></li><li><a href="/ru/learn/plan-before-code/">03 Сначала план</a></li></ol></li><li><h3>Модуль 2 · Тестбенчи</h3><p>Стимулы, проверки и вердикт, записанные как один spec рядом с дизайном, который они судят.</p><ol><li><a href="/ru/learn/anatomy-of-a-tb/">04 Анатомия тестбенча</a></li><li><a href="/ru/learn/self-checking/">05 Самопроверка вместо всматривания</a></li><li><a href="/ru/learn/directed-vs-random/">06 Направленные против случайных</a></li></ol></li><li><h3>Модуль 3 · Временные диаграммы</h3><p>Трасса каждого сигнала, прочитанная так, как её читает инженер по железу, и два прогона, сравнённые между собой.</p><ol><li><a href="/ru/learn/vcd-format/">07 Формат VCD</a></li><li><a href="/ru/learn/debug-with-waves/">08 Отладка по волнам</a></li><li><a href="/ru/learn/compare-traces/">09 Сравнение двух трасс</a></li></ol></li><li><h3>Модуль 4 · Векторы соответствия</h3><p>Случаи с ответом, записанным рядом, там, откуда их достанет компилятор.</p><ol><li><a href="/ru/learn/what-a-vector-is/">10 Что такое вектор</a></li><li><a href="/ru/learn/golden-ruler-vectors/">11 Векторы как эталонная линейка</a></li><li><a href="/ru/learn/end-to-end-scenarios/">12 Сквозные сценарии</a></li></ol></li><li><h3>Модуль 5 · Косимуляция</h3><p>Spec, симулятор и плата сходятся в одном ответе на стенде Artix-7 XC7A200T, и что делать, когда не сходятся.</p><ol><li><a href="/ru/learn/three-way-agreement/">13 Тройное согласие</a></li><li><a href="/ru/learn/simulator-internals/">14 Внутри симулятора</a></li><li><a href="/ru/learn/when-they-disagree/">15 Когда они не сходятся</a></li></ol></li><li><h3>Модуль 6 · Покрытие</h3><p>Чего коснулись тесты: строки, переключения, состояния — и что прячет это число.</p><ol><li><a href="/ru/learn/line-and-toggle/">16 Строки и переключения</a></li><li><a href="/ru/learn/fsm-coverage/">17 Покрытие автоматов</a></li><li><a href="/ru/learn/coverage-lies/">18 Покрытие лжёт</a></li></ol></li><li><h3>Модуль 7 · Формальные методы</h3><p>Ассерты, верные каждый такт; ограниченный поиск контрпримера; и почему доказательству нужна индукция.</p><ol><li><a href="/ru/learn/assertions/">19 Ассерты</a></li><li class="on" aria-current="page"><a href="/ru/learn/bounded-model-checking/">20 Ограниченная проверка моделей</a></li><li><a href="/ru/learn/induction/">21 Индукция</a></li></ol></li><li><h3>Модуль 8 · Мутации</h3><p>Ломайте дизайн нарочно и считайте, что заметили тесты.</p><ol><li><a href="/ru/learn/kill-the-mutant/">22 Убить мутанта</a></li><li><a href="/ru/learn/surviving-mutants/">23 Выжившие мутанты</a></li><li><a href="/ru/learn/test-strength/">24 Сила тестов</a></li></ol></li><li><h3>Модуль 9 · Приёмка</h3><p>Одна команда, все квитанции, чистый вердикт, который можно показать.</p><ol><li><a href="/ru/learn/build-verify/">25 Сборка и проверка</a></li><li><a href="/ru/learn/board-selftest/">26 Стенд проверяет себя</a></li><li><a href="/ru/learn/verify-capstone/">27 Капстоун: проверить модуль целиком</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">Этот курс сам является spec t27: прочитайте его</a></p><p>Виджеты сохраняют свои английские слова: каждый живёт в собственном spec t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/bring-up-on-the-bench/index.html b/apps/website/public/ru/learn/bring-up-on-the-bench/index.html
index 2796b141bf..ff04654153 100644
--- a/apps/website/public/ru/learn/bring-up-on-the-bench/index.html
+++ b/apps/website/public/ru/learn/bring-up-on-the-bench/index.html
@@ -109,5 +109,6 @@ <h1>Запуск на стенде</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое шина</h3><p>Зачем вообще существует шина: разговор по проводам, с кадрами и адресами, и кому позволено говорить.</p><ol><li><a href="/ru/learn/wires-that-talk/">01 Провода, которые говорят</a></li><li><a href="/ru/learn/framing-a-conversation/">02 Кадрирование разговора</a></li><li><a href="/ru/learn/master-and-servant/">03 Хозяин и слуга</a></li></ol></li><li><h3>Модуль 2 · Разговор по UART</h3><p>Двухпроводная шина без такта: кадр, делитель, задающий скорость, и статус, который опрашивает драйвер.</p><ol><li><a href="/ru/learn/the-frame/">04 Кадр: старт, данные, стоп</a></li><li><a href="/ru/learn/baud-and-the-divisor/">05 Бод и делитель</a></li><li><a href="/ru/learn/status-and-fifos/">06 Статус и FIFO</a></li></ol></li><li><h3>Модуль 3 · SPI под тактом</h3><p>Разговор под тактом: четыре режима, лестница предделителей и выбор кристалла на каждого подчинённого.</p><ol><li><a href="/ru/learn/modes-cpol-cpha/">07 Режимы: CPOL и CPHA</a></li><li><a href="/ru/learn/the-prescaler-ladder/">08 Лестница предделителей</a></li><li><a href="/ru/learn/chip-select-and-widths/">09 Выбор кристалла и ширины</a></li></ol></li><li><h3>Модуль 4 · Регистровая шина APB</h3><p>Шина регистров: PSEL и PENABLE, строобы и ожидания, и сколько адресных бит стоит количество периферии.</p><ol><li><a href="/ru/learn/psel-and-penable/">10 PSEL и PENABLE</a></li><li><a href="/ru/learn/strobes-and-waits/">11 Строобы и ожидания</a></li><li><a href="/ru/learn/address-decode/">12 Дешифрация адреса</a></li></ol></li><li><h3>Модуль 5 · Пять каналов AXI4</h3><p>Пять каналов: адрес, данные и отклик в обе стороны, lite или full, пакеты и идентификаторы.</p><ol><li><a href="/ru/learn/five-channels/">13 Пять каналов</a></li><li><a href="/ru/learn/lite-or-full/">14 Lite или full</a></li><li><a href="/ru/learn/bursts-and-ids/">15 Пакеты и идентификаторы</a></li></ol></li><li><h3>Модуль 6 · Память</h3><p>Что находится на дальнем конце каждой шины: карты памяти, типы портов и задержка, которую должно покрывать ожидание.</p><ol><li><a href="/ru/learn/memory-maps/">16 Карты памяти</a></li><li><a href="/ru/learn/ports-and-kinds/">17 Порты и типы</a></li><li><a href="/ru/learn/latency/">18 Задержка</a></li></ol></li><li><h3>Модуль 7 · Мосты</h3><p>Почему в дизайне появляется больше одной шины и пакетный мост, переносящий работу между ними.</p><ol><li><a href="/ru/learn/why-bridges/">19 Зачем мосты</a></li><li><a href="/ru/learn/the-packet-bridge/">20 Пакетный мост</a></li><li><a href="/ru/learn/everything-in-one-bench/">21 Всё в одном стенде</a></li></ol></li><li><h3>Модуль 8 · Ethernet: кадры и тайминг</h3><p>Кадры, контрольная сумма кадра, тайминг RGMII и заранее зарегистрированные шаги настоящего запуска.</p><ol><li><a href="/ru/learn/frames-and-the-fcs/">22 Кадры и контрольная сумма</a></li><li><a href="/ru/learn/rgmii-timing/">23 Тайминг RGMII</a></li><li class="on" aria-current="page"><a href="/ru/learn/bring-up-on-the-bench/">24 Запуск на стенде</a></li></ol></li><li><h3>Модуль 9 · Стенд</h3><p>Дисциплина, оберегающая настоящее железо: кто держит IO, как взять и вернуть захват, и что запускается дальше.</p><ol><li><a href="/ru/learn/who-holds-the-io/">25 Кто держит IO</a></li><li><a href="/ru/learn/claim-and-release/">26 Захват и возврат</a></li><li><a href="/ru/learn/capstone-what-runs-next/">27 Капстоун: что запускается дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/bugs-that-compile/index.html b/apps/website/public/ru/learn/bugs-that-compile/index.html
index 64f2d8cdaa..b9d0355349 100644
--- a/apps/website/public/ru/learn/bugs-that-compile/index.html
+++ b/apps/website/public/ru/learn/bugs-that-compile/index.html
@@ -466,5 +466,6 @@ <h1>Баги, которые компилируются</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Зачем проверять</h3><p>Дизайны, которые компилируются и ошибаются; модель, которая выносит вердикт; план, записанный до кода.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/bugs-that-compile/">01 Баги, которые компилируются</a></li><li><a href="/ru/learn/golden-model/">02 Золотая модель</a></li><li><a href="/ru/learn/plan-before-code/">03 Сначала план</a></li></ol></li><li><h3>Модуль 2 · Тестбенчи</h3><p>Стимулы, проверки и вердикт, записанные как один spec рядом с дизайном, который они судят.</p><ol><li><a href="/ru/learn/anatomy-of-a-tb/">04 Анатомия тестбенча</a></li><li><a href="/ru/learn/self-checking/">05 Самопроверка вместо всматривания</a></li><li><a href="/ru/learn/directed-vs-random/">06 Направленные против случайных</a></li></ol></li><li><h3>Модуль 3 · Временные диаграммы</h3><p>Трасса каждого сигнала, прочитанная так, как её читает инженер по железу, и два прогона, сравнённые между собой.</p><ol><li><a href="/ru/learn/vcd-format/">07 Формат VCD</a></li><li><a href="/ru/learn/debug-with-waves/">08 Отладка по волнам</a></li><li><a href="/ru/learn/compare-traces/">09 Сравнение двух трасс</a></li></ol></li><li><h3>Модуль 4 · Векторы соответствия</h3><p>Случаи с ответом, записанным рядом, там, откуда их достанет компилятор.</p><ol><li><a href="/ru/learn/what-a-vector-is/">10 Что такое вектор</a></li><li><a href="/ru/learn/golden-ruler-vectors/">11 Векторы как эталонная линейка</a></li><li><a href="/ru/learn/end-to-end-scenarios/">12 Сквозные сценарии</a></li></ol></li><li><h3>Модуль 5 · Косимуляция</h3><p>Spec, симулятор и плата сходятся в одном ответе на стенде Artix-7 XC7A200T, и что делать, когда не сходятся.</p><ol><li><a href="/ru/learn/three-way-agreement/">13 Тройное согласие</a></li><li><a href="/ru/learn/simulator-internals/">14 Внутри симулятора</a></li><li><a href="/ru/learn/when-they-disagree/">15 Когда они не сходятся</a></li></ol></li><li><h3>Модуль 6 · Покрытие</h3><p>Чего коснулись тесты: строки, переключения, состояния — и что прячет это число.</p><ol><li><a href="/ru/learn/line-and-toggle/">16 Строки и переключения</a></li><li><a href="/ru/learn/fsm-coverage/">17 Покрытие автоматов</a></li><li><a href="/ru/learn/coverage-lies/">18 Покрытие лжёт</a></li></ol></li><li><h3>Модуль 7 · Формальные методы</h3><p>Ассерты, верные каждый такт; ограниченный поиск контрпримера; и почему доказательству нужна индукция.</p><ol><li><a href="/ru/learn/assertions/">19 Ассерты</a></li><li><a href="/ru/learn/bounded-model-checking/">20 Ограниченная проверка моделей</a></li><li><a href="/ru/learn/induction/">21 Индукция</a></li></ol></li><li><h3>Модуль 8 · Мутации</h3><p>Ломайте дизайн нарочно и считайте, что заметили тесты.</p><ol><li><a href="/ru/learn/kill-the-mutant/">22 Убить мутанта</a></li><li><a href="/ru/learn/surviving-mutants/">23 Выжившие мутанты</a></li><li><a href="/ru/learn/test-strength/">24 Сила тестов</a></li></ol></li><li><h3>Модуль 9 · Приёмка</h3><p>Одна команда, все квитанции, чистый вердикт, который можно показать.</p><ol><li><a href="/ru/learn/build-verify/">25 Сборка и проверка</a></li><li><a href="/ru/learn/board-selftest/">26 Стенд проверяет себя</a></li><li><a href="/ru/learn/verify-capstone/">27 Капстоун: проверить модуль целиком</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">Этот курс сам является spec t27: прочитайте его</a></p><p>Виджеты сохраняют свои английские слова: каждый живёт в собственном spec t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/build-verify/index.html b/apps/website/public/ru/learn/build-verify/index.html
index d64cb24fd9..d525ab915b 100644
--- a/apps/website/public/ru/learn/build-verify/index.html
+++ b/apps/website/public/ru/learn/build-verify/index.html
@@ -193,5 +193,6 @@ <h1>Сборка и проверка</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Зачем проверять</h3><p>Дизайны, которые компилируются и ошибаются; модель, которая выносит вердикт; план, записанный до кода.</p><ol><li><a href="/ru/learn/bugs-that-compile/">01 Баги, которые компилируются</a></li><li><a href="/ru/learn/golden-model/">02 Золотая модель</a></li><li><a href="/ru/learn/plan-before-code/">03 Сначала план</a></li></ol></li><li><h3>Модуль 2 · Тестбенчи</h3><p>Стимулы, проверки и вердикт, записанные как один spec рядом с дизайном, который они судят.</p><ol><li><a href="/ru/learn/anatomy-of-a-tb/">04 Анатомия тестбенча</a></li><li><a href="/ru/learn/self-checking/">05 Самопроверка вместо всматривания</a></li><li><a href="/ru/learn/directed-vs-random/">06 Направленные против случайных</a></li></ol></li><li><h3>Модуль 3 · Временные диаграммы</h3><p>Трасса каждого сигнала, прочитанная так, как её читает инженер по железу, и два прогона, сравнённые между собой.</p><ol><li><a href="/ru/learn/vcd-format/">07 Формат VCD</a></li><li><a href="/ru/learn/debug-with-waves/">08 Отладка по волнам</a></li><li><a href="/ru/learn/compare-traces/">09 Сравнение двух трасс</a></li></ol></li><li><h3>Модуль 4 · Векторы соответствия</h3><p>Случаи с ответом, записанным рядом, там, откуда их достанет компилятор.</p><ol><li><a href="/ru/learn/what-a-vector-is/">10 Что такое вектор</a></li><li><a href="/ru/learn/golden-ruler-vectors/">11 Векторы как эталонная линейка</a></li><li><a href="/ru/learn/end-to-end-scenarios/">12 Сквозные сценарии</a></li></ol></li><li><h3>Модуль 5 · Косимуляция</h3><p>Spec, симулятор и плата сходятся в одном ответе на стенде Artix-7 XC7A200T, и что делать, когда не сходятся.</p><ol><li><a href="/ru/learn/three-way-agreement/">13 Тройное согласие</a></li><li><a href="/ru/learn/simulator-internals/">14 Внутри симулятора</a></li><li><a href="/ru/learn/when-they-disagree/">15 Когда они не сходятся</a></li></ol></li><li><h3>Модуль 6 · Покрытие</h3><p>Чего коснулись тесты: строки, переключения, состояния — и что прячет это число.</p><ol><li><a href="/ru/learn/line-and-toggle/">16 Строки и переключения</a></li><li><a href="/ru/learn/fsm-coverage/">17 Покрытие автоматов</a></li><li><a href="/ru/learn/coverage-lies/">18 Покрытие лжёт</a></li></ol></li><li><h3>Модуль 7 · Формальные методы</h3><p>Ассерты, верные каждый такт; ограниченный поиск контрпримера; и почему доказательству нужна индукция.</p><ol><li><a href="/ru/learn/assertions/">19 Ассерты</a></li><li><a href="/ru/learn/bounded-model-checking/">20 Ограниченная проверка моделей</a></li><li><a href="/ru/learn/induction/">21 Индукция</a></li></ol></li><li><h3>Модуль 8 · Мутации</h3><p>Ломайте дизайн нарочно и считайте, что заметили тесты.</p><ol><li><a href="/ru/learn/kill-the-mutant/">22 Убить мутанта</a></li><li><a href="/ru/learn/surviving-mutants/">23 Выжившие мутанты</a></li><li><a href="/ru/learn/test-strength/">24 Сила тестов</a></li></ol></li><li><h3>Модуль 9 · Приёмка</h3><p>Одна команда, все квитанции, чистый вердикт, который можно показать.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/build-verify/">25 Сборка и проверка</a></li><li><a href="/ru/learn/board-selftest/">26 Стенд проверяет себя</a></li><li><a href="/ru/learn/verify-capstone/">27 Капстоун: проверить модуль целиком</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">Этот курс сам является spec t27: прочитайте его</a></p><p>Виджеты сохраняют свои английские слова: каждый живёт в собственном spec t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/bursts-and-ids/index.html b/apps/website/public/ru/learn/bursts-and-ids/index.html
index b9b56a3f2a..32657054b5 100644
--- a/apps/website/public/ru/learn/bursts-and-ids/index.html
+++ b/apps/website/public/ru/learn/bursts-and-ids/index.html
@@ -280,5 +280,6 @@ <h1>Пакеты и идентификаторы</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое шина</h3><p>Зачем вообще существует шина: разговор по проводам, с кадрами и адресами, и кому позволено говорить.</p><ol><li><a href="/ru/learn/wires-that-talk/">01 Провода, которые говорят</a></li><li><a href="/ru/learn/framing-a-conversation/">02 Кадрирование разговора</a></li><li><a href="/ru/learn/master-and-servant/">03 Хозяин и слуга</a></li></ol></li><li><h3>Модуль 2 · Разговор по UART</h3><p>Двухпроводная шина без такта: кадр, делитель, задающий скорость, и статус, который опрашивает драйвер.</p><ol><li><a href="/ru/learn/the-frame/">04 Кадр: старт, данные, стоп</a></li><li><a href="/ru/learn/baud-and-the-divisor/">05 Бод и делитель</a></li><li><a href="/ru/learn/status-and-fifos/">06 Статус и FIFO</a></li></ol></li><li><h3>Модуль 3 · SPI под тактом</h3><p>Разговор под тактом: четыре режима, лестница предделителей и выбор кристалла на каждого подчинённого.</p><ol><li><a href="/ru/learn/modes-cpol-cpha/">07 Режимы: CPOL и CPHA</a></li><li><a href="/ru/learn/the-prescaler-ladder/">08 Лестница предделителей</a></li><li><a href="/ru/learn/chip-select-and-widths/">09 Выбор кристалла и ширины</a></li></ol></li><li><h3>Модуль 4 · Регистровая шина APB</h3><p>Шина регистров: PSEL и PENABLE, строобы и ожидания, и сколько адресных бит стоит количество периферии.</p><ol><li><a href="/ru/learn/psel-and-penable/">10 PSEL и PENABLE</a></li><li><a href="/ru/learn/strobes-and-waits/">11 Строобы и ожидания</a></li><li><a href="/ru/learn/address-decode/">12 Дешифрация адреса</a></li></ol></li><li><h3>Модуль 5 · Пять каналов AXI4</h3><p>Пять каналов: адрес, данные и отклик в обе стороны, lite или full, пакеты и идентификаторы.</p><ol><li><a href="/ru/learn/five-channels/">13 Пять каналов</a></li><li><a href="/ru/learn/lite-or-full/">14 Lite или full</a></li><li class="on" aria-current="page"><a href="/ru/learn/bursts-and-ids/">15 Пакеты и идентификаторы</a></li></ol></li><li><h3>Модуль 6 · Память</h3><p>Что находится на дальнем конце каждой шины: карты памяти, типы портов и задержка, которую должно покрывать ожидание.</p><ol><li><a href="/ru/learn/memory-maps/">16 Карты памяти</a></li><li><a href="/ru/learn/ports-and-kinds/">17 Порты и типы</a></li><li><a href="/ru/learn/latency/">18 Задержка</a></li></ol></li><li><h3>Модуль 7 · Мосты</h3><p>Почему в дизайне появляется больше одной шины и пакетный мост, переносящий работу между ними.</p><ol><li><a href="/ru/learn/why-bridges/">19 Зачем мосты</a></li><li><a href="/ru/learn/the-packet-bridge/">20 Пакетный мост</a></li><li><a href="/ru/learn/everything-in-one-bench/">21 Всё в одном стенде</a></li></ol></li><li><h3>Модуль 8 · Ethernet: кадры и тайминг</h3><p>Кадры, контрольная сумма кадра, тайминг RGMII и заранее зарегистрированные шаги настоящего запуска.</p><ol><li><a href="/ru/learn/frames-and-the-fcs/">22 Кадры и контрольная сумма</a></li><li><a href="/ru/learn/rgmii-timing/">23 Тайминг RGMII</a></li><li><a href="/ru/learn/bring-up-on-the-bench/">24 Запуск на стенде</a></li></ol></li><li><h3>Модуль 9 · Стенд</h3><p>Дисциплина, оберегающая настоящее железо: кто держит IO, как взять и вернуть захват, и что запускается дальше.</p><ol><li><a href="/ru/learn/who-holds-the-io/">25 Кто держит IO</a></li><li><a href="/ru/learn/claim-and-release/">26 Захват и возврат</a></li><li><a href="/ru/learn/capstone-what-runs-next/">27 Капстоун: что запускается дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/buses-and-peripherals/index.html b/apps/website/public/ru/learn/buses-and-peripherals/index.html
index dbd29411ff..ceb398b481 100644
--- a/apps/website/public/ru/learn/buses-and-peripherals/index.html
+++ b/apps/website/public/ru/learn/buses-and-peripherals/index.html
@@ -67,5 +67,6 @@ <h1>Шины и периферия на t27: UART, SPI, APB, AXI4, Ethernet</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое шина</h3><p>Зачем вообще существует шина: разговор по проводам, с кадрами и адресами, и кому позволено говорить.</p><ol><li><a href="/ru/learn/wires-that-talk/">01 Провода, которые говорят</a></li><li><a href="/ru/learn/framing-a-conversation/">02 Кадрирование разговора</a></li><li><a href="/ru/learn/master-and-servant/">03 Хозяин и слуга</a></li></ol></li><li><h3>Модуль 2 · Разговор по UART</h3><p>Двухпроводная шина без такта: кадр, делитель, задающий скорость, и статус, который опрашивает драйвер.</p><ol><li><a href="/ru/learn/the-frame/">04 Кадр: старт, данные, стоп</a></li><li><a href="/ru/learn/baud-and-the-divisor/">05 Бод и делитель</a></li><li><a href="/ru/learn/status-and-fifos/">06 Статус и FIFO</a></li></ol></li><li><h3>Модуль 3 · SPI под тактом</h3><p>Разговор под тактом: четыре режима, лестница предделителей и выбор кристалла на каждого подчинённого.</p><ol><li><a href="/ru/learn/modes-cpol-cpha/">07 Режимы: CPOL и CPHA</a></li><li><a href="/ru/learn/the-prescaler-ladder/">08 Лестница предделителей</a></li><li><a href="/ru/learn/chip-select-and-widths/">09 Выбор кристалла и ширины</a></li></ol></li><li><h3>Модуль 4 · Регистровая шина APB</h3><p>Шина регистров: PSEL и PENABLE, строобы и ожидания, и сколько адресных бит стоит количество периферии.</p><ol><li><a href="/ru/learn/psel-and-penable/">10 PSEL и PENABLE</a></li><li><a href="/ru/learn/strobes-and-waits/">11 Строобы и ожидания</a></li><li><a href="/ru/learn/address-decode/">12 Дешифрация адреса</a></li></ol></li><li><h3>Модуль 5 · Пять каналов AXI4</h3><p>Пять каналов: адрес, данные и отклик в обе стороны, lite или full, пакеты и идентификаторы.</p><ol><li><a href="/ru/learn/five-channels/">13 Пять каналов</a></li><li><a href="/ru/learn/lite-or-full/">14 Lite или full</a></li><li><a href="/ru/learn/bursts-and-ids/">15 Пакеты и идентификаторы</a></li></ol></li><li><h3>Модуль 6 · Память</h3><p>Что находится на дальнем конце каждой шины: карты памяти, типы портов и задержка, которую должно покрывать ожидание.</p><ol><li><a href="/ru/learn/memory-maps/">16 Карты памяти</a></li><li><a href="/ru/learn/ports-and-kinds/">17 Порты и типы</a></li><li><a href="/ru/learn/latency/">18 Задержка</a></li></ol></li><li><h3>Модуль 7 · Мосты</h3><p>Почему в дизайне появляется больше одной шины и пакетный мост, переносящий работу между ними.</p><ol><li><a href="/ru/learn/why-bridges/">19 Зачем мосты</a></li><li><a href="/ru/learn/the-packet-bridge/">20 Пакетный мост</a></li><li><a href="/ru/learn/everything-in-one-bench/">21 Всё в одном стенде</a></li></ol></li><li><h3>Модуль 8 · Ethernet: кадры и тайминг</h3><p>Кадры, контрольная сумма кадра, тайминг RGMII и заранее зарегистрированные шаги настоящего запуска.</p><ol><li><a href="/ru/learn/frames-and-the-fcs/">22 Кадры и контрольная сумма</a></li><li><a href="/ru/learn/rgmii-timing/">23 Тайминг RGMII</a></li><li><a href="/ru/learn/bring-up-on-the-bench/">24 Запуск на стенде</a></li></ol></li><li><h3>Модуль 9 · Стенд</h3><p>Дисциплина, оберегающая настоящее железо: кто держит IO, как взять и вернуть захват, и что запускается дальше.</p><ol><li><a href="/ru/learn/who-holds-the-io/">25 Кто держит IO</a></li><li><a href="/ru/learn/claim-and-release/">26 Захват и возврат</a></li><li><a href="/ru/learn/capstone-what-runs-next/">27 Капстоун: что запускается дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/capstone-a-filter-on-the-board/index.html b/apps/website/public/ru/learn/capstone-a-filter-on-the-board/index.html
index 02622e87c1..d252d6b837 100644
--- a/apps/website/public/ru/learn/capstone-a-filter-on-the-board/index.html
+++ b/apps/website/public/ru/learn/capstone-a-filter-on-the-board/index.html
@@ -1122,5 +1122,6 @@ <h1>Итог: фильтр на плате</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Сложение</h3><p>Почему сложение медленное: перенос, который идёт вверх по слову, выделенная цепь в каждом слайсе, префиксные сети и деревья сохранения переноса.</p><ol><li><a href="/ru/learn/the-carry-chain/">01 Цепь переноса</a></li><li><a href="/ru/learn/prefix-adders/">02 Взгляд вперёд: префиксные сумматоры</a></li><li><a href="/ru/learn/carry-save/">03 Три числа сразу</a></li></ol></li><li><h3>Модуль 2 · Умножение</h3><p>Произведение — сумма сдвинутых копий; перекодирование Бута вдвое сокращает их, а слайс DSP48E1 делает остальное одним блоком.</p><ol><li><a href="/ru/learn/shift-and-add/">04 Сдвиг и сложение</a></li><li><a href="/ru/learn/booths-trick/">05 Приём Бута</a></li><li><a href="/ru/learn/the-dsp48e1-slice/">06 Слайс DSP48E1</a></li></ol></li><li><h3>Модуль 3 · Фиксированная точка</h3><p>Где стоит двоичная точка, что округление делает со значением и с его средним и что даёт каждый бит квантователя.</p><ol><li><a href="/ru/learn/q-formats/">07 Форматы Q</a></li><li><a href="/ru/learn/rounding-and-saturation/">08 Округление и насыщение</a></li><li><a href="/ru/learn/quantization-noise/">09 Шум квантования</a></li></ol></li><li><h3>Модуль 4 · Функции в железе</h3><p>Синус, косинус, угол и длина из сдвигов и сложений — и когда таблица оказывается выгоднее.</p><ol><li><a href="/ru/learn/rotating-by-shifts/">10 Поворот сдвигами</a></li><li><a href="/ru/learn/angle-and-magnitude/">11 Угол и модуль</a></li><li><a href="/ru/learn/tables-and-interpolation/">12 Таблицы и интерполяция</a></li></ol></li><li><h3>Модуль 5 · Сигналы и дискретизация</h3><p>Что дискретизация делает с частотой, как генератор строится из сумматора и что измеряет бин ДПФ.</p><ol><li><a href="/ru/learn/sampling-and-aliasing/">13 Дискретизация и наложение спектров</a></li><li><a href="/ru/learn/the-nco/">14 Генератор с числовым управлением</a></li><li><a href="/ru/learn/bins-of-a-spectrum/">15 Бины спектра</a></li></ol></li><li><h3>Модуль 6 · Фильтры FIR</h3><p>Скользящее среднее, расчёт окном sinc с целыми коэффициентами и раскладка отводов по слайсам DSP.</p><ol><li><a href="/ru/learn/the-moving-average/">16 Скользящее среднее</a></li><li><a href="/ru/learn/designing-taps/">17 Расчёт коэффициентов</a></li><li><a href="/ru/learn/folding-taps-onto-slices/">18 Раскладка отводов по слайсам</a></li></ol></li><li><h3>Модуль 7 · Многоскоростная обработка</h3><p>Понижение частоты дискретизации без заворота шума: децимация, фильтр CIC и полифазная форма.</p><ol><li><a href="/ru/learn/lowering-the-rate/">19 Понижение частоты</a></li><li><a href="/ru/learn/the-cic-filter/">20 Фильтр CIC</a></li><li><a href="/ru/learn/polyphase/">21 Полифазная форма</a></li></ol></li><li><h3>Модуль 8 · БПФ</h3><p>N log N вместо N^2: бабочки, бит-реверсный порядок входа и биты, которые добавляет каждый этап.</p><ol><li><a href="/ru/learn/the-butterfly/">22 Бабочка</a></li><li><a href="/ru/learn/twiddles-and-bit-reversal/">23 Поворачивающие множители и бит-реверс</a></li><li><a href="/ru/learn/growth-through-the-stages/">24 Рост через этапы</a></li></ol></li><li><h3>Модуль 9 · На стенде</h3><p>От арифметики к плате: коррелятор из спеки модема, бюджет слайсов и тайминга и итоговый фильтр.</p><ol><li><a href="/ru/learn/a-correlator-finds-the-frame/">25 Коррелятор находит кадр</a></li><li><a href="/ru/learn/does-it-fit-and-close-timing/">26 Помещается ли и сходится ли тайминг?</a></li><li class="on" aria-current="page"><a href="/ru/learn/capstone-a-filter-on-the-board/">27 Итог: фильтр на плате</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/capstone-diff/index.html b/apps/website/public/ru/learn/capstone-diff/index.html
index 51e8b17a3e..5161309c72 100644
--- a/apps/website/public/ru/learn/capstone-diff/index.html
+++ b/apps/website/public/ru/learn/capstone-diff/index.html
@@ -434,5 +434,6 @@ <h1>Капстоун: поменяй и сравни</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое тактовый сигнал</h3><p>Один фронт, один мир: всё, что делит фронт тактового сигнала, делит и мир; период и джиттер настоящего фронта и место, где тактовый сигнал входит в плату.</p><ol><li><a href="/ru/learn/one-edge-one-world/">01 Один фронт, один мир</a></li><li><a href="/ru/learn/period-and-jitter/">02 Период, фронт и джиттер</a></li><li><a href="/ru/learn/where-the-clock-enters/">03 Где входит тактовый сигнал</a></li></ol></li><li><h3>Модуль 2 · Деревья тактового сигнала</h3><p>Перекос и задержка прохождения, глобальная сеть буферов и ловушка стробирования клока логикой.</p><ol><li><a href="/ru/learn/skew-and-insertion/">04 Перекос и задержка прохождения</a></li><li><a href="/ru/learn/global-buffers/">05 Глобальный буфер</a></li><li><a href="/ru/learn/gated-clock-trap/">06 Ловушка стробированного клока</a></li></ol></li><li><h3>Модуль 3 · PLL и MMCM</h3><p>Умножить и поделить один тактовый сигнал в другой, сдвинуть его фазу шагами VCO и какие тактовые сигналы анализатор считает родственными.</p><ol><li><a href="/ru/learn/multiply-and-divide/">07 Умножить и поделить</a></li><li><a href="/ru/learn/phase-and-lock/">08 Фаза и захват</a></li><li><a href="/ru/learn/related-or-not/">09 Родственные или чужие</a></li></ol></li><li><h3>Модуль 4 · Сбросы</h3><p>Ассертировать асинхронно, снимать синхронно: три вида сброса, конвейер снятия и дерево, которое растит сброс.</p><ol><li><a href="/ru/learn/sync-or-async/">10 Синхронный или асинхронный</a></li><li><a href="/ru/learn/reset-release/">11 Конвейер снятия</a></li><li><a href="/ru/learn/reset-fanout/">12 Разветвление сброса</a></li></ol></li><li><h3>Модуль 5 · Метастабильность</h3><p>Окно setup-hold, среднее время между отказами в целочисленной арифметике и два триггера, которые это чинят.</p><ol><li><a href="/ru/learn/setup-hold-window/">13 Окно setup-hold</a></li><li><a href="/ru/learn/mtbf/">14 Среднее время между отказами</a></li><li><a href="/ru/learn/two-flop-sync/">15 Двухтриггерный синхронизатор</a></li></ol></li><li><h3>Модуль 6 · Пересечение многих бит</h3><p>Почему двоичная шина рвётся, почему код Грея нет, и рукопожатие, переносящее импульс между мирами.</p><ol><li><a href="/ru/learn/why-binary-fails/">16 Почему двоичный код рвётся</a></li><li><a href="/ru/learn/gray-code/">17 Код Грея</a></li><li><a href="/ru/learn/handshakes-and-pulses/">18 Рукопожатия и импульсы</a></li></ol></li><li><h3>Модуль 7 · Асинхронный FIFO</h3><p>Указатели, флаги и глубина: буфер, переносящий поток между двумя тактовыми сигналами.</p><ol><li><a href="/ru/learn/pointers-and-flags/">19 Указатели и флаги</a></li><li><a href="/ru/learn/depth-sizing/">20 Выбор глубины</a></li><li><a href="/ru/learn/fifo-testbench/">21 Тестбенч FIFO</a></li></ol></li><li><h3>Модуль 8 · Ограничения</h3><p>Строки, сообщающие анализатору, что такое тактовый сигнал, какие пути не проверять и чему должны соответствовать выводы.</p><ol><li><a href="/ru/learn/create-clock/">22 Строка create_clock</a></li><li><a href="/ru/learn/false-path-and-max-delay/">23 False path и max delay</a></li><li><a href="/ru/learn/io-timing/">24 Тайминг выводов</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Отчёт CDC, одно пересечение, захваченное на триггерах, и дифф битстрима, замыкающий курс.</p><ol><li><a href="/ru/learn/cdc-report/">25 Отчёт CDC</a></li><li><a href="/ru/learn/capture-a-crossing/">26 Захватить пересечение</a></li><li class="on" aria-current="page"><a href="/ru/learn/capstone-diff/">27 Капстоун: поменяй и сравни</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/capstone-what-runs-next/index.html b/apps/website/public/ru/learn/capstone-what-runs-next/index.html
index dc60036454..a315adbfe0 100644
--- a/apps/website/public/ru/learn/capstone-what-runs-next/index.html
+++ b/apps/website/public/ru/learn/capstone-what-runs-next/index.html
@@ -300,5 +300,6 @@ <h1>Капстоун: что запускается дальше</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое шина</h3><p>Зачем вообще существует шина: разговор по проводам, с кадрами и адресами, и кому позволено говорить.</p><ol><li><a href="/ru/learn/wires-that-talk/">01 Провода, которые говорят</a></li><li><a href="/ru/learn/framing-a-conversation/">02 Кадрирование разговора</a></li><li><a href="/ru/learn/master-and-servant/">03 Хозяин и слуга</a></li></ol></li><li><h3>Модуль 2 · Разговор по UART</h3><p>Двухпроводная шина без такта: кадр, делитель, задающий скорость, и статус, который опрашивает драйвер.</p><ol><li><a href="/ru/learn/the-frame/">04 Кадр: старт, данные, стоп</a></li><li><a href="/ru/learn/baud-and-the-divisor/">05 Бод и делитель</a></li><li><a href="/ru/learn/status-and-fifos/">06 Статус и FIFO</a></li></ol></li><li><h3>Модуль 3 · SPI под тактом</h3><p>Разговор под тактом: четыре режима, лестница предделителей и выбор кристалла на каждого подчинённого.</p><ol><li><a href="/ru/learn/modes-cpol-cpha/">07 Режимы: CPOL и CPHA</a></li><li><a href="/ru/learn/the-prescaler-ladder/">08 Лестница предделителей</a></li><li><a href="/ru/learn/chip-select-and-widths/">09 Выбор кристалла и ширины</a></li></ol></li><li><h3>Модуль 4 · Регистровая шина APB</h3><p>Шина регистров: PSEL и PENABLE, строобы и ожидания, и сколько адресных бит стоит количество периферии.</p><ol><li><a href="/ru/learn/psel-and-penable/">10 PSEL и PENABLE</a></li><li><a href="/ru/learn/strobes-and-waits/">11 Строобы и ожидания</a></li><li><a href="/ru/learn/address-decode/">12 Дешифрация адреса</a></li></ol></li><li><h3>Модуль 5 · Пять каналов AXI4</h3><p>Пять каналов: адрес, данные и отклик в обе стороны, lite или full, пакеты и идентификаторы.</p><ol><li><a href="/ru/learn/five-channels/">13 Пять каналов</a></li><li><a href="/ru/learn/lite-or-full/">14 Lite или full</a></li><li><a href="/ru/learn/bursts-and-ids/">15 Пакеты и идентификаторы</a></li></ol></li><li><h3>Модуль 6 · Память</h3><p>Что находится на дальнем конце каждой шины: карты памяти, типы портов и задержка, которую должно покрывать ожидание.</p><ol><li><a href="/ru/learn/memory-maps/">16 Карты памяти</a></li><li><a href="/ru/learn/ports-and-kinds/">17 Порты и типы</a></li><li><a href="/ru/learn/latency/">18 Задержка</a></li></ol></li><li><h3>Модуль 7 · Мосты</h3><p>Почему в дизайне появляется больше одной шины и пакетный мост, переносящий работу между ними.</p><ol><li><a href="/ru/learn/why-bridges/">19 Зачем мосты</a></li><li><a href="/ru/learn/the-packet-bridge/">20 Пакетный мост</a></li><li><a href="/ru/learn/everything-in-one-bench/">21 Всё в одном стенде</a></li></ol></li><li><h3>Модуль 8 · Ethernet: кадры и тайминг</h3><p>Кадры, контрольная сумма кадра, тайминг RGMII и заранее зарегистрированные шаги настоящего запуска.</p><ol><li><a href="/ru/learn/frames-and-the-fcs/">22 Кадры и контрольная сумма</a></li><li><a href="/ru/learn/rgmii-timing/">23 Тайминг RGMII</a></li><li><a href="/ru/learn/bring-up-on-the-bench/">24 Запуск на стенде</a></li></ol></li><li><h3>Модуль 9 · Стенд</h3><p>Дисциплина, оберегающая настоящее железо: кто держит IO, как взять и вернуть захват, и что запускается дальше.</p><ol><li><a href="/ru/learn/who-holds-the-io/">25 Кто держит IO</a></li><li><a href="/ru/learn/claim-and-release/">26 Захват и возврат</a></li><li class="on" aria-current="page"><a href="/ru/learn/capstone-what-runs-next/">27 Капстоун: что запускается дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/capstone/index.html b/apps/website/public/ru/learn/capstone/index.html
index f5fbd1f321..b157c57da3 100644
--- a/apps/website/public/ru/learn/capstone/index.html
+++ b/apps/website/public/ru/learn/capstone/index.html
@@ -322,5 +322,6 @@ <h1>Итог: переугадайте yosys</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Чип</h3><p>Что такое FPGA, какой чип у нас и как его выводы встречаются с платой.</p><ol><li><a href="/ru/learn/fpga-is-a-grid/">01 FPGA — это сетка, которую вы настраиваете</a></li><li><a href="/ru/learn/which-chip/">02 Какой это чип?</a></li><li><a href="/ru/learn/pins-and-package/">03 Выводы, шарики и корпус</a></li></ol></li><li><h3>Модуль 2 · Числа в железе</h3><p>Биты, триты и форматы чисел, и сколько стоит арифметика в логике.</p><ol><li><a href="/ru/learn/bits-and-trits/">04 Биты, триты и GF16</a></li><li><a href="/ru/learn/what-arithmetic-costs/">05 Сколько стоит арифметика</a></li><li><a href="/ru/learn/how-much-fits/">06 Сколько помещается на чип</a></li></ol></li><li><h3>Модуль 3 · Ваша программа на t27</h3><p>Напишите спеку, проверьте её и увидьте, почему «компилируется» не значит «верно».</p><ol><li><a href="/ru/learn/hello-t27/">07 Привет, t27</a></li><li><a href="/ru/learn/tests-are-the-spec/">08 Тесты — часть спеки</a></li><li><a href="/ru/learn/compiles-is-not-correct/">09 Компилируется — не значит верно</a></li></ol></li><li><h3>Модуль 4 · Внутри t27c</h3><p>Как компилятор читает спеку и что он пишет, включая нативный t27b.</p><ol><li><a href="/ru/learn/compiler-stages/">10 Стадии t27c</a></li><li><a href="/ru/learn/t27c-on-a-real-machine/">11 t27c на настоящей машине</a></li><li><a href="/ru/learn/seven-backends-and-t27b/">12 Семь бэкендов и t27b</a></li></ol></li><li><h3>Модуль 5 · От спеки к железу</h3><p>Verilog, который пишет t27c, ячейки, в которые он превращается, и что делает одна LUT.</p><ol><li><a href="/ru/learn/spec-to-verilog/">13 От спеки к Verilog</a></li><li><a href="/ru/learn/count-the-cells/">14 Посчитайте ячейки</a></li><li><a href="/ru/learn/one-lut-up-close/">15 Одна LUT вблизи</a></li></ol></li><li><h3>Модуль 6 · Читаем синтез</h3><p>Что yosys сообщает о вашем дизайне и какие предупреждения важны.</p><ol><li><a href="/ru/learn/synthesis-warnings/">16 Важные предупреждения синтеза</a></li><li><a href="/ru/learn/state-machines/">17 Конечные автоматы</a></li><li><a href="/ru/learn/where-the-luts-go/">18 Куда уходят LUT</a></li></ol></li><li><h3>Модуль 7 · Размещение, трассировка, тайминг</h3><p>Где ячейки оказываются на кристалле и успевает ли тактовая частота.</p><ol><li><a href="/ru/learn/placement/">19 Размещение: где оказываются ячейки</a></li><li><a href="/ru/learn/logic-depth/">20 Глубина логики</a></li><li><a href="/ru/learn/timing-slack/">21 Тайминг и запас</a></li></ol></li><li><h3>Модуль 8 · Битстрим</h3><p>Как оттрассированный дизайн становится битами, которые загружает чип, без Vivado.</p><ol><li><a href="/ru/learn/fasm-to-frames/">22 От FASM к фреймам</a></li><li><a href="/ru/learn/a-build-receipt/">23 Сборка по пунктам</a></li><li><a href="/ru/learn/read-a-bitstream/">24 Читаем битстрим</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Спросите чип, кто он, загрузите биты и проверьте их.</p><ol><li><a href="/ru/learn/jtag-asks-the-chip/">25 JTAG спрашивает чип</a></li><li><a href="/ru/learn/same-bits-on-the-board/">26 Те же биты — на плате</a></li><li class="on" aria-current="page"><a href="/ru/learn/capstone/">27 Итог: переугадайте yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/capture-a-crossing/index.html b/apps/website/public/ru/learn/capture-a-crossing/index.html
index 874a83057a..664b687fed 100644
--- a/apps/website/public/ru/learn/capture-a-crossing/index.html
+++ b/apps/website/public/ru/learn/capture-a-crossing/index.html
@@ -110,5 +110,6 @@ <h1>Захватить пересечение</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое тактовый сигнал</h3><p>Один фронт, один мир: всё, что делит фронт тактового сигнала, делит и мир; период и джиттер настоящего фронта и место, где тактовый сигнал входит в плату.</p><ol><li><a href="/ru/learn/one-edge-one-world/">01 Один фронт, один мир</a></li><li><a href="/ru/learn/period-and-jitter/">02 Период, фронт и джиттер</a></li><li><a href="/ru/learn/where-the-clock-enters/">03 Где входит тактовый сигнал</a></li></ol></li><li><h3>Модуль 2 · Деревья тактового сигнала</h3><p>Перекос и задержка прохождения, глобальная сеть буферов и ловушка стробирования клока логикой.</p><ol><li><a href="/ru/learn/skew-and-insertion/">04 Перекос и задержка прохождения</a></li><li><a href="/ru/learn/global-buffers/">05 Глобальный буфер</a></li><li><a href="/ru/learn/gated-clock-trap/">06 Ловушка стробированного клока</a></li></ol></li><li><h3>Модуль 3 · PLL и MMCM</h3><p>Умножить и поделить один тактовый сигнал в другой, сдвинуть его фазу шагами VCO и какие тактовые сигналы анализатор считает родственными.</p><ol><li><a href="/ru/learn/multiply-and-divide/">07 Умножить и поделить</a></li><li><a href="/ru/learn/phase-and-lock/">08 Фаза и захват</a></li><li><a href="/ru/learn/related-or-not/">09 Родственные или чужие</a></li></ol></li><li><h3>Модуль 4 · Сбросы</h3><p>Ассертировать асинхронно, снимать синхронно: три вида сброса, конвейер снятия и дерево, которое растит сброс.</p><ol><li><a href="/ru/learn/sync-or-async/">10 Синхронный или асинхронный</a></li><li><a href="/ru/learn/reset-release/">11 Конвейер снятия</a></li><li><a href="/ru/learn/reset-fanout/">12 Разветвление сброса</a></li></ol></li><li><h3>Модуль 5 · Метастабильность</h3><p>Окно setup-hold, среднее время между отказами в целочисленной арифметике и два триггера, которые это чинят.</p><ol><li><a href="/ru/learn/setup-hold-window/">13 Окно setup-hold</a></li><li><a href="/ru/learn/mtbf/">14 Среднее время между отказами</a></li><li><a href="/ru/learn/two-flop-sync/">15 Двухтриггерный синхронизатор</a></li></ol></li><li><h3>Модуль 6 · Пересечение многих бит</h3><p>Почему двоичная шина рвётся, почему код Грея нет, и рукопожатие, переносящее импульс между мирами.</p><ol><li><a href="/ru/learn/why-binary-fails/">16 Почему двоичный код рвётся</a></li><li><a href="/ru/learn/gray-code/">17 Код Грея</a></li><li><a href="/ru/learn/handshakes-and-pulses/">18 Рукопожатия и импульсы</a></li></ol></li><li><h3>Модуль 7 · Асинхронный FIFO</h3><p>Указатели, флаги и глубина: буфер, переносящий поток между двумя тактовыми сигналами.</p><ol><li><a href="/ru/learn/pointers-and-flags/">19 Указатели и флаги</a></li><li><a href="/ru/learn/depth-sizing/">20 Выбор глубины</a></li><li><a href="/ru/learn/fifo-testbench/">21 Тестбенч FIFO</a></li></ol></li><li><h3>Модуль 8 · Ограничения</h3><p>Строки, сообщающие анализатору, что такое тактовый сигнал, какие пути не проверять и чему должны соответствовать выводы.</p><ol><li><a href="/ru/learn/create-clock/">22 Строка create_clock</a></li><li><a href="/ru/learn/false-path-and-max-delay/">23 False path и max delay</a></li><li><a href="/ru/learn/io-timing/">24 Тайминг выводов</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Отчёт CDC, одно пересечение, захваченное на триггерах, и дифф битстрима, замыкающий курс.</p><ol><li><a href="/ru/learn/cdc-report/">25 Отчёт CDC</a></li><li class="on" aria-current="page"><a href="/ru/learn/capture-a-crossing/">26 Захватить пересечение</a></li><li><a href="/ru/learn/capstone-diff/">27 Капстоун: поменяй и сравни</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/carry-save/index.html b/apps/website/public/ru/learn/carry-save/index.html
index dbf55741c6..3fb7ee5be6 100644
--- a/apps/website/public/ru/learn/carry-save/index.html
+++ b/apps/website/public/ru/learn/carry-save/index.html
@@ -554,5 +554,6 @@ <h1>Три числа сразу</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Сложение</h3><p>Почему сложение медленное: перенос, который идёт вверх по слову, выделенная цепь в каждом слайсе, префиксные сети и деревья сохранения переноса.</p><ol><li><a href="/ru/learn/the-carry-chain/">01 Цепь переноса</a></li><li><a href="/ru/learn/prefix-adders/">02 Взгляд вперёд: префиксные сумматоры</a></li><li class="on" aria-current="page"><a href="/ru/learn/carry-save/">03 Три числа сразу</a></li></ol></li><li><h3>Модуль 2 · Умножение</h3><p>Произведение — сумма сдвинутых копий; перекодирование Бута вдвое сокращает их, а слайс DSP48E1 делает остальное одним блоком.</p><ol><li><a href="/ru/learn/shift-and-add/">04 Сдвиг и сложение</a></li><li><a href="/ru/learn/booths-trick/">05 Приём Бута</a></li><li><a href="/ru/learn/the-dsp48e1-slice/">06 Слайс DSP48E1</a></li></ol></li><li><h3>Модуль 3 · Фиксированная точка</h3><p>Где стоит двоичная точка, что округление делает со значением и с его средним и что даёт каждый бит квантователя.</p><ol><li><a href="/ru/learn/q-formats/">07 Форматы Q</a></li><li><a href="/ru/learn/rounding-and-saturation/">08 Округление и насыщение</a></li><li><a href="/ru/learn/quantization-noise/">09 Шум квантования</a></li></ol></li><li><h3>Модуль 4 · Функции в железе</h3><p>Синус, косинус, угол и длина из сдвигов и сложений — и когда таблица оказывается выгоднее.</p><ol><li><a href="/ru/learn/rotating-by-shifts/">10 Поворот сдвигами</a></li><li><a href="/ru/learn/angle-and-magnitude/">11 Угол и модуль</a></li><li><a href="/ru/learn/tables-and-interpolation/">12 Таблицы и интерполяция</a></li></ol></li><li><h3>Модуль 5 · Сигналы и дискретизация</h3><p>Что дискретизация делает с частотой, как генератор строится из сумматора и что измеряет бин ДПФ.</p><ol><li><a href="/ru/learn/sampling-and-aliasing/">13 Дискретизация и наложение спектров</a></li><li><a href="/ru/learn/the-nco/">14 Генератор с числовым управлением</a></li><li><a href="/ru/learn/bins-of-a-spectrum/">15 Бины спектра</a></li></ol></li><li><h3>Модуль 6 · Фильтры FIR</h3><p>Скользящее среднее, расчёт окном sinc с целыми коэффициентами и раскладка отводов по слайсам DSP.</p><ol><li><a href="/ru/learn/the-moving-average/">16 Скользящее среднее</a></li><li><a href="/ru/learn/designing-taps/">17 Расчёт коэффициентов</a></li><li><a href="/ru/learn/folding-taps-onto-slices/">18 Раскладка отводов по слайсам</a></li></ol></li><li><h3>Модуль 7 · Многоскоростная обработка</h3><p>Понижение частоты дискретизации без заворота шума: децимация, фильтр CIC и полифазная форма.</p><ol><li><a href="/ru/learn/lowering-the-rate/">19 Понижение частоты</a></li><li><a href="/ru/learn/the-cic-filter/">20 Фильтр CIC</a></li><li><a href="/ru/learn/polyphase/">21 Полифазная форма</a></li></ol></li><li><h3>Модуль 8 · БПФ</h3><p>N log N вместо N^2: бабочки, бит-реверсный порядок входа и биты, которые добавляет каждый этап.</p><ol><li><a href="/ru/learn/the-butterfly/">22 Бабочка</a></li><li><a href="/ru/learn/twiddles-and-bit-reversal/">23 Поворачивающие множители и бит-реверс</a></li><li><a href="/ru/learn/growth-through-the-stages/">24 Рост через этапы</a></li></ol></li><li><h3>Модуль 9 · На стенде</h3><p>От арифметики к плате: коррелятор из спеки модема, бюджет слайсов и тайминга и итоговый фильтр.</p><ol><li><a href="/ru/learn/a-correlator-finds-the-frame/">25 Коррелятор находит кадр</a></li><li><a href="/ru/learn/does-it-fit-and-close-timing/">26 Помещается ли и сходится ли тайминг?</a></li><li><a href="/ru/learn/capstone-a-filter-on-the-board/">27 Итог: фильтр на плате</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/cdc-report/index.html b/apps/website/public/ru/learn/cdc-report/index.html
index 97e8d26082..a4ced7c45f 100644
--- a/apps/website/public/ru/learn/cdc-report/index.html
+++ b/apps/website/public/ru/learn/cdc-report/index.html
@@ -448,5 +448,6 @@ <h1>Отчёт CDC</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое тактовый сигнал</h3><p>Один фронт, один мир: всё, что делит фронт тактового сигнала, делит и мир; период и джиттер настоящего фронта и место, где тактовый сигнал входит в плату.</p><ol><li><a href="/ru/learn/one-edge-one-world/">01 Один фронт, один мир</a></li><li><a href="/ru/learn/period-and-jitter/">02 Период, фронт и джиттер</a></li><li><a href="/ru/learn/where-the-clock-enters/">03 Где входит тактовый сигнал</a></li></ol></li><li><h3>Модуль 2 · Деревья тактового сигнала</h3><p>Перекос и задержка прохождения, глобальная сеть буферов и ловушка стробирования клока логикой.</p><ol><li><a href="/ru/learn/skew-and-insertion/">04 Перекос и задержка прохождения</a></li><li><a href="/ru/learn/global-buffers/">05 Глобальный буфер</a></li><li><a href="/ru/learn/gated-clock-trap/">06 Ловушка стробированного клока</a></li></ol></li><li><h3>Модуль 3 · PLL и MMCM</h3><p>Умножить и поделить один тактовый сигнал в другой, сдвинуть его фазу шагами VCO и какие тактовые сигналы анализатор считает родственными.</p><ol><li><a href="/ru/learn/multiply-and-divide/">07 Умножить и поделить</a></li><li><a href="/ru/learn/phase-and-lock/">08 Фаза и захват</a></li><li><a href="/ru/learn/related-or-not/">09 Родственные или чужие</a></li></ol></li><li><h3>Модуль 4 · Сбросы</h3><p>Ассертировать асинхронно, снимать синхронно: три вида сброса, конвейер снятия и дерево, которое растит сброс.</p><ol><li><a href="/ru/learn/sync-or-async/">10 Синхронный или асинхронный</a></li><li><a href="/ru/learn/reset-release/">11 Конвейер снятия</a></li><li><a href="/ru/learn/reset-fanout/">12 Разветвление сброса</a></li></ol></li><li><h3>Модуль 5 · Метастабильность</h3><p>Окно setup-hold, среднее время между отказами в целочисленной арифметике и два триггера, которые это чинят.</p><ol><li><a href="/ru/learn/setup-hold-window/">13 Окно setup-hold</a></li><li><a href="/ru/learn/mtbf/">14 Среднее время между отказами</a></li><li><a href="/ru/learn/two-flop-sync/">15 Двухтриггерный синхронизатор</a></li></ol></li><li><h3>Модуль 6 · Пересечение многих бит</h3><p>Почему двоичная шина рвётся, почему код Грея нет, и рукопожатие, переносящее импульс между мирами.</p><ol><li><a href="/ru/learn/why-binary-fails/">16 Почему двоичный код рвётся</a></li><li><a href="/ru/learn/gray-code/">17 Код Грея</a></li><li><a href="/ru/learn/handshakes-and-pulses/">18 Рукопожатия и импульсы</a></li></ol></li><li><h3>Модуль 7 · Асинхронный FIFO</h3><p>Указатели, флаги и глубина: буфер, переносящий поток между двумя тактовыми сигналами.</p><ol><li><a href="/ru/learn/pointers-and-flags/">19 Указатели и флаги</a></li><li><a href="/ru/learn/depth-sizing/">20 Выбор глубины</a></li><li><a href="/ru/learn/fifo-testbench/">21 Тестбенч FIFO</a></li></ol></li><li><h3>Модуль 8 · Ограничения</h3><p>Строки, сообщающие анализатору, что такое тактовый сигнал, какие пути не проверять и чему должны соответствовать выводы.</p><ol><li><a href="/ru/learn/create-clock/">22 Строка create_clock</a></li><li><a href="/ru/learn/false-path-and-max-delay/">23 False path и max delay</a></li><li><a href="/ru/learn/io-timing/">24 Тайминг выводов</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Отчёт CDC, одно пересечение, захваченное на триггерах, и дифф битстрима, замыкающий курс.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/cdc-report/">25 Отчёт CDC</a></li><li><a href="/ru/learn/capture-a-crossing/">26 Захватить пересечение</a></li><li><a href="/ru/learn/capstone-diff/">27 Капстоун: поменяй и сравни</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/chip-select-and-widths/index.html b/apps/website/public/ru/learn/chip-select-and-widths/index.html
index 1d247819fc..934314737d 100644
--- a/apps/website/public/ru/learn/chip-select-and-widths/index.html
+++ b/apps/website/public/ru/learn/chip-select-and-widths/index.html
@@ -209,5 +209,6 @@ <h1>Выбор кристалла и ширины</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое шина</h3><p>Зачем вообще существует шина: разговор по проводам, с кадрами и адресами, и кому позволено говорить.</p><ol><li><a href="/ru/learn/wires-that-talk/">01 Провода, которые говорят</a></li><li><a href="/ru/learn/framing-a-conversation/">02 Кадрирование разговора</a></li><li><a href="/ru/learn/master-and-servant/">03 Хозяин и слуга</a></li></ol></li><li><h3>Модуль 2 · Разговор по UART</h3><p>Двухпроводная шина без такта: кадр, делитель, задающий скорость, и статус, который опрашивает драйвер.</p><ol><li><a href="/ru/learn/the-frame/">04 Кадр: старт, данные, стоп</a></li><li><a href="/ru/learn/baud-and-the-divisor/">05 Бод и делитель</a></li><li><a href="/ru/learn/status-and-fifos/">06 Статус и FIFO</a></li></ol></li><li><h3>Модуль 3 · SPI под тактом</h3><p>Разговор под тактом: четыре режима, лестница предделителей и выбор кристалла на каждого подчинённого.</p><ol><li><a href="/ru/learn/modes-cpol-cpha/">07 Режимы: CPOL и CPHA</a></li><li><a href="/ru/learn/the-prescaler-ladder/">08 Лестница предделителей</a></li><li class="on" aria-current="page"><a href="/ru/learn/chip-select-and-widths/">09 Выбор кристалла и ширины</a></li></ol></li><li><h3>Модуль 4 · Регистровая шина APB</h3><p>Шина регистров: PSEL и PENABLE, строобы и ожидания, и сколько адресных бит стоит количество периферии.</p><ol><li><a href="/ru/learn/psel-and-penable/">10 PSEL и PENABLE</a></li><li><a href="/ru/learn/strobes-and-waits/">11 Строобы и ожидания</a></li><li><a href="/ru/learn/address-decode/">12 Дешифрация адреса</a></li></ol></li><li><h3>Модуль 5 · Пять каналов AXI4</h3><p>Пять каналов: адрес, данные и отклик в обе стороны, lite или full, пакеты и идентификаторы.</p><ol><li><a href="/ru/learn/five-channels/">13 Пять каналов</a></li><li><a href="/ru/learn/lite-or-full/">14 Lite или full</a></li><li><a href="/ru/learn/bursts-and-ids/">15 Пакеты и идентификаторы</a></li></ol></li><li><h3>Модуль 6 · Память</h3><p>Что находится на дальнем конце каждой шины: карты памяти, типы портов и задержка, которую должно покрывать ожидание.</p><ol><li><a href="/ru/learn/memory-maps/">16 Карты памяти</a></li><li><a href="/ru/learn/ports-and-kinds/">17 Порты и типы</a></li><li><a href="/ru/learn/latency/">18 Задержка</a></li></ol></li><li><h3>Модуль 7 · Мосты</h3><p>Почему в дизайне появляется больше одной шины и пакетный мост, переносящий работу между ними.</p><ol><li><a href="/ru/learn/why-bridges/">19 Зачем мосты</a></li><li><a href="/ru/learn/the-packet-bridge/">20 Пакетный мост</a></li><li><a href="/ru/learn/everything-in-one-bench/">21 Всё в одном стенде</a></li></ol></li><li><h3>Модуль 8 · Ethernet: кадры и тайминг</h3><p>Кадры, контрольная сумма кадра, тайминг RGMII и заранее зарегистрированные шаги настоящего запуска.</p><ol><li><a href="/ru/learn/frames-and-the-fcs/">22 Кадры и контрольная сумма</a></li><li><a href="/ru/learn/rgmii-timing/">23 Тайминг RGMII</a></li><li><a href="/ru/learn/bring-up-on-the-bench/">24 Запуск на стенде</a></li></ol></li><li><h3>Модуль 9 · Стенд</h3><p>Дисциплина, оберегающая настоящее железо: кто держит IO, как взять и вернуть захват, и что запускается дальше.</p><ol><li><a href="/ru/learn/who-holds-the-io/">25 Кто держит IO</a></li><li><a href="/ru/learn/claim-and-release/">26 Захват и возврат</a></li><li><a href="/ru/learn/capstone-what-runs-next/">27 Капстоун: что запускается дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/claim-and-release/index.html b/apps/website/public/ru/learn/claim-and-release/index.html
index 20ebcbb3d3..7b5e817704 100644
--- a/apps/website/public/ru/learn/claim-and-release/index.html
+++ b/apps/website/public/ru/learn/claim-and-release/index.html
@@ -109,5 +109,6 @@ <h1>Захват и возврат</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое шина</h3><p>Зачем вообще существует шина: разговор по проводам, с кадрами и адресами, и кому позволено говорить.</p><ol><li><a href="/ru/learn/wires-that-talk/">01 Провода, которые говорят</a></li><li><a href="/ru/learn/framing-a-conversation/">02 Кадрирование разговора</a></li><li><a href="/ru/learn/master-and-servant/">03 Хозяин и слуга</a></li></ol></li><li><h3>Модуль 2 · Разговор по UART</h3><p>Двухпроводная шина без такта: кадр, делитель, задающий скорость, и статус, который опрашивает драйвер.</p><ol><li><a href="/ru/learn/the-frame/">04 Кадр: старт, данные, стоп</a></li><li><a href="/ru/learn/baud-and-the-divisor/">05 Бод и делитель</a></li><li><a href="/ru/learn/status-and-fifos/">06 Статус и FIFO</a></li></ol></li><li><h3>Модуль 3 · SPI под тактом</h3><p>Разговор под тактом: четыре режима, лестница предделителей и выбор кристалла на каждого подчинённого.</p><ol><li><a href="/ru/learn/modes-cpol-cpha/">07 Режимы: CPOL и CPHA</a></li><li><a href="/ru/learn/the-prescaler-ladder/">08 Лестница предделителей</a></li><li><a href="/ru/learn/chip-select-and-widths/">09 Выбор кристалла и ширины</a></li></ol></li><li><h3>Модуль 4 · Регистровая шина APB</h3><p>Шина регистров: PSEL и PENABLE, строобы и ожидания, и сколько адресных бит стоит количество периферии.</p><ol><li><a href="/ru/learn/psel-and-penable/">10 PSEL и PENABLE</a></li><li><a href="/ru/learn/strobes-and-waits/">11 Строобы и ожидания</a></li><li><a href="/ru/learn/address-decode/">12 Дешифрация адреса</a></li></ol></li><li><h3>Модуль 5 · Пять каналов AXI4</h3><p>Пять каналов: адрес, данные и отклик в обе стороны, lite или full, пакеты и идентификаторы.</p><ol><li><a href="/ru/learn/five-channels/">13 Пять каналов</a></li><li><a href="/ru/learn/lite-or-full/">14 Lite или full</a></li><li><a href="/ru/learn/bursts-and-ids/">15 Пакеты и идентификаторы</a></li></ol></li><li><h3>Модуль 6 · Память</h3><p>Что находится на дальнем конце каждой шины: карты памяти, типы портов и задержка, которую должно покрывать ожидание.</p><ol><li><a href="/ru/learn/memory-maps/">16 Карты памяти</a></li><li><a href="/ru/learn/ports-and-kinds/">17 Порты и типы</a></li><li><a href="/ru/learn/latency/">18 Задержка</a></li></ol></li><li><h3>Модуль 7 · Мосты</h3><p>Почему в дизайне появляется больше одной шины и пакетный мост, переносящий работу между ними.</p><ol><li><a href="/ru/learn/why-bridges/">19 Зачем мосты</a></li><li><a href="/ru/learn/the-packet-bridge/">20 Пакетный мост</a></li><li><a href="/ru/learn/everything-in-one-bench/">21 Всё в одном стенде</a></li></ol></li><li><h3>Модуль 8 · Ethernet: кадры и тайминг</h3><p>Кадры, контрольная сумма кадра, тайминг RGMII и заранее зарегистрированные шаги настоящего запуска.</p><ol><li><a href="/ru/learn/frames-and-the-fcs/">22 Кадры и контрольная сумма</a></li><li><a href="/ru/learn/rgmii-timing/">23 Тайминг RGMII</a></li><li><a href="/ru/learn/bring-up-on-the-bench/">24 Запуск на стенде</a></li></ol></li><li><h3>Модуль 9 · Стенд</h3><p>Дисциплина, оберегающая настоящее железо: кто держит IO, как взять и вернуть захват, и что запускается дальше.</p><ol><li><a href="/ru/learn/who-holds-the-io/">25 Кто держит IO</a></li><li class="on" aria-current="page"><a href="/ru/learn/claim-and-release/">26 Захват и возврат</a></li><li><a href="/ru/learn/capstone-what-runs-next/">27 Капстоун: что запускается дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/clocks-and-cdc/index.html b/apps/website/public/ru/learn/clocks-and-cdc/index.html
index 08f4fa6df4..b2bc6e7baf 100644
--- a/apps/website/public/ru/learn/clocks-and-cdc/index.html
+++ b/apps/website/public/ru/learn/clocks-and-cdc/index.html
@@ -67,5 +67,6 @@ <h1>Тактирование, сброс и пересечения тактов
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое тактовый сигнал</h3><p>Один фронт, один мир: всё, что делит фронт тактового сигнала, делит и мир; период и джиттер настоящего фронта и место, где тактовый сигнал входит в плату.</p><ol><li><a href="/ru/learn/one-edge-one-world/">01 Один фронт, один мир</a></li><li><a href="/ru/learn/period-and-jitter/">02 Период, фронт и джиттер</a></li><li><a href="/ru/learn/where-the-clock-enters/">03 Где входит тактовый сигнал</a></li></ol></li><li><h3>Модуль 2 · Деревья тактового сигнала</h3><p>Перекос и задержка прохождения, глобальная сеть буферов и ловушка стробирования клока логикой.</p><ol><li><a href="/ru/learn/skew-and-insertion/">04 Перекос и задержка прохождения</a></li><li><a href="/ru/learn/global-buffers/">05 Глобальный буфер</a></li><li><a href="/ru/learn/gated-clock-trap/">06 Ловушка стробированного клока</a></li></ol></li><li><h3>Модуль 3 · PLL и MMCM</h3><p>Умножить и поделить один тактовый сигнал в другой, сдвинуть его фазу шагами VCO и какие тактовые сигналы анализатор считает родственными.</p><ol><li><a href="/ru/learn/multiply-and-divide/">07 Умножить и поделить</a></li><li><a href="/ru/learn/phase-and-lock/">08 Фаза и захват</a></li><li><a href="/ru/learn/related-or-not/">09 Родственные или чужие</a></li></ol></li><li><h3>Модуль 4 · Сбросы</h3><p>Ассертировать асинхронно, снимать синхронно: три вида сброса, конвейер снятия и дерево, которое растит сброс.</p><ol><li><a href="/ru/learn/sync-or-async/">10 Синхронный или асинхронный</a></li><li><a href="/ru/learn/reset-release/">11 Конвейер снятия</a></li><li><a href="/ru/learn/reset-fanout/">12 Разветвление сброса</a></li></ol></li><li><h3>Модуль 5 · Метастабильность</h3><p>Окно setup-hold, среднее время между отказами в целочисленной арифметике и два триггера, которые это чинят.</p><ol><li><a href="/ru/learn/setup-hold-window/">13 Окно setup-hold</a></li><li><a href="/ru/learn/mtbf/">14 Среднее время между отказами</a></li><li><a href="/ru/learn/two-flop-sync/">15 Двухтриггерный синхронизатор</a></li></ol></li><li><h3>Модуль 6 · Пересечение многих бит</h3><p>Почему двоичная шина рвётся, почему код Грея нет, и рукопожатие, переносящее импульс между мирами.</p><ol><li><a href="/ru/learn/why-binary-fails/">16 Почему двоичный код рвётся</a></li><li><a href="/ru/learn/gray-code/">17 Код Грея</a></li><li><a href="/ru/learn/handshakes-and-pulses/">18 Рукопожатия и импульсы</a></li></ol></li><li><h3>Модуль 7 · Асинхронный FIFO</h3><p>Указатели, флаги и глубина: буфер, переносящий поток между двумя тактовыми сигналами.</p><ol><li><a href="/ru/learn/pointers-and-flags/">19 Указатели и флаги</a></li><li><a href="/ru/learn/depth-sizing/">20 Выбор глубины</a></li><li><a href="/ru/learn/fifo-testbench/">21 Тестбенч FIFO</a></li></ol></li><li><h3>Модуль 8 · Ограничения</h3><p>Строки, сообщающие анализатору, что такое тактовый сигнал, какие пути не проверять и чему должны соответствовать выводы.</p><ol><li><a href="/ru/learn/create-clock/">22 Строка create_clock</a></li><li><a href="/ru/learn/false-path-and-max-delay/">23 False path и max delay</a></li><li><a href="/ru/learn/io-timing/">24 Тайминг выводов</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Отчёт CDC, одно пересечение, захваченное на триггерах, и дифф битстрима, замыкающий курс.</p><ol><li><a href="/ru/learn/cdc-report/">25 Отчёт CDC</a></li><li><a href="/ru/learn/capture-a-crossing/">26 Захватить пересечение</a></li><li><a href="/ru/learn/capstone-diff/">27 Капстоун: поменяй и сравни</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/comments-and-prose/index.html b/apps/website/public/ru/learn/comments-and-prose/index.html
index 0b18afef93..389fae8641 100644
--- a/apps/website/public/ru/learn/comments-and-prose/index.html
+++ b/apps/website/public/ru/learn/comments-and-prose/index.html
@@ -85,5 +85,6 @@ <h1>Комментарии и строки прозы</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Файл спеки</h3><p>Что лежит в файле .t27: строка module, строки прозы и комментарии.</p><ol><li><a href="/ru/learn/what-a-spec-is/">01 Что такое спека</a></li><li><a href="/ru/learn/the-module-line/">02 Строка module</a></li><li class="on" aria-current="page"><a href="/ru/learn/comments-and-prose/">03 Комментарии и строки прозы</a></li></ol></li><li><h3>Модуль 2 · Значения и типы</h3><p>Константы, ширина целого числа, а также true, false и текст.</p><ol><li><a href="/ru/learn/constants/">04 Константы</a></li><li><a href="/ru/learn/whole-numbers/">05 Целые числа и их ширина</a></li><li><a href="/ru/learn/true-false-and-text/">06 Истина, ложь и текст</a></li></ol></li><li><h3>Модуль 3 · Массивы, триты и выражения</h3><p>Списки значений, три значения трита и что вычисляют операторы.</p><ol><li><a href="/ru/learn/arrays/">07 Массивы</a></li><li><a href="/ru/learn/trits/">08 Триты: минус один, ноль, один</a></li><li><a href="/ru/learn/expressions/">09 Выражения</a></li></ol></li><li><h3>Модуль 4 · Тесты и функции</h3><p>Блок test, несколько маленьких тестов в одной спеке и функция с типизированными входами.</p><ol><li><a href="/ru/learn/a-test-block/">10 Блок test</a></li><li><a href="/ru/learn/many-tests/">11 Много тестов в одной спеке</a></li><li><a href="/ru/learn/functions/">12 Функции</a></li></ol></li><li><h3>Модуль 5 · Внутри функции</h3><p>Имена через let и var, выбор через if и switch и циклы.</p><ol><li><a href="/ru/learn/local-names/">13 Локальные имена: let и var</a></li><li><a href="/ru/learn/if-and-else/">14 Выбор: if, else и switch</a></li><li><a href="/ru/learn/loops/">15 Циклы: while и for</a></li></ol></li><li><h3>Модуль 6 · Видимость и свои типы</h3><p>Что pub открывает другим модулям, и структуры и перечисления, которые вы объявляете.</p><ol><li><a href="/ru/learn/pub-or-private/">16 pub или закрытое имя</a></li><li><a href="/ru/learn/structs/">17 Структуры</a></li><li><a href="/ru/learn/enums/">18 Перечисления</a></li></ol></li><li><h3>Модуль 7 · Модули, правила и gen-ts</h3><p>Чтение другого модуля, правило, которое выполняется всегда, и спека, превращённая в TypeScript.</p><ol><li><a href="/ru/learn/use-other-modules/">19 Другие модули</a></li><li><a href="/ru/learn/invariants/">20 Инварианты и замеры</a></li><li><a href="/ru/learn/gen-ts/">21 gen-ts: спека превращается в TypeScript</a></li></ol></li><li><h3>Модуль 8 · Компилятор и tri</h3><p>Бэкенды, которые пишет t27c, стадии, в которых он читает спеку, и команда tri.</p><ol><li><a href="/ru/learn/seven-backends/">22 Семь бэкендов</a></li><li><a href="/ru/learn/stages-of-t27c/">23 Стадии t27c</a></li><li><a href="/ru/learn/tri-commands/">24 tri: командная строка</a></li></ol></li><li><h3>Модуль 9 · Ошибки, программа и что дальше</h3><p>Как читать ошибку компилятора, одна маленькая законченная программа и какой курс пройти дальше.</p><ol><li><a href="/ru/learn/reading-errors/">25 Как читать ошибки компилятора</a></li><li><a href="/ru/learn/a-small-program/">26 Маленькая законченная программа</a></li><li><a href="/ru/learn/where-next/">27 Куда дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/compare-at-the-top/index.html b/apps/website/public/ru/learn/compare-at-the-top/index.html
index 3d3b17c0b8..92ead28354 100644
--- a/apps/website/public/ru/learn/compare-at-the-top/index.html
+++ b/apps/website/public/ru/learn/compare-at-the-top/index.html
@@ -769,5 +769,6 @@ <h1>Сравнение сверху</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Три значения</h3><p>Почему три, как сбалансированная троичная система пишет любое число без знака и что дают переворот и отсечение тритов.</p><ol><li><a href="/ru/learn/why-three/">01 Почему три</a></li><li><a href="/ru/learn/balanced-ternary/">02 Сбалансированная троичная</a></li><li><a href="/ru/learn/flip-and-cut/">03 Переворот вместо минуса, отсечение вместо округления</a></li></ol></li><li><h3>Модуль 2 · Логика с «не знаю»</h3><p>Трёхзначные вентили Клини, два вентиля без двоичного двойника и сложение как пара таблиц.</p><ol><li><a href="/ru/learn/kleene-logic/">04 Трёхзначная логика</a></li><li><a href="/ru/learn/consensus-and-any/">05 Консенсус и «любой»</a></li><li><a href="/ru/learn/sum-and-carry/">06 Сумма и перенос</a></li></ol></li><li><h3>Модуль 3 · Складываем триты</h3><p>Полусумматор, полный сумматор и перенос, бегущий влево разряд за разрядом.</p><ol><li><a href="/ru/learn/half-adder/">07 Полусумматор</a></li><li><a href="/ru/learn/full-adder/">08 Полный сумматор</a></li><li><a href="/ru/learn/ripple-carry/">09 Сквозной перенос</a></li></ol></li><li><h3>Модуль 4 · Умножение без умножителя</h3><p>Копия, ноль или переворот: произведение на один трит, умножение столбиком и MAC, который только складывает.</p><ol><li><a href="/ru/learn/times-a-trit/">10 Умножить на один трит</a></li><li><a href="/ru/learn/long-multiplication/">11 Умножение столбиком</a></li><li><a href="/ru/learn/ternary-mac/">12 Троичный MAC</a></li></ol></li><li><h3>Модуль 5 · Триты в двоичной памяти</h3><p>Два бита на трит, пять тритов на байт и сколько бит нужно слову из 27 тритов.</p><ol><li><a href="/ru/learn/two-bits-per-trit/">13 Два бита на трит</a></li><li><a href="/ru/learn/five-trits-a-byte/">14 Пять тритов в байте</a></li><li><a href="/ru/learn/a-word-of-27/">15 Слово из 27 тритов</a></li></ol></li><li><h3>Модуль 6 · Командное слово TRI-27</h3><p>32-битное слово, которое декодирует эмулятор Trinity: его поля, 47 опкодов и 15-битное непосредственное значение.</p><ol><li><a href="/ru/learn/the-instruction-word/">16 Командное слово</a></li><li><a href="/ru/learn/the-opcode-map/">17 Карта опкодов</a></li><li><a href="/ru/learn/the-immediate/">18 Непосредственное значение</a></li></ol></li><li><h3>Модуль 7 · Троичная машина</h3><p>АЛУ из сумматоров этого курса, переход на три стороны и программа, которую можно пройти по шагам.</p><ol><li><a href="/ru/learn/a-ternary-alu/">19 Троичное АЛУ</a></li><li><a href="/ru/learn/three-way-jump/">20 Переход на три стороны</a></li><li><a href="/ru/learn/run-a-program/">21 Запустить программу</a></li></ol></li><li><h3>Модуль 8 · Три ответа</h3><p>Сравнение по старшему различному триту, поиск числа по третям и сортировка трёхзначными сравнениями.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/compare-at-the-top/">22 Сравнение сверху</a></li><li><a href="/ru/learn/questions-in-thirds/">23 Вопросы по третям</a></li><li><a href="/ru/learn/sort-by-three/">24 Сортировка трёхзначными сравнениями</a></li></ol></li><li><h3>Модуль 9 · Троичные нейроны</h3><p>Веса -1, 0, +1: один нейрон, детектор и маленький слой, и нигде нет умножителя.</p><ol><li><a href="/ru/learn/one-ternary-neuron/">25 Один троичный нейрон</a></li><li><a href="/ru/learn/a-bar-detector/">26 Детектор полосы</a></li><li><a href="/ru/learn/a-tiny-layer/">27 Маленький слой</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/compare-traces/index.html b/apps/website/public/ru/learn/compare-traces/index.html
index b5c55e4752..d9af2b87f9 100644
--- a/apps/website/public/ru/learn/compare-traces/index.html
+++ b/apps/website/public/ru/learn/compare-traces/index.html
@@ -181,5 +181,6 @@ <h1>Сравнение двух трасс</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Зачем проверять</h3><p>Дизайны, которые компилируются и ошибаются; модель, которая выносит вердикт; план, записанный до кода.</p><ol><li><a href="/ru/learn/bugs-that-compile/">01 Баги, которые компилируются</a></li><li><a href="/ru/learn/golden-model/">02 Золотая модель</a></li><li><a href="/ru/learn/plan-before-code/">03 Сначала план</a></li></ol></li><li><h3>Модуль 2 · Тестбенчи</h3><p>Стимулы, проверки и вердикт, записанные как один spec рядом с дизайном, который они судят.</p><ol><li><a href="/ru/learn/anatomy-of-a-tb/">04 Анатомия тестбенча</a></li><li><a href="/ru/learn/self-checking/">05 Самопроверка вместо всматривания</a></li><li><a href="/ru/learn/directed-vs-random/">06 Направленные против случайных</a></li></ol></li><li><h3>Модуль 3 · Временные диаграммы</h3><p>Трасса каждого сигнала, прочитанная так, как её читает инженер по железу, и два прогона, сравнённые между собой.</p><ol><li><a href="/ru/learn/vcd-format/">07 Формат VCD</a></li><li><a href="/ru/learn/debug-with-waves/">08 Отладка по волнам</a></li><li class="on" aria-current="page"><a href="/ru/learn/compare-traces/">09 Сравнение двух трасс</a></li></ol></li><li><h3>Модуль 4 · Векторы соответствия</h3><p>Случаи с ответом, записанным рядом, там, откуда их достанет компилятор.</p><ol><li><a href="/ru/learn/what-a-vector-is/">10 Что такое вектор</a></li><li><a href="/ru/learn/golden-ruler-vectors/">11 Векторы как эталонная линейка</a></li><li><a href="/ru/learn/end-to-end-scenarios/">12 Сквозные сценарии</a></li></ol></li><li><h3>Модуль 5 · Косимуляция</h3><p>Spec, симулятор и плата сходятся в одном ответе на стенде Artix-7 XC7A200T, и что делать, когда не сходятся.</p><ol><li><a href="/ru/learn/three-way-agreement/">13 Тройное согласие</a></li><li><a href="/ru/learn/simulator-internals/">14 Внутри симулятора</a></li><li><a href="/ru/learn/when-they-disagree/">15 Когда они не сходятся</a></li></ol></li><li><h3>Модуль 6 · Покрытие</h3><p>Чего коснулись тесты: строки, переключения, состояния — и что прячет это число.</p><ol><li><a href="/ru/learn/line-and-toggle/">16 Строки и переключения</a></li><li><a href="/ru/learn/fsm-coverage/">17 Покрытие автоматов</a></li><li><a href="/ru/learn/coverage-lies/">18 Покрытие лжёт</a></li></ol></li><li><h3>Модуль 7 · Формальные методы</h3><p>Ассерты, верные каждый такт; ограниченный поиск контрпримера; и почему доказательству нужна индукция.</p><ol><li><a href="/ru/learn/assertions/">19 Ассерты</a></li><li><a href="/ru/learn/bounded-model-checking/">20 Ограниченная проверка моделей</a></li><li><a href="/ru/learn/induction/">21 Индукция</a></li></ol></li><li><h3>Модуль 8 · Мутации</h3><p>Ломайте дизайн нарочно и считайте, что заметили тесты.</p><ol><li><a href="/ru/learn/kill-the-mutant/">22 Убить мутанта</a></li><li><a href="/ru/learn/surviving-mutants/">23 Выжившие мутанты</a></li><li><a href="/ru/learn/test-strength/">24 Сила тестов</a></li></ol></li><li><h3>Модуль 9 · Приёмка</h3><p>Одна команда, все квитанции, чистый вердикт, который можно показать.</p><ol><li><a href="/ru/learn/build-verify/">25 Сборка и проверка</a></li><li><a href="/ru/learn/board-selftest/">26 Стенд проверяет себя</a></li><li><a href="/ru/learn/verify-capstone/">27 Капстоун: проверить модуль целиком</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">Этот курс сам является spec t27: прочитайте его</a></p><p>Виджеты сохраняют свои английские слова: каждый живёт в собственном spec t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/compiler-stages/index.html b/apps/website/public/ru/learn/compiler-stages/index.html
index 9bc0a86e31..4ac2f59bce 100644
--- a/apps/website/public/ru/learn/compiler-stages/index.html
+++ b/apps/website/public/ru/learn/compiler-stages/index.html
@@ -210,5 +210,6 @@ <h1>Стадии t27c</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Чип</h3><p>Что такое FPGA, какой чип у нас и как его выводы встречаются с платой.</p><ol><li><a href="/ru/learn/fpga-is-a-grid/">01 FPGA — это сетка, которую вы настраиваете</a></li><li><a href="/ru/learn/which-chip/">02 Какой это чип?</a></li><li><a href="/ru/learn/pins-and-package/">03 Выводы, шарики и корпус</a></li></ol></li><li><h3>Модуль 2 · Числа в железе</h3><p>Биты, триты и форматы чисел, и сколько стоит арифметика в логике.</p><ol><li><a href="/ru/learn/bits-and-trits/">04 Биты, триты и GF16</a></li><li><a href="/ru/learn/what-arithmetic-costs/">05 Сколько стоит арифметика</a></li><li><a href="/ru/learn/how-much-fits/">06 Сколько помещается на чип</a></li></ol></li><li><h3>Модуль 3 · Ваша программа на t27</h3><p>Напишите спеку, проверьте её и увидьте, почему «компилируется» не значит «верно».</p><ol><li><a href="/ru/learn/hello-t27/">07 Привет, t27</a></li><li><a href="/ru/learn/tests-are-the-spec/">08 Тесты — часть спеки</a></li><li><a href="/ru/learn/compiles-is-not-correct/">09 Компилируется — не значит верно</a></li></ol></li><li><h3>Модуль 4 · Внутри t27c</h3><p>Как компилятор читает спеку и что он пишет, включая нативный t27b.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/compiler-stages/">10 Стадии t27c</a></li><li><a href="/ru/learn/t27c-on-a-real-machine/">11 t27c на настоящей машине</a></li><li><a href="/ru/learn/seven-backends-and-t27b/">12 Семь бэкендов и t27b</a></li></ol></li><li><h3>Модуль 5 · От спеки к железу</h3><p>Verilog, который пишет t27c, ячейки, в которые он превращается, и что делает одна LUT.</p><ol><li><a href="/ru/learn/spec-to-verilog/">13 От спеки к Verilog</a></li><li><a href="/ru/learn/count-the-cells/">14 Посчитайте ячейки</a></li><li><a href="/ru/learn/one-lut-up-close/">15 Одна LUT вблизи</a></li></ol></li><li><h3>Модуль 6 · Читаем синтез</h3><p>Что yosys сообщает о вашем дизайне и какие предупреждения важны.</p><ol><li><a href="/ru/learn/synthesis-warnings/">16 Важные предупреждения синтеза</a></li><li><a href="/ru/learn/state-machines/">17 Конечные автоматы</a></li><li><a href="/ru/learn/where-the-luts-go/">18 Куда уходят LUT</a></li></ol></li><li><h3>Модуль 7 · Размещение, трассировка, тайминг</h3><p>Где ячейки оказываются на кристалле и успевает ли тактовая частота.</p><ol><li><a href="/ru/learn/placement/">19 Размещение: где оказываются ячейки</a></li><li><a href="/ru/learn/logic-depth/">20 Глубина логики</a></li><li><a href="/ru/learn/timing-slack/">21 Тайминг и запас</a></li></ol></li><li><h3>Модуль 8 · Битстрим</h3><p>Как оттрассированный дизайн становится битами, которые загружает чип, без Vivado.</p><ol><li><a href="/ru/learn/fasm-to-frames/">22 От FASM к фреймам</a></li><li><a href="/ru/learn/a-build-receipt/">23 Сборка по пунктам</a></li><li><a href="/ru/learn/read-a-bitstream/">24 Читаем битстрим</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Спросите чип, кто он, загрузите биты и проверьте их.</p><ol><li><a href="/ru/learn/jtag-asks-the-chip/">25 JTAG спрашивает чип</a></li><li><a href="/ru/learn/same-bits-on-the-board/">26 Те же биты — на плате</a></li><li><a href="/ru/learn/capstone/">27 Итог: переугадайте yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/compiles-is-not-correct/index.html b/apps/website/public/ru/learn/compiles-is-not-correct/index.html
index daf93a8194..6bcebcfc17 100644
--- a/apps/website/public/ru/learn/compiles-is-not-correct/index.html
+++ b/apps/website/public/ru/learn/compiles-is-not-correct/index.html
@@ -179,5 +179,6 @@ <h1>Компилируется — не значит верно</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Чип</h3><p>Что такое FPGA, какой чип у нас и как его выводы встречаются с платой.</p><ol><li><a href="/ru/learn/fpga-is-a-grid/">01 FPGA — это сетка, которую вы настраиваете</a></li><li><a href="/ru/learn/which-chip/">02 Какой это чип?</a></li><li><a href="/ru/learn/pins-and-package/">03 Выводы, шарики и корпус</a></li></ol></li><li><h3>Модуль 2 · Числа в железе</h3><p>Биты, триты и форматы чисел, и сколько стоит арифметика в логике.</p><ol><li><a href="/ru/learn/bits-and-trits/">04 Биты, триты и GF16</a></li><li><a href="/ru/learn/what-arithmetic-costs/">05 Сколько стоит арифметика</a></li><li><a href="/ru/learn/how-much-fits/">06 Сколько помещается на чип</a></li></ol></li><li><h3>Модуль 3 · Ваша программа на t27</h3><p>Напишите спеку, проверьте её и увидьте, почему «компилируется» не значит «верно».</p><ol><li><a href="/ru/learn/hello-t27/">07 Привет, t27</a></li><li><a href="/ru/learn/tests-are-the-spec/">08 Тесты — часть спеки</a></li><li class="on" aria-current="page"><a href="/ru/learn/compiles-is-not-correct/">09 Компилируется — не значит верно</a></li></ol></li><li><h3>Модуль 4 · Внутри t27c</h3><p>Как компилятор читает спеку и что он пишет, включая нативный t27b.</p><ol><li><a href="/ru/learn/compiler-stages/">10 Стадии t27c</a></li><li><a href="/ru/learn/t27c-on-a-real-machine/">11 t27c на настоящей машине</a></li><li><a href="/ru/learn/seven-backends-and-t27b/">12 Семь бэкендов и t27b</a></li></ol></li><li><h3>Модуль 5 · От спеки к железу</h3><p>Verilog, который пишет t27c, ячейки, в которые он превращается, и что делает одна LUT.</p><ol><li><a href="/ru/learn/spec-to-verilog/">13 От спеки к Verilog</a></li><li><a href="/ru/learn/count-the-cells/">14 Посчитайте ячейки</a></li><li><a href="/ru/learn/one-lut-up-close/">15 Одна LUT вблизи</a></li></ol></li><li><h3>Модуль 6 · Читаем синтез</h3><p>Что yosys сообщает о вашем дизайне и какие предупреждения важны.</p><ol><li><a href="/ru/learn/synthesis-warnings/">16 Важные предупреждения синтеза</a></li><li><a href="/ru/learn/state-machines/">17 Конечные автоматы</a></li><li><a href="/ru/learn/where-the-luts-go/">18 Куда уходят LUT</a></li></ol></li><li><h3>Модуль 7 · Размещение, трассировка, тайминг</h3><p>Где ячейки оказываются на кристалле и успевает ли тактовая частота.</p><ol><li><a href="/ru/learn/placement/">19 Размещение: где оказываются ячейки</a></li><li><a href="/ru/learn/logic-depth/">20 Глубина логики</a></li><li><a href="/ru/learn/timing-slack/">21 Тайминг и запас</a></li></ol></li><li><h3>Модуль 8 · Битстрим</h3><p>Как оттрассированный дизайн становится битами, которые загружает чип, без Vivado.</p><ol><li><a href="/ru/learn/fasm-to-frames/">22 От FASM к фреймам</a></li><li><a href="/ru/learn/a-build-receipt/">23 Сборка по пунктам</a></li><li><a href="/ru/learn/read-a-bitstream/">24 Читаем битстрим</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Спросите чип, кто он, загрузите биты и проверьте их.</p><ol><li><a href="/ru/learn/jtag-asks-the-chip/">25 JTAG спрашивает чип</a></li><li><a href="/ru/learn/same-bits-on-the-board/">26 Те же биты — на плате</a></li><li><a href="/ru/learn/capstone/">27 Итог: переугадайте yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/consensus-and-any/index.html b/apps/website/public/ru/learn/consensus-and-any/index.html
index 246fcf1b1b..d73d66eef6 100644
--- a/apps/website/public/ru/learn/consensus-and-any/index.html
+++ b/apps/website/public/ru/learn/consensus-and-any/index.html
@@ -769,5 +769,6 @@ <h1>Консенсус и «любой»</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Три значения</h3><p>Почему три, как сбалансированная троичная система пишет любое число без знака и что дают переворот и отсечение тритов.</p><ol><li><a href="/ru/learn/why-three/">01 Почему три</a></li><li><a href="/ru/learn/balanced-ternary/">02 Сбалансированная троичная</a></li><li><a href="/ru/learn/flip-and-cut/">03 Переворот вместо минуса, отсечение вместо округления</a></li></ol></li><li><h3>Модуль 2 · Логика с «не знаю»</h3><p>Трёхзначные вентили Клини, два вентиля без двоичного двойника и сложение как пара таблиц.</p><ol><li><a href="/ru/learn/kleene-logic/">04 Трёхзначная логика</a></li><li class="on" aria-current="page"><a href="/ru/learn/consensus-and-any/">05 Консенсус и «любой»</a></li><li><a href="/ru/learn/sum-and-carry/">06 Сумма и перенос</a></li></ol></li><li><h3>Модуль 3 · Складываем триты</h3><p>Полусумматор, полный сумматор и перенос, бегущий влево разряд за разрядом.</p><ol><li><a href="/ru/learn/half-adder/">07 Полусумматор</a></li><li><a href="/ru/learn/full-adder/">08 Полный сумматор</a></li><li><a href="/ru/learn/ripple-carry/">09 Сквозной перенос</a></li></ol></li><li><h3>Модуль 4 · Умножение без умножителя</h3><p>Копия, ноль или переворот: произведение на один трит, умножение столбиком и MAC, который только складывает.</p><ol><li><a href="/ru/learn/times-a-trit/">10 Умножить на один трит</a></li><li><a href="/ru/learn/long-multiplication/">11 Умножение столбиком</a></li><li><a href="/ru/learn/ternary-mac/">12 Троичный MAC</a></li></ol></li><li><h3>Модуль 5 · Триты в двоичной памяти</h3><p>Два бита на трит, пять тритов на байт и сколько бит нужно слову из 27 тритов.</p><ol><li><a href="/ru/learn/two-bits-per-trit/">13 Два бита на трит</a></li><li><a href="/ru/learn/five-trits-a-byte/">14 Пять тритов в байте</a></li><li><a href="/ru/learn/a-word-of-27/">15 Слово из 27 тритов</a></li></ol></li><li><h3>Модуль 6 · Командное слово TRI-27</h3><p>32-битное слово, которое декодирует эмулятор Trinity: его поля, 47 опкодов и 15-битное непосредственное значение.</p><ol><li><a href="/ru/learn/the-instruction-word/">16 Командное слово</a></li><li><a href="/ru/learn/the-opcode-map/">17 Карта опкодов</a></li><li><a href="/ru/learn/the-immediate/">18 Непосредственное значение</a></li></ol></li><li><h3>Модуль 7 · Троичная машина</h3><p>АЛУ из сумматоров этого курса, переход на три стороны и программа, которую можно пройти по шагам.</p><ol><li><a href="/ru/learn/a-ternary-alu/">19 Троичное АЛУ</a></li><li><a href="/ru/learn/three-way-jump/">20 Переход на три стороны</a></li><li><a href="/ru/learn/run-a-program/">21 Запустить программу</a></li></ol></li><li><h3>Модуль 8 · Три ответа</h3><p>Сравнение по старшему различному триту, поиск числа по третям и сортировка трёхзначными сравнениями.</p><ol><li><a href="/ru/learn/compare-at-the-top/">22 Сравнение сверху</a></li><li><a href="/ru/learn/questions-in-thirds/">23 Вопросы по третям</a></li><li><a href="/ru/learn/sort-by-three/">24 Сортировка трёхзначными сравнениями</a></li></ol></li><li><h3>Модуль 9 · Троичные нейроны</h3><p>Веса -1, 0, +1: один нейрон, детектор и маленький слой, и нигде нет умножителя.</p><ol><li><a href="/ru/learn/one-ternary-neuron/">25 Один троичный нейрон</a></li><li><a href="/ru/learn/a-bar-detector/">26 Детектор полосы</a></li><li><a href="/ru/learn/a-tiny-layer/">27 Маленький слой</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/constants/index.html b/apps/website/public/ru/learn/constants/index.html
index cd8789cc3a..2db59d8ba4 100644
--- a/apps/website/public/ru/learn/constants/index.html
+++ b/apps/website/public/ru/learn/constants/index.html
@@ -83,5 +83,6 @@ <h1>Константы</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Файл спеки</h3><p>Что лежит в файле .t27: строка module, строки прозы и комментарии.</p><ol><li><a href="/ru/learn/what-a-spec-is/">01 Что такое спека</a></li><li><a href="/ru/learn/the-module-line/">02 Строка module</a></li><li><a href="/ru/learn/comments-and-prose/">03 Комментарии и строки прозы</a></li></ol></li><li><h3>Модуль 2 · Значения и типы</h3><p>Константы, ширина целого числа, а также true, false и текст.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/constants/">04 Константы</a></li><li><a href="/ru/learn/whole-numbers/">05 Целые числа и их ширина</a></li><li><a href="/ru/learn/true-false-and-text/">06 Истина, ложь и текст</a></li></ol></li><li><h3>Модуль 3 · Массивы, триты и выражения</h3><p>Списки значений, три значения трита и что вычисляют операторы.</p><ol><li><a href="/ru/learn/arrays/">07 Массивы</a></li><li><a href="/ru/learn/trits/">08 Триты: минус один, ноль, один</a></li><li><a href="/ru/learn/expressions/">09 Выражения</a></li></ol></li><li><h3>Модуль 4 · Тесты и функции</h3><p>Блок test, несколько маленьких тестов в одной спеке и функция с типизированными входами.</p><ol><li><a href="/ru/learn/a-test-block/">10 Блок test</a></li><li><a href="/ru/learn/many-tests/">11 Много тестов в одной спеке</a></li><li><a href="/ru/learn/functions/">12 Функции</a></li></ol></li><li><h3>Модуль 5 · Внутри функции</h3><p>Имена через let и var, выбор через if и switch и циклы.</p><ol><li><a href="/ru/learn/local-names/">13 Локальные имена: let и var</a></li><li><a href="/ru/learn/if-and-else/">14 Выбор: if, else и switch</a></li><li><a href="/ru/learn/loops/">15 Циклы: while и for</a></li></ol></li><li><h3>Модуль 6 · Видимость и свои типы</h3><p>Что pub открывает другим модулям, и структуры и перечисления, которые вы объявляете.</p><ol><li><a href="/ru/learn/pub-or-private/">16 pub или закрытое имя</a></li><li><a href="/ru/learn/structs/">17 Структуры</a></li><li><a href="/ru/learn/enums/">18 Перечисления</a></li></ol></li><li><h3>Модуль 7 · Модули, правила и gen-ts</h3><p>Чтение другого модуля, правило, которое выполняется всегда, и спека, превращённая в TypeScript.</p><ol><li><a href="/ru/learn/use-other-modules/">19 Другие модули</a></li><li><a href="/ru/learn/invariants/">20 Инварианты и замеры</a></li><li><a href="/ru/learn/gen-ts/">21 gen-ts: спека превращается в TypeScript</a></li></ol></li><li><h3>Модуль 8 · Компилятор и tri</h3><p>Бэкенды, которые пишет t27c, стадии, в которых он читает спеку, и команда tri.</p><ol><li><a href="/ru/learn/seven-backends/">22 Семь бэкендов</a></li><li><a href="/ru/learn/stages-of-t27c/">23 Стадии t27c</a></li><li><a href="/ru/learn/tri-commands/">24 tri: командная строка</a></li></ol></li><li><h3>Модуль 9 · Ошибки, программа и что дальше</h3><p>Как читать ошибку компилятора, одна маленькая законченная программа и какой курс пройти дальше.</p><ol><li><a href="/ru/learn/reading-errors/">25 Как читать ошибки компилятора</a></li><li><a href="/ru/learn/a-small-program/">26 Маленькая законченная программа</a></li><li><a href="/ru/learn/where-next/">27 Куда дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/count-the-cells/index.html b/apps/website/public/ru/learn/count-the-cells/index.html
index a2a589219e..8220e46144 100644
--- a/apps/website/public/ru/learn/count-the-cells/index.html
+++ b/apps/website/public/ru/learn/count-the-cells/index.html
@@ -222,5 +222,6 @@ <h1>Посчитайте ячейки</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Чип</h3><p>Что такое FPGA, какой чип у нас и как его выводы встречаются с платой.</p><ol><li><a href="/ru/learn/fpga-is-a-grid/">01 FPGA — это сетка, которую вы настраиваете</a></li><li><a href="/ru/learn/which-chip/">02 Какой это чип?</a></li><li><a href="/ru/learn/pins-and-package/">03 Выводы, шарики и корпус</a></li></ol></li><li><h3>Модуль 2 · Числа в железе</h3><p>Биты, триты и форматы чисел, и сколько стоит арифметика в логике.</p><ol><li><a href="/ru/learn/bits-and-trits/">04 Биты, триты и GF16</a></li><li><a href="/ru/learn/what-arithmetic-costs/">05 Сколько стоит арифметика</a></li><li><a href="/ru/learn/how-much-fits/">06 Сколько помещается на чип</a></li></ol></li><li><h3>Модуль 3 · Ваша программа на t27</h3><p>Напишите спеку, проверьте её и увидьте, почему «компилируется» не значит «верно».</p><ol><li><a href="/ru/learn/hello-t27/">07 Привет, t27</a></li><li><a href="/ru/learn/tests-are-the-spec/">08 Тесты — часть спеки</a></li><li><a href="/ru/learn/compiles-is-not-correct/">09 Компилируется — не значит верно</a></li></ol></li><li><h3>Модуль 4 · Внутри t27c</h3><p>Как компилятор читает спеку и что он пишет, включая нативный t27b.</p><ol><li><a href="/ru/learn/compiler-stages/">10 Стадии t27c</a></li><li><a href="/ru/learn/t27c-on-a-real-machine/">11 t27c на настоящей машине</a></li><li><a href="/ru/learn/seven-backends-and-t27b/">12 Семь бэкендов и t27b</a></li></ol></li><li><h3>Модуль 5 · От спеки к железу</h3><p>Verilog, который пишет t27c, ячейки, в которые он превращается, и что делает одна LUT.</p><ol><li><a href="/ru/learn/spec-to-verilog/">13 От спеки к Verilog</a></li><li class="on" aria-current="page"><a href="/ru/learn/count-the-cells/">14 Посчитайте ячейки</a></li><li><a href="/ru/learn/one-lut-up-close/">15 Одна LUT вблизи</a></li></ol></li><li><h3>Модуль 6 · Читаем синтез</h3><p>Что yosys сообщает о вашем дизайне и какие предупреждения важны.</p><ol><li><a href="/ru/learn/synthesis-warnings/">16 Важные предупреждения синтеза</a></li><li><a href="/ru/learn/state-machines/">17 Конечные автоматы</a></li><li><a href="/ru/learn/where-the-luts-go/">18 Куда уходят LUT</a></li></ol></li><li><h3>Модуль 7 · Размещение, трассировка, тайминг</h3><p>Где ячейки оказываются на кристалле и успевает ли тактовая частота.</p><ol><li><a href="/ru/learn/placement/">19 Размещение: где оказываются ячейки</a></li><li><a href="/ru/learn/logic-depth/">20 Глубина логики</a></li><li><a href="/ru/learn/timing-slack/">21 Тайминг и запас</a></li></ol></li><li><h3>Модуль 8 · Битстрим</h3><p>Как оттрассированный дизайн становится битами, которые загружает чип, без Vivado.</p><ol><li><a href="/ru/learn/fasm-to-frames/">22 От FASM к фреймам</a></li><li><a href="/ru/learn/a-build-receipt/">23 Сборка по пунктам</a></li><li><a href="/ru/learn/read-a-bitstream/">24 Читаем битстрим</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Спросите чип, кто он, загрузите биты и проверьте их.</p><ol><li><a href="/ru/learn/jtag-asks-the-chip/">25 JTAG спрашивает чип</a></li><li><a href="/ru/learn/same-bits-on-the-board/">26 Те же биты — на плате</a></li><li><a href="/ru/learn/capstone/">27 Итог: переугадайте yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/course/index.html b/apps/website/public/ru/learn/course/index.html
index 81ad5c1efc..f9ed6b7402 100644
--- a/apps/website/public/ru/learn/course/index.html
+++ b/apps/website/public/ru/learn/course/index.html
@@ -67,5 +67,6 @@ <h1>С нуля до чипа: программирование FPGA на t27</h
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Чип</h3><p>Что такое FPGA, какой чип у нас и как его выводы встречаются с платой.</p><ol><li><a href="/ru/learn/fpga-is-a-grid/">01 FPGA — это сетка, которую вы настраиваете</a></li><li><a href="/ru/learn/which-chip/">02 Какой это чип?</a></li><li><a href="/ru/learn/pins-and-package/">03 Выводы, шарики и корпус</a></li></ol></li><li><h3>Модуль 2 · Числа в железе</h3><p>Биты, триты и форматы чисел, и сколько стоит арифметика в логике.</p><ol><li><a href="/ru/learn/bits-and-trits/">04 Биты, триты и GF16</a></li><li><a href="/ru/learn/what-arithmetic-costs/">05 Сколько стоит арифметика</a></li><li><a href="/ru/learn/how-much-fits/">06 Сколько помещается на чип</a></li></ol></li><li><h3>Модуль 3 · Ваша программа на t27</h3><p>Напишите спеку, проверьте её и увидьте, почему «компилируется» не значит «верно».</p><ol><li><a href="/ru/learn/hello-t27/">07 Привет, t27</a></li><li><a href="/ru/learn/tests-are-the-spec/">08 Тесты — часть спеки</a></li><li><a href="/ru/learn/compiles-is-not-correct/">09 Компилируется — не значит верно</a></li></ol></li><li><h3>Модуль 4 · Внутри t27c</h3><p>Как компилятор читает спеку и что он пишет, включая нативный t27b.</p><ol><li><a href="/ru/learn/compiler-stages/">10 Стадии t27c</a></li><li><a href="/ru/learn/t27c-on-a-real-machine/">11 t27c на настоящей машине</a></li><li><a href="/ru/learn/seven-backends-and-t27b/">12 Семь бэкендов и t27b</a></li></ol></li><li><h3>Модуль 5 · От спеки к железу</h3><p>Verilog, который пишет t27c, ячейки, в которые он превращается, и что делает одна LUT.</p><ol><li><a href="/ru/learn/spec-to-verilog/">13 От спеки к Verilog</a></li><li><a href="/ru/learn/count-the-cells/">14 Посчитайте ячейки</a></li><li><a href="/ru/learn/one-lut-up-close/">15 Одна LUT вблизи</a></li></ol></li><li><h3>Модуль 6 · Читаем синтез</h3><p>Что yosys сообщает о вашем дизайне и какие предупреждения важны.</p><ol><li><a href="/ru/learn/synthesis-warnings/">16 Важные предупреждения синтеза</a></li><li><a href="/ru/learn/state-machines/">17 Конечные автоматы</a></li><li><a href="/ru/learn/where-the-luts-go/">18 Куда уходят LUT</a></li></ol></li><li><h3>Модуль 7 · Размещение, трассировка, тайминг</h3><p>Где ячейки оказываются на кристалле и успевает ли тактовая частота.</p><ol><li><a href="/ru/learn/placement/">19 Размещение: где оказываются ячейки</a></li><li><a href="/ru/learn/logic-depth/">20 Глубина логики</a></li><li><a href="/ru/learn/timing-slack/">21 Тайминг и запас</a></li></ol></li><li><h3>Модуль 8 · Битстрим</h3><p>Как оттрассированный дизайн становится битами, которые загружает чип, без Vivado.</p><ol><li><a href="/ru/learn/fasm-to-frames/">22 От FASM к фреймам</a></li><li><a href="/ru/learn/a-build-receipt/">23 Сборка по пунктам</a></li><li><a href="/ru/learn/read-a-bitstream/">24 Читаем битстрим</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Спросите чип, кто он, загрузите биты и проверьте их.</p><ol><li><a href="/ru/learn/jtag-asks-the-chip/">25 JTAG спрашивает чип</a></li><li><a href="/ru/learn/same-bits-on-the-board/">26 Те же биты — на плате</a></li><li><a href="/ru/learn/capstone/">27 Итог: переугадайте yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/coverage-lies/index.html b/apps/website/public/ru/learn/coverage-lies/index.html
index 49070047d6..cd776602af 100644
--- a/apps/website/public/ru/learn/coverage-lies/index.html
+++ b/apps/website/public/ru/learn/coverage-lies/index.html
@@ -304,5 +304,6 @@ <h1>Покрытие лжёт</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Зачем проверять</h3><p>Дизайны, которые компилируются и ошибаются; модель, которая выносит вердикт; план, записанный до кода.</p><ol><li><a href="/ru/learn/bugs-that-compile/">01 Баги, которые компилируются</a></li><li><a href="/ru/learn/golden-model/">02 Золотая модель</a></li><li><a href="/ru/learn/plan-before-code/">03 Сначала план</a></li></ol></li><li><h3>Модуль 2 · Тестбенчи</h3><p>Стимулы, проверки и вердикт, записанные как один spec рядом с дизайном, который они судят.</p><ol><li><a href="/ru/learn/anatomy-of-a-tb/">04 Анатомия тестбенча</a></li><li><a href="/ru/learn/self-checking/">05 Самопроверка вместо всматривания</a></li><li><a href="/ru/learn/directed-vs-random/">06 Направленные против случайных</a></li></ol></li><li><h3>Модуль 3 · Временные диаграммы</h3><p>Трасса каждого сигнала, прочитанная так, как её читает инженер по железу, и два прогона, сравнённые между собой.</p><ol><li><a href="/ru/learn/vcd-format/">07 Формат VCD</a></li><li><a href="/ru/learn/debug-with-waves/">08 Отладка по волнам</a></li><li><a href="/ru/learn/compare-traces/">09 Сравнение двух трасс</a></li></ol></li><li><h3>Модуль 4 · Векторы соответствия</h3><p>Случаи с ответом, записанным рядом, там, откуда их достанет компилятор.</p><ol><li><a href="/ru/learn/what-a-vector-is/">10 Что такое вектор</a></li><li><a href="/ru/learn/golden-ruler-vectors/">11 Векторы как эталонная линейка</a></li><li><a href="/ru/learn/end-to-end-scenarios/">12 Сквозные сценарии</a></li></ol></li><li><h3>Модуль 5 · Косимуляция</h3><p>Spec, симулятор и плата сходятся в одном ответе на стенде Artix-7 XC7A200T, и что делать, когда не сходятся.</p><ol><li><a href="/ru/learn/three-way-agreement/">13 Тройное согласие</a></li><li><a href="/ru/learn/simulator-internals/">14 Внутри симулятора</a></li><li><a href="/ru/learn/when-they-disagree/">15 Когда они не сходятся</a></li></ol></li><li><h3>Модуль 6 · Покрытие</h3><p>Чего коснулись тесты: строки, переключения, состояния — и что прячет это число.</p><ol><li><a href="/ru/learn/line-and-toggle/">16 Строки и переключения</a></li><li><a href="/ru/learn/fsm-coverage/">17 Покрытие автоматов</a></li><li class="on" aria-current="page"><a href="/ru/learn/coverage-lies/">18 Покрытие лжёт</a></li></ol></li><li><h3>Модуль 7 · Формальные методы</h3><p>Ассерты, верные каждый такт; ограниченный поиск контрпримера; и почему доказательству нужна индукция.</p><ol><li><a href="/ru/learn/assertions/">19 Ассерты</a></li><li><a href="/ru/learn/bounded-model-checking/">20 Ограниченная проверка моделей</a></li><li><a href="/ru/learn/induction/">21 Индукция</a></li></ol></li><li><h3>Модуль 8 · Мутации</h3><p>Ломайте дизайн нарочно и считайте, что заметили тесты.</p><ol><li><a href="/ru/learn/kill-the-mutant/">22 Убить мутанта</a></li><li><a href="/ru/learn/surviving-mutants/">23 Выжившие мутанты</a></li><li><a href="/ru/learn/test-strength/">24 Сила тестов</a></li></ol></li><li><h3>Модуль 9 · Приёмка</h3><p>Одна команда, все квитанции, чистый вердикт, который можно показать.</p><ol><li><a href="/ru/learn/build-verify/">25 Сборка и проверка</a></li><li><a href="/ru/learn/board-selftest/">26 Стенд проверяет себя</a></li><li><a href="/ru/learn/verify-capstone/">27 Капстоун: проверить модуль целиком</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">Этот курс сам является spec t27: прочитайте его</a></p><p>Виджеты сохраняют свои английские слова: каждый живёт в собственном spec t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/create-clock/index.html b/apps/website/public/ru/learn/create-clock/index.html
index dd487619fc..645eb475d7 100644
--- a/apps/website/public/ru/learn/create-clock/index.html
+++ b/apps/website/public/ru/learn/create-clock/index.html
@@ -364,5 +364,6 @@ <h1>Строка create_clock</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое тактовый сигнал</h3><p>Один фронт, один мир: всё, что делит фронт тактового сигнала, делит и мир; период и джиттер настоящего фронта и место, где тактовый сигнал входит в плату.</p><ol><li><a href="/ru/learn/one-edge-one-world/">01 Один фронт, один мир</a></li><li><a href="/ru/learn/period-and-jitter/">02 Период, фронт и джиттер</a></li><li><a href="/ru/learn/where-the-clock-enters/">03 Где входит тактовый сигнал</a></li></ol></li><li><h3>Модуль 2 · Деревья тактового сигнала</h3><p>Перекос и задержка прохождения, глобальная сеть буферов и ловушка стробирования клока логикой.</p><ol><li><a href="/ru/learn/skew-and-insertion/">04 Перекос и задержка прохождения</a></li><li><a href="/ru/learn/global-buffers/">05 Глобальный буфер</a></li><li><a href="/ru/learn/gated-clock-trap/">06 Ловушка стробированного клока</a></li></ol></li><li><h3>Модуль 3 · PLL и MMCM</h3><p>Умножить и поделить один тактовый сигнал в другой, сдвинуть его фазу шагами VCO и какие тактовые сигналы анализатор считает родственными.</p><ol><li><a href="/ru/learn/multiply-and-divide/">07 Умножить и поделить</a></li><li><a href="/ru/learn/phase-and-lock/">08 Фаза и захват</a></li><li><a href="/ru/learn/related-or-not/">09 Родственные или чужие</a></li></ol></li><li><h3>Модуль 4 · Сбросы</h3><p>Ассертировать асинхронно, снимать синхронно: три вида сброса, конвейер снятия и дерево, которое растит сброс.</p><ol><li><a href="/ru/learn/sync-or-async/">10 Синхронный или асинхронный</a></li><li><a href="/ru/learn/reset-release/">11 Конвейер снятия</a></li><li><a href="/ru/learn/reset-fanout/">12 Разветвление сброса</a></li></ol></li><li><h3>Модуль 5 · Метастабильность</h3><p>Окно setup-hold, среднее время между отказами в целочисленной арифметике и два триггера, которые это чинят.</p><ol><li><a href="/ru/learn/setup-hold-window/">13 Окно setup-hold</a></li><li><a href="/ru/learn/mtbf/">14 Среднее время между отказами</a></li><li><a href="/ru/learn/two-flop-sync/">15 Двухтриггерный синхронизатор</a></li></ol></li><li><h3>Модуль 6 · Пересечение многих бит</h3><p>Почему двоичная шина рвётся, почему код Грея нет, и рукопожатие, переносящее импульс между мирами.</p><ol><li><a href="/ru/learn/why-binary-fails/">16 Почему двоичный код рвётся</a></li><li><a href="/ru/learn/gray-code/">17 Код Грея</a></li><li><a href="/ru/learn/handshakes-and-pulses/">18 Рукопожатия и импульсы</a></li></ol></li><li><h3>Модуль 7 · Асинхронный FIFO</h3><p>Указатели, флаги и глубина: буфер, переносящий поток между двумя тактовыми сигналами.</p><ol><li><a href="/ru/learn/pointers-and-flags/">19 Указатели и флаги</a></li><li><a href="/ru/learn/depth-sizing/">20 Выбор глубины</a></li><li><a href="/ru/learn/fifo-testbench/">21 Тестбенч FIFO</a></li></ol></li><li><h3>Модуль 8 · Ограничения</h3><p>Строки, сообщающие анализатору, что такое тактовый сигнал, какие пути не проверять и чему должны соответствовать выводы.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/create-clock/">22 Строка create_clock</a></li><li><a href="/ru/learn/false-path-and-max-delay/">23 False path и max delay</a></li><li><a href="/ru/learn/io-timing/">24 Тайминг выводов</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Отчёт CDC, одно пересечение, захваченное на триггерах, и дифф битстрима, замыкающий курс.</p><ol><li><a href="/ru/learn/cdc-report/">25 Отчёт CDC</a></li><li><a href="/ru/learn/capture-a-crossing/">26 Захватить пересечение</a></li><li><a href="/ru/learn/capstone-diff/">27 Капстоун: поменяй и сравни</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/debug-with-waves/index.html b/apps/website/public/ru/learn/debug-with-waves/index.html
index 0e2af60cca..019442f285 100644
--- a/apps/website/public/ru/learn/debug-with-waves/index.html
+++ b/apps/website/public/ru/learn/debug-with-waves/index.html
@@ -161,5 +161,6 @@ <h1>Отладка по волнам</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Зачем проверять</h3><p>Дизайны, которые компилируются и ошибаются; модель, которая выносит вердикт; план, записанный до кода.</p><ol><li><a href="/ru/learn/bugs-that-compile/">01 Баги, которые компилируются</a></li><li><a href="/ru/learn/golden-model/">02 Золотая модель</a></li><li><a href="/ru/learn/plan-before-code/">03 Сначала план</a></li></ol></li><li><h3>Модуль 2 · Тестбенчи</h3><p>Стимулы, проверки и вердикт, записанные как один spec рядом с дизайном, который они судят.</p><ol><li><a href="/ru/learn/anatomy-of-a-tb/">04 Анатомия тестбенча</a></li><li><a href="/ru/learn/self-checking/">05 Самопроверка вместо всматривания</a></li><li><a href="/ru/learn/directed-vs-random/">06 Направленные против случайных</a></li></ol></li><li><h3>Модуль 3 · Временные диаграммы</h3><p>Трасса каждого сигнала, прочитанная так, как её читает инженер по железу, и два прогона, сравнённые между собой.</p><ol><li><a href="/ru/learn/vcd-format/">07 Формат VCD</a></li><li class="on" aria-current="page"><a href="/ru/learn/debug-with-waves/">08 Отладка по волнам</a></li><li><a href="/ru/learn/compare-traces/">09 Сравнение двух трасс</a></li></ol></li><li><h3>Модуль 4 · Векторы соответствия</h3><p>Случаи с ответом, записанным рядом, там, откуда их достанет компилятор.</p><ol><li><a href="/ru/learn/what-a-vector-is/">10 Что такое вектор</a></li><li><a href="/ru/learn/golden-ruler-vectors/">11 Векторы как эталонная линейка</a></li><li><a href="/ru/learn/end-to-end-scenarios/">12 Сквозные сценарии</a></li></ol></li><li><h3>Модуль 5 · Косимуляция</h3><p>Spec, симулятор и плата сходятся в одном ответе на стенде Artix-7 XC7A200T, и что делать, когда не сходятся.</p><ol><li><a href="/ru/learn/three-way-agreement/">13 Тройное согласие</a></li><li><a href="/ru/learn/simulator-internals/">14 Внутри симулятора</a></li><li><a href="/ru/learn/when-they-disagree/">15 Когда они не сходятся</a></li></ol></li><li><h3>Модуль 6 · Покрытие</h3><p>Чего коснулись тесты: строки, переключения, состояния — и что прячет это число.</p><ol><li><a href="/ru/learn/line-and-toggle/">16 Строки и переключения</a></li><li><a href="/ru/learn/fsm-coverage/">17 Покрытие автоматов</a></li><li><a href="/ru/learn/coverage-lies/">18 Покрытие лжёт</a></li></ol></li><li><h3>Модуль 7 · Формальные методы</h3><p>Ассерты, верные каждый такт; ограниченный поиск контрпримера; и почему доказательству нужна индукция.</p><ol><li><a href="/ru/learn/assertions/">19 Ассерты</a></li><li><a href="/ru/learn/bounded-model-checking/">20 Ограниченная проверка моделей</a></li><li><a href="/ru/learn/induction/">21 Индукция</a></li></ol></li><li><h3>Модуль 8 · Мутации</h3><p>Ломайте дизайн нарочно и считайте, что заметили тесты.</p><ol><li><a href="/ru/learn/kill-the-mutant/">22 Убить мутанта</a></li><li><a href="/ru/learn/surviving-mutants/">23 Выжившие мутанты</a></li><li><a href="/ru/learn/test-strength/">24 Сила тестов</a></li></ol></li><li><h3>Модуль 9 · Приёмка</h3><p>Одна команда, все квитанции, чистый вердикт, который можно показать.</p><ol><li><a href="/ru/learn/build-verify/">25 Сборка и проверка</a></li><li><a href="/ru/learn/board-selftest/">26 Стенд проверяет себя</a></li><li><a href="/ru/learn/verify-capstone/">27 Капстоун: проверить модуль целиком</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">Этот курс сам является spec t27: прочитайте его</a></p><p>Виджеты сохраняют свои английские слова: каждый живёт в собственном spec t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/decode-in-one-step/index.html b/apps/website/public/ru/learn/decode-in-one-step/index.html
index ceec9d79fb..e3b4e724f2 100644
--- a/apps/website/public/ru/learn/decode-in-one-step/index.html
+++ b/apps/website/public/ru/learn/decode-in-one-step/index.html
@@ -191,5 +191,6 @@ <h1>Раскодировать за один шаг</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Правило и его числа</h3><p>Одно правило делит каждую ширину, отношение, к которому оно стремится, и числа Люка за тройкой 3.</p><ol><li><a href="/ru/learn/a-float-cut-by-phi/">01 Число с плавающей точкой, разрезанное по phi</a></li><li><a href="/ru/learn/phi-as-a-ratio/">02 Phi как отношение</a></li><li><a href="/ru/learn/lucas-numbers-stay-whole/">03 Числа Люка остаются целыми</a></li></ol></li><li><h3>Модуль 2 · Почему phi, почему три</h3><p>Почему деление идёт по phi, почему основание три и как спека проверяет, что GF16 хранит phi.</p><ol><li><a href="/ru/learn/why-the-split-is-phi/">04 Почему деление по phi</a></li><li><a href="/ru/learn/why-base-three/">05 Почему основание три</a></li><li><a href="/ru/learn/phi-in-sixteen-bits/">06 Phi в шестнадцати битах</a></li></ol></li><li><h3>Модуль 3 · Малые ступени: от GF4 до GF8</h3><p>GF4, GF6 и GF8 — меньше всего битов, и округление до целых битов стоит здесь дороже всего.</p><ol><li><a href="/ru/learn/gf4-four-bits/">07 GF4: четыре бита</a></li><li><a href="/ru/learn/gf6-six-bits/">08 GF6: шесть битов</a></li><li><a href="/ru/learn/gf8-one-byte/">09 GF8: один байт</a></li></ol></li><li><h3>Модуль 4 · От десяти до четырнадцати битов</h3><p>GF10, GF12 и GF14 и то, как расстояние до 1 / phi меняется с ростом слова.</p><ol><li><a href="/ru/learn/gf10-ten-bits/">10 GF10: десять битов</a></li><li><a href="/ru/learn/gf12-twelve-bits/">11 GF12: двенадцать битов</a></li><li><a href="/ru/learn/gf14-fourteen-bits/">12 GF14: четырнадцать битов</a></li></ol></li><li><h3>Модуль 5 · GF16 в работе</h3><p>Основной 16-битный формат, скалярное произведение из двух слагаемых в GF-T16, затем GF20 и GF24.</p><ol><li><a href="/ru/learn/gf16-the-primary/">13 GF16: основной формат</a></li><li><a href="/ru/learn/two-products-one-sum/">14 Два произведения, одна сумма</a></li><li><a href="/ru/learn/gf24-twenty-four-bits/">15 GF20 и GF24</a></li></ol></li><li><h3>Модуль 6 · От GF32 до GF64</h3><p>GF32 рядом с IEEE single, GF48 без пары в IEEE, GF64 рядом с IEEE double.</p><ol><li><a href="/ru/learn/gf32-a-single/">16 GF32: одинарная точность</a></li><li><a href="/ru/learn/gf48-forty-eight-bits/">17 GF48: сорок восемь битов</a></li><li><a href="/ru/learn/gf64-a-double/">18 GF64: двойная точность</a></li></ol></li><li><h3>Модуль 7 · От GF96 до GF256</h3><p>GF96, GF128 и GF256, где спеки держат раскладку инвариантами.</p><ol><li><a href="/ru/learn/gf96-ninety-six-bits/">19 GF96: девяносто шесть битов</a></li><li><a href="/ru/learn/gf128-a-quad/">20 GF128: четверная точность</a></li><li><a href="/ru/learn/gf256-two-five-six/">21 GF256: двести пятьдесят шесть битов</a></li></ol></li><li><h3>Модуль 8 · Самые широкие ступени, затем триты</h3><p>GF512 и GF1024, две самые широкие ступени, затем GF-T8, где порядок уходит в триты.</p><ol><li><a href="/ru/learn/gf512-five-twelve/">22 GF512: пятьсот двенадцать битов</a></li><li><a href="/ru/learn/gf1024-one-kilobit/">23 GF1024: один килобит</a></li><li><a href="/ru/learn/gft8-exponent-in-trits/">24 GF-T8: порядок в тритах</a></li></ol></li><li><h3>Модуль 9 · Ещё триты, затем декодирование</h3><p>GF-T16 и GF-T32, затем почему фиксированные поля декодируются параллельно, а posit — нет.</p><ol><li><a href="/ru/learn/gft16-six-trits/">25 GF-T16: шесть тритов</a></li><li><a href="/ru/learn/gft32-twelve-trits/">26 GF-T32: двенадцать тритов</a></li><li class="on" aria-current="page"><a href="/ru/learn/decode-in-one-step/">27 Раскодировать за один шаг</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/depth-sizing/index.html b/apps/website/public/ru/learn/depth-sizing/index.html
index 291a5dd366..04c9da8821 100644
--- a/apps/website/public/ru/learn/depth-sizing/index.html
+++ b/apps/website/public/ru/learn/depth-sizing/index.html
@@ -434,5 +434,6 @@ <h1>Выбор глубины</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое тактовый сигнал</h3><p>Один фронт, один мир: всё, что делит фронт тактового сигнала, делит и мир; период и джиттер настоящего фронта и место, где тактовый сигнал входит в плату.</p><ol><li><a href="/ru/learn/one-edge-one-world/">01 Один фронт, один мир</a></li><li><a href="/ru/learn/period-and-jitter/">02 Период, фронт и джиттер</a></li><li><a href="/ru/learn/where-the-clock-enters/">03 Где входит тактовый сигнал</a></li></ol></li><li><h3>Модуль 2 · Деревья тактового сигнала</h3><p>Перекос и задержка прохождения, глобальная сеть буферов и ловушка стробирования клока логикой.</p><ol><li><a href="/ru/learn/skew-and-insertion/">04 Перекос и задержка прохождения</a></li><li><a href="/ru/learn/global-buffers/">05 Глобальный буфер</a></li><li><a href="/ru/learn/gated-clock-trap/">06 Ловушка стробированного клока</a></li></ol></li><li><h3>Модуль 3 · PLL и MMCM</h3><p>Умножить и поделить один тактовый сигнал в другой, сдвинуть его фазу шагами VCO и какие тактовые сигналы анализатор считает родственными.</p><ol><li><a href="/ru/learn/multiply-and-divide/">07 Умножить и поделить</a></li><li><a href="/ru/learn/phase-and-lock/">08 Фаза и захват</a></li><li><a href="/ru/learn/related-or-not/">09 Родственные или чужие</a></li></ol></li><li><h3>Модуль 4 · Сбросы</h3><p>Ассертировать асинхронно, снимать синхронно: три вида сброса, конвейер снятия и дерево, которое растит сброс.</p><ol><li><a href="/ru/learn/sync-or-async/">10 Синхронный или асинхронный</a></li><li><a href="/ru/learn/reset-release/">11 Конвейер снятия</a></li><li><a href="/ru/learn/reset-fanout/">12 Разветвление сброса</a></li></ol></li><li><h3>Модуль 5 · Метастабильность</h3><p>Окно setup-hold, среднее время между отказами в целочисленной арифметике и два триггера, которые это чинят.</p><ol><li><a href="/ru/learn/setup-hold-window/">13 Окно setup-hold</a></li><li><a href="/ru/learn/mtbf/">14 Среднее время между отказами</a></li><li><a href="/ru/learn/two-flop-sync/">15 Двухтриггерный синхронизатор</a></li></ol></li><li><h3>Модуль 6 · Пересечение многих бит</h3><p>Почему двоичная шина рвётся, почему код Грея нет, и рукопожатие, переносящее импульс между мирами.</p><ol><li><a href="/ru/learn/why-binary-fails/">16 Почему двоичный код рвётся</a></li><li><a href="/ru/learn/gray-code/">17 Код Грея</a></li><li><a href="/ru/learn/handshakes-and-pulses/">18 Рукопожатия и импульсы</a></li></ol></li><li><h3>Модуль 7 · Асинхронный FIFO</h3><p>Указатели, флаги и глубина: буфер, переносящий поток между двумя тактовыми сигналами.</p><ol><li><a href="/ru/learn/pointers-and-flags/">19 Указатели и флаги</a></li><li class="on" aria-current="page"><a href="/ru/learn/depth-sizing/">20 Выбор глубины</a></li><li><a href="/ru/learn/fifo-testbench/">21 Тестбенч FIFO</a></li></ol></li><li><h3>Модуль 8 · Ограничения</h3><p>Строки, сообщающие анализатору, что такое тактовый сигнал, какие пути не проверять и чему должны соответствовать выводы.</p><ol><li><a href="/ru/learn/create-clock/">22 Строка create_clock</a></li><li><a href="/ru/learn/false-path-and-max-delay/">23 False path и max delay</a></li><li><a href="/ru/learn/io-timing/">24 Тайминг выводов</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Отчёт CDC, одно пересечение, захваченное на триггерах, и дифф битстрима, замыкающий курс.</p><ol><li><a href="/ru/learn/cdc-report/">25 Отчёт CDC</a></li><li><a href="/ru/learn/capture-a-crossing/">26 Захватить пересечение</a></li><li><a href="/ru/learn/capstone-diff/">27 Капстоун: поменяй и сравни</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/designing-taps/index.html b/apps/website/public/ru/learn/designing-taps/index.html
index 59a4c3b32c..8b8fd65db2 100644
--- a/apps/website/public/ru/learn/designing-taps/index.html
+++ b/apps/website/public/ru/learn/designing-taps/index.html
@@ -1122,5 +1122,6 @@ <h1>Расчёт коэффициентов</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Сложение</h3><p>Почему сложение медленное: перенос, который идёт вверх по слову, выделенная цепь в каждом слайсе, префиксные сети и деревья сохранения переноса.</p><ol><li><a href="/ru/learn/the-carry-chain/">01 Цепь переноса</a></li><li><a href="/ru/learn/prefix-adders/">02 Взгляд вперёд: префиксные сумматоры</a></li><li><a href="/ru/learn/carry-save/">03 Три числа сразу</a></li></ol></li><li><h3>Модуль 2 · Умножение</h3><p>Произведение — сумма сдвинутых копий; перекодирование Бута вдвое сокращает их, а слайс DSP48E1 делает остальное одним блоком.</p><ol><li><a href="/ru/learn/shift-and-add/">04 Сдвиг и сложение</a></li><li><a href="/ru/learn/booths-trick/">05 Приём Бута</a></li><li><a href="/ru/learn/the-dsp48e1-slice/">06 Слайс DSP48E1</a></li></ol></li><li><h3>Модуль 3 · Фиксированная точка</h3><p>Где стоит двоичная точка, что округление делает со значением и с его средним и что даёт каждый бит квантователя.</p><ol><li><a href="/ru/learn/q-formats/">07 Форматы Q</a></li><li><a href="/ru/learn/rounding-and-saturation/">08 Округление и насыщение</a></li><li><a href="/ru/learn/quantization-noise/">09 Шум квантования</a></li></ol></li><li><h3>Модуль 4 · Функции в железе</h3><p>Синус, косинус, угол и длина из сдвигов и сложений — и когда таблица оказывается выгоднее.</p><ol><li><a href="/ru/learn/rotating-by-shifts/">10 Поворот сдвигами</a></li><li><a href="/ru/learn/angle-and-magnitude/">11 Угол и модуль</a></li><li><a href="/ru/learn/tables-and-interpolation/">12 Таблицы и интерполяция</a></li></ol></li><li><h3>Модуль 5 · Сигналы и дискретизация</h3><p>Что дискретизация делает с частотой, как генератор строится из сумматора и что измеряет бин ДПФ.</p><ol><li><a href="/ru/learn/sampling-and-aliasing/">13 Дискретизация и наложение спектров</a></li><li><a href="/ru/learn/the-nco/">14 Генератор с числовым управлением</a></li><li><a href="/ru/learn/bins-of-a-spectrum/">15 Бины спектра</a></li></ol></li><li><h3>Модуль 6 · Фильтры FIR</h3><p>Скользящее среднее, расчёт окном sinc с целыми коэффициентами и раскладка отводов по слайсам DSP.</p><ol><li><a href="/ru/learn/the-moving-average/">16 Скользящее среднее</a></li><li class="on" aria-current="page"><a href="/ru/learn/designing-taps/">17 Расчёт коэффициентов</a></li><li><a href="/ru/learn/folding-taps-onto-slices/">18 Раскладка отводов по слайсам</a></li></ol></li><li><h3>Модуль 7 · Многоскоростная обработка</h3><p>Понижение частоты дискретизации без заворота шума: децимация, фильтр CIC и полифазная форма.</p><ol><li><a href="/ru/learn/lowering-the-rate/">19 Понижение частоты</a></li><li><a href="/ru/learn/the-cic-filter/">20 Фильтр CIC</a></li><li><a href="/ru/learn/polyphase/">21 Полифазная форма</a></li></ol></li><li><h3>Модуль 8 · БПФ</h3><p>N log N вместо N^2: бабочки, бит-реверсный порядок входа и биты, которые добавляет каждый этап.</p><ol><li><a href="/ru/learn/the-butterfly/">22 Бабочка</a></li><li><a href="/ru/learn/twiddles-and-bit-reversal/">23 Поворачивающие множители и бит-реверс</a></li><li><a href="/ru/learn/growth-through-the-stages/">24 Рост через этапы</a></li></ol></li><li><h3>Модуль 9 · На стенде</h3><p>От арифметики к плате: коррелятор из спеки модема, бюджет слайсов и тайминга и итоговый фильтр.</p><ol><li><a href="/ru/learn/a-correlator-finds-the-frame/">25 Коррелятор находит кадр</a></li><li><a href="/ru/learn/does-it-fit-and-close-timing/">26 Помещается ли и сходится ли тайминг?</a></li><li><a href="/ru/learn/capstone-a-filter-on-the-board/">27 Итог: фильтр на плате</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/directed-vs-random/index.html b/apps/website/public/ru/learn/directed-vs-random/index.html
index c8dc905480..cff4a7a47c 100644
--- a/apps/website/public/ru/learn/directed-vs-random/index.html
+++ b/apps/website/public/ru/learn/directed-vs-random/index.html
@@ -466,5 +466,6 @@ <h1>Направленные против случайных</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Зачем проверять</h3><p>Дизайны, которые компилируются и ошибаются; модель, которая выносит вердикт; план, записанный до кода.</p><ol><li><a href="/ru/learn/bugs-that-compile/">01 Баги, которые компилируются</a></li><li><a href="/ru/learn/golden-model/">02 Золотая модель</a></li><li><a href="/ru/learn/plan-before-code/">03 Сначала план</a></li></ol></li><li><h3>Модуль 2 · Тестбенчи</h3><p>Стимулы, проверки и вердикт, записанные как один spec рядом с дизайном, который они судят.</p><ol><li><a href="/ru/learn/anatomy-of-a-tb/">04 Анатомия тестбенча</a></li><li><a href="/ru/learn/self-checking/">05 Самопроверка вместо всматривания</a></li><li class="on" aria-current="page"><a href="/ru/learn/directed-vs-random/">06 Направленные против случайных</a></li></ol></li><li><h3>Модуль 3 · Временные диаграммы</h3><p>Трасса каждого сигнала, прочитанная так, как её читает инженер по железу, и два прогона, сравнённые между собой.</p><ol><li><a href="/ru/learn/vcd-format/">07 Формат VCD</a></li><li><a href="/ru/learn/debug-with-waves/">08 Отладка по волнам</a></li><li><a href="/ru/learn/compare-traces/">09 Сравнение двух трасс</a></li></ol></li><li><h3>Модуль 4 · Векторы соответствия</h3><p>Случаи с ответом, записанным рядом, там, откуда их достанет компилятор.</p><ol><li><a href="/ru/learn/what-a-vector-is/">10 Что такое вектор</a></li><li><a href="/ru/learn/golden-ruler-vectors/">11 Векторы как эталонная линейка</a></li><li><a href="/ru/learn/end-to-end-scenarios/">12 Сквозные сценарии</a></li></ol></li><li><h3>Модуль 5 · Косимуляция</h3><p>Spec, симулятор и плата сходятся в одном ответе на стенде Artix-7 XC7A200T, и что делать, когда не сходятся.</p><ol><li><a href="/ru/learn/three-way-agreement/">13 Тройное согласие</a></li><li><a href="/ru/learn/simulator-internals/">14 Внутри симулятора</a></li><li><a href="/ru/learn/when-they-disagree/">15 Когда они не сходятся</a></li></ol></li><li><h3>Модуль 6 · Покрытие</h3><p>Чего коснулись тесты: строки, переключения, состояния — и что прячет это число.</p><ol><li><a href="/ru/learn/line-and-toggle/">16 Строки и переключения</a></li><li><a href="/ru/learn/fsm-coverage/">17 Покрытие автоматов</a></li><li><a href="/ru/learn/coverage-lies/">18 Покрытие лжёт</a></li></ol></li><li><h3>Модуль 7 · Формальные методы</h3><p>Ассерты, верные каждый такт; ограниченный поиск контрпримера; и почему доказательству нужна индукция.</p><ol><li><a href="/ru/learn/assertions/">19 Ассерты</a></li><li><a href="/ru/learn/bounded-model-checking/">20 Ограниченная проверка моделей</a></li><li><a href="/ru/learn/induction/">21 Индукция</a></li></ol></li><li><h3>Модуль 8 · Мутации</h3><p>Ломайте дизайн нарочно и считайте, что заметили тесты.</p><ol><li><a href="/ru/learn/kill-the-mutant/">22 Убить мутанта</a></li><li><a href="/ru/learn/surviving-mutants/">23 Выжившие мутанты</a></li><li><a href="/ru/learn/test-strength/">24 Сила тестов</a></li></ol></li><li><h3>Модуль 9 · Приёмка</h3><p>Одна команда, все квитанции, чистый вердикт, который можно показать.</p><ol><li><a href="/ru/learn/build-verify/">25 Сборка и проверка</a></li><li><a href="/ru/learn/board-selftest/">26 Стенд проверяет себя</a></li><li><a href="/ru/learn/verify-capstone/">27 Капстоун: проверить модуль целиком</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">Этот курс сам является spec t27: прочитайте его</a></p><p>Виджеты сохраняют свои английские слова: каждый живёт в собственном spec t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/does-it-fit-and-close-timing/index.html b/apps/website/public/ru/learn/does-it-fit-and-close-timing/index.html
index 28406be411..383d17c538 100644
--- a/apps/website/public/ru/learn/does-it-fit-and-close-timing/index.html
+++ b/apps/website/public/ru/learn/does-it-fit-and-close-timing/index.html
@@ -1122,5 +1122,6 @@ <h1>Помещается ли и сходится ли тайминг?</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Сложение</h3><p>Почему сложение медленное: перенос, который идёт вверх по слову, выделенная цепь в каждом слайсе, префиксные сети и деревья сохранения переноса.</p><ol><li><a href="/ru/learn/the-carry-chain/">01 Цепь переноса</a></li><li><a href="/ru/learn/prefix-adders/">02 Взгляд вперёд: префиксные сумматоры</a></li><li><a href="/ru/learn/carry-save/">03 Три числа сразу</a></li></ol></li><li><h3>Модуль 2 · Умножение</h3><p>Произведение — сумма сдвинутых копий; перекодирование Бута вдвое сокращает их, а слайс DSP48E1 делает остальное одним блоком.</p><ol><li><a href="/ru/learn/shift-and-add/">04 Сдвиг и сложение</a></li><li><a href="/ru/learn/booths-trick/">05 Приём Бута</a></li><li><a href="/ru/learn/the-dsp48e1-slice/">06 Слайс DSP48E1</a></li></ol></li><li><h3>Модуль 3 · Фиксированная точка</h3><p>Где стоит двоичная точка, что округление делает со значением и с его средним и что даёт каждый бит квантователя.</p><ol><li><a href="/ru/learn/q-formats/">07 Форматы Q</a></li><li><a href="/ru/learn/rounding-and-saturation/">08 Округление и насыщение</a></li><li><a href="/ru/learn/quantization-noise/">09 Шум квантования</a></li></ol></li><li><h3>Модуль 4 · Функции в железе</h3><p>Синус, косинус, угол и длина из сдвигов и сложений — и когда таблица оказывается выгоднее.</p><ol><li><a href="/ru/learn/rotating-by-shifts/">10 Поворот сдвигами</a></li><li><a href="/ru/learn/angle-and-magnitude/">11 Угол и модуль</a></li><li><a href="/ru/learn/tables-and-interpolation/">12 Таблицы и интерполяция</a></li></ol></li><li><h3>Модуль 5 · Сигналы и дискретизация</h3><p>Что дискретизация делает с частотой, как генератор строится из сумматора и что измеряет бин ДПФ.</p><ol><li><a href="/ru/learn/sampling-and-aliasing/">13 Дискретизация и наложение спектров</a></li><li><a href="/ru/learn/the-nco/">14 Генератор с числовым управлением</a></li><li><a href="/ru/learn/bins-of-a-spectrum/">15 Бины спектра</a></li></ol></li><li><h3>Модуль 6 · Фильтры FIR</h3><p>Скользящее среднее, расчёт окном sinc с целыми коэффициентами и раскладка отводов по слайсам DSP.</p><ol><li><a href="/ru/learn/the-moving-average/">16 Скользящее среднее</a></li><li><a href="/ru/learn/designing-taps/">17 Расчёт коэффициентов</a></li><li><a href="/ru/learn/folding-taps-onto-slices/">18 Раскладка отводов по слайсам</a></li></ol></li><li><h3>Модуль 7 · Многоскоростная обработка</h3><p>Понижение частоты дискретизации без заворота шума: децимация, фильтр CIC и полифазная форма.</p><ol><li><a href="/ru/learn/lowering-the-rate/">19 Понижение частоты</a></li><li><a href="/ru/learn/the-cic-filter/">20 Фильтр CIC</a></li><li><a href="/ru/learn/polyphase/">21 Полифазная форма</a></li></ol></li><li><h3>Модуль 8 · БПФ</h3><p>N log N вместо N^2: бабочки, бит-реверсный порядок входа и биты, которые добавляет каждый этап.</p><ol><li><a href="/ru/learn/the-butterfly/">22 Бабочка</a></li><li><a href="/ru/learn/twiddles-and-bit-reversal/">23 Поворачивающие множители и бит-реверс</a></li><li><a href="/ru/learn/growth-through-the-stages/">24 Рост через этапы</a></li></ol></li><li><h3>Модуль 9 · На стенде</h3><p>От арифметики к плате: коррелятор из спеки модема, бюджет слайсов и тайминга и итоговый фильтр.</p><ol><li><a href="/ru/learn/a-correlator-finds-the-frame/">25 Коррелятор находит кадр</a></li><li class="on" aria-current="page"><a href="/ru/learn/does-it-fit-and-close-timing/">26 Помещается ли и сходится ли тайминг?</a></li><li><a href="/ru/learn/capstone-a-filter-on-the-board/">27 Итог: фильтр на плате</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/end-to-end-scenarios/index.html b/apps/website/public/ru/learn/end-to-end-scenarios/index.html
index a82c2c44de..4818a1fa2f 100644
--- a/apps/website/public/ru/learn/end-to-end-scenarios/index.html
+++ b/apps/website/public/ru/learn/end-to-end-scenarios/index.html
@@ -322,5 +322,6 @@ <h1>Сквозные сценарии</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Зачем проверять</h3><p>Дизайны, которые компилируются и ошибаются; модель, которая выносит вердикт; план, записанный до кода.</p><ol><li><a href="/ru/learn/bugs-that-compile/">01 Баги, которые компилируются</a></li><li><a href="/ru/learn/golden-model/">02 Золотая модель</a></li><li><a href="/ru/learn/plan-before-code/">03 Сначала план</a></li></ol></li><li><h3>Модуль 2 · Тестбенчи</h3><p>Стимулы, проверки и вердикт, записанные как один spec рядом с дизайном, который они судят.</p><ol><li><a href="/ru/learn/anatomy-of-a-tb/">04 Анатомия тестбенча</a></li><li><a href="/ru/learn/self-checking/">05 Самопроверка вместо всматривания</a></li><li><a href="/ru/learn/directed-vs-random/">06 Направленные против случайных</a></li></ol></li><li><h3>Модуль 3 · Временные диаграммы</h3><p>Трасса каждого сигнала, прочитанная так, как её читает инженер по железу, и два прогона, сравнённые между собой.</p><ol><li><a href="/ru/learn/vcd-format/">07 Формат VCD</a></li><li><a href="/ru/learn/debug-with-waves/">08 Отладка по волнам</a></li><li><a href="/ru/learn/compare-traces/">09 Сравнение двух трасс</a></li></ol></li><li><h3>Модуль 4 · Векторы соответствия</h3><p>Случаи с ответом, записанным рядом, там, откуда их достанет компилятор.</p><ol><li><a href="/ru/learn/what-a-vector-is/">10 Что такое вектор</a></li><li><a href="/ru/learn/golden-ruler-vectors/">11 Векторы как эталонная линейка</a></li><li class="on" aria-current="page"><a href="/ru/learn/end-to-end-scenarios/">12 Сквозные сценарии</a></li></ol></li><li><h3>Модуль 5 · Косимуляция</h3><p>Spec, симулятор и плата сходятся в одном ответе на стенде Artix-7 XC7A200T, и что делать, когда не сходятся.</p><ol><li><a href="/ru/learn/three-way-agreement/">13 Тройное согласие</a></li><li><a href="/ru/learn/simulator-internals/">14 Внутри симулятора</a></li><li><a href="/ru/learn/when-they-disagree/">15 Когда они не сходятся</a></li></ol></li><li><h3>Модуль 6 · Покрытие</h3><p>Чего коснулись тесты: строки, переключения, состояния — и что прячет это число.</p><ol><li><a href="/ru/learn/line-and-toggle/">16 Строки и переключения</a></li><li><a href="/ru/learn/fsm-coverage/">17 Покрытие автоматов</a></li><li><a href="/ru/learn/coverage-lies/">18 Покрытие лжёт</a></li></ol></li><li><h3>Модуль 7 · Формальные методы</h3><p>Ассерты, верные каждый такт; ограниченный поиск контрпримера; и почему доказательству нужна индукция.</p><ol><li><a href="/ru/learn/assertions/">19 Ассерты</a></li><li><a href="/ru/learn/bounded-model-checking/">20 Ограниченная проверка моделей</a></li><li><a href="/ru/learn/induction/">21 Индукция</a></li></ol></li><li><h3>Модуль 8 · Мутации</h3><p>Ломайте дизайн нарочно и считайте, что заметили тесты.</p><ol><li><a href="/ru/learn/kill-the-mutant/">22 Убить мутанта</a></li><li><a href="/ru/learn/surviving-mutants/">23 Выжившие мутанты</a></li><li><a href="/ru/learn/test-strength/">24 Сила тестов</a></li></ol></li><li><h3>Модуль 9 · Приёмка</h3><p>Одна команда, все квитанции, чистый вердикт, который можно показать.</p><ol><li><a href="/ru/learn/build-verify/">25 Сборка и проверка</a></li><li><a href="/ru/learn/board-selftest/">26 Стенд проверяет себя</a></li><li><a href="/ru/learn/verify-capstone/">27 Капстоун: проверить модуль целиком</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">Этот курс сам является spec t27: прочитайте его</a></p><p>Виджеты сохраняют свои английские слова: каждый живёт в собственном spec t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/enums/index.html b/apps/website/public/ru/learn/enums/index.html
index e7c90e95e6..b8a5e4d072 100644
--- a/apps/website/public/ru/learn/enums/index.html
+++ b/apps/website/public/ru/learn/enums/index.html
@@ -92,5 +92,6 @@ <h1>Перечисления</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Файл спеки</h3><p>Что лежит в файле .t27: строка module, строки прозы и комментарии.</p><ol><li><a href="/ru/learn/what-a-spec-is/">01 Что такое спека</a></li><li><a href="/ru/learn/the-module-line/">02 Строка module</a></li><li><a href="/ru/learn/comments-and-prose/">03 Комментарии и строки прозы</a></li></ol></li><li><h3>Модуль 2 · Значения и типы</h3><p>Константы, ширина целого числа, а также true, false и текст.</p><ol><li><a href="/ru/learn/constants/">04 Константы</a></li><li><a href="/ru/learn/whole-numbers/">05 Целые числа и их ширина</a></li><li><a href="/ru/learn/true-false-and-text/">06 Истина, ложь и текст</a></li></ol></li><li><h3>Модуль 3 · Массивы, триты и выражения</h3><p>Списки значений, три значения трита и что вычисляют операторы.</p><ol><li><a href="/ru/learn/arrays/">07 Массивы</a></li><li><a href="/ru/learn/trits/">08 Триты: минус один, ноль, один</a></li><li><a href="/ru/learn/expressions/">09 Выражения</a></li></ol></li><li><h3>Модуль 4 · Тесты и функции</h3><p>Блок test, несколько маленьких тестов в одной спеке и функция с типизированными входами.</p><ol><li><a href="/ru/learn/a-test-block/">10 Блок test</a></li><li><a href="/ru/learn/many-tests/">11 Много тестов в одной спеке</a></li><li><a href="/ru/learn/functions/">12 Функции</a></li></ol></li><li><h3>Модуль 5 · Внутри функции</h3><p>Имена через let и var, выбор через if и switch и циклы.</p><ol><li><a href="/ru/learn/local-names/">13 Локальные имена: let и var</a></li><li><a href="/ru/learn/if-and-else/">14 Выбор: if, else и switch</a></li><li><a href="/ru/learn/loops/">15 Циклы: while и for</a></li></ol></li><li><h3>Модуль 6 · Видимость и свои типы</h3><p>Что pub открывает другим модулям, и структуры и перечисления, которые вы объявляете.</p><ol><li><a href="/ru/learn/pub-or-private/">16 pub или закрытое имя</a></li><li><a href="/ru/learn/structs/">17 Структуры</a></li><li class="on" aria-current="page"><a href="/ru/learn/enums/">18 Перечисления</a></li></ol></li><li><h3>Модуль 7 · Модули, правила и gen-ts</h3><p>Чтение другого модуля, правило, которое выполняется всегда, и спека, превращённая в TypeScript.</p><ol><li><a href="/ru/learn/use-other-modules/">19 Другие модули</a></li><li><a href="/ru/learn/invariants/">20 Инварианты и замеры</a></li><li><a href="/ru/learn/gen-ts/">21 gen-ts: спека превращается в TypeScript</a></li></ol></li><li><h3>Модуль 8 · Компилятор и tri</h3><p>Бэкенды, которые пишет t27c, стадии, в которых он читает спеку, и команда tri.</p><ol><li><a href="/ru/learn/seven-backends/">22 Семь бэкендов</a></li><li><a href="/ru/learn/stages-of-t27c/">23 Стадии t27c</a></li><li><a href="/ru/learn/tri-commands/">24 tri: командная строка</a></li></ol></li><li><h3>Модуль 9 · Ошибки, программа и что дальше</h3><p>Как читать ошибку компилятора, одна маленькая законченная программа и какой курс пройти дальше.</p><ol><li><a href="/ru/learn/reading-errors/">25 Как читать ошибки компилятора</a></li><li><a href="/ru/learn/a-small-program/">26 Маленькая законченная программа</a></li><li><a href="/ru/learn/where-next/">27 Куда дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/everything-in-one-bench/index.html b/apps/website/public/ru/learn/everything-in-one-bench/index.html
index 222d98e3e6..ccba531587 100644
--- a/apps/website/public/ru/learn/everything-in-one-bench/index.html
+++ b/apps/website/public/ru/learn/everything-in-one-bench/index.html
@@ -202,5 +202,6 @@ <h1>Всё в одном стенде</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое шина</h3><p>Зачем вообще существует шина: разговор по проводам, с кадрами и адресами, и кому позволено говорить.</p><ol><li><a href="/ru/learn/wires-that-talk/">01 Провода, которые говорят</a></li><li><a href="/ru/learn/framing-a-conversation/">02 Кадрирование разговора</a></li><li><a href="/ru/learn/master-and-servant/">03 Хозяин и слуга</a></li></ol></li><li><h3>Модуль 2 · Разговор по UART</h3><p>Двухпроводная шина без такта: кадр, делитель, задающий скорость, и статус, который опрашивает драйвер.</p><ol><li><a href="/ru/learn/the-frame/">04 Кадр: старт, данные, стоп</a></li><li><a href="/ru/learn/baud-and-the-divisor/">05 Бод и делитель</a></li><li><a href="/ru/learn/status-and-fifos/">06 Статус и FIFO</a></li></ol></li><li><h3>Модуль 3 · SPI под тактом</h3><p>Разговор под тактом: четыре режима, лестница предделителей и выбор кристалла на каждого подчинённого.</p><ol><li><a href="/ru/learn/modes-cpol-cpha/">07 Режимы: CPOL и CPHA</a></li><li><a href="/ru/learn/the-prescaler-ladder/">08 Лестница предделителей</a></li><li><a href="/ru/learn/chip-select-and-widths/">09 Выбор кристалла и ширины</a></li></ol></li><li><h3>Модуль 4 · Регистровая шина APB</h3><p>Шина регистров: PSEL и PENABLE, строобы и ожидания, и сколько адресных бит стоит количество периферии.</p><ol><li><a href="/ru/learn/psel-and-penable/">10 PSEL и PENABLE</a></li><li><a href="/ru/learn/strobes-and-waits/">11 Строобы и ожидания</a></li><li><a href="/ru/learn/address-decode/">12 Дешифрация адреса</a></li></ol></li><li><h3>Модуль 5 · Пять каналов AXI4</h3><p>Пять каналов: адрес, данные и отклик в обе стороны, lite или full, пакеты и идентификаторы.</p><ol><li><a href="/ru/learn/five-channels/">13 Пять каналов</a></li><li><a href="/ru/learn/lite-or-full/">14 Lite или full</a></li><li><a href="/ru/learn/bursts-and-ids/">15 Пакеты и идентификаторы</a></li></ol></li><li><h3>Модуль 6 · Память</h3><p>Что находится на дальнем конце каждой шины: карты памяти, типы портов и задержка, которую должно покрывать ожидание.</p><ol><li><a href="/ru/learn/memory-maps/">16 Карты памяти</a></li><li><a href="/ru/learn/ports-and-kinds/">17 Порты и типы</a></li><li><a href="/ru/learn/latency/">18 Задержка</a></li></ol></li><li><h3>Модуль 7 · Мосты</h3><p>Почему в дизайне появляется больше одной шины и пакетный мост, переносящий работу между ними.</p><ol><li><a href="/ru/learn/why-bridges/">19 Зачем мосты</a></li><li><a href="/ru/learn/the-packet-bridge/">20 Пакетный мост</a></li><li class="on" aria-current="page"><a href="/ru/learn/everything-in-one-bench/">21 Всё в одном стенде</a></li></ol></li><li><h3>Модуль 8 · Ethernet: кадры и тайминг</h3><p>Кадры, контрольная сумма кадра, тайминг RGMII и заранее зарегистрированные шаги настоящего запуска.</p><ol><li><a href="/ru/learn/frames-and-the-fcs/">22 Кадры и контрольная сумма</a></li><li><a href="/ru/learn/rgmii-timing/">23 Тайминг RGMII</a></li><li><a href="/ru/learn/bring-up-on-the-bench/">24 Запуск на стенде</a></li></ol></li><li><h3>Модуль 9 · Стенд</h3><p>Дисциплина, оберегающая настоящее железо: кто держит IO, как взять и вернуть захват, и что запускается дальше.</p><ol><li><a href="/ru/learn/who-holds-the-io/">25 Кто держит IO</a></li><li><a href="/ru/learn/claim-and-release/">26 Захват и возврат</a></li><li><a href="/ru/learn/capstone-what-runs-next/">27 Капстоун: что запускается дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/expressions/index.html b/apps/website/public/ru/learn/expressions/index.html
index f5651d3798..a6ee62d614 100644
--- a/apps/website/public/ru/learn/expressions/index.html
+++ b/apps/website/public/ru/learn/expressions/index.html
@@ -92,5 +92,6 @@ <h1>Выражения</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Файл спеки</h3><p>Что лежит в файле .t27: строка module, строки прозы и комментарии.</p><ol><li><a href="/ru/learn/what-a-spec-is/">01 Что такое спека</a></li><li><a href="/ru/learn/the-module-line/">02 Строка module</a></li><li><a href="/ru/learn/comments-and-prose/">03 Комментарии и строки прозы</a></li></ol></li><li><h3>Модуль 2 · Значения и типы</h3><p>Константы, ширина целого числа, а также true, false и текст.</p><ol><li><a href="/ru/learn/constants/">04 Константы</a></li><li><a href="/ru/learn/whole-numbers/">05 Целые числа и их ширина</a></li><li><a href="/ru/learn/true-false-and-text/">06 Истина, ложь и текст</a></li></ol></li><li><h3>Модуль 3 · Массивы, триты и выражения</h3><p>Списки значений, три значения трита и что вычисляют операторы.</p><ol><li><a href="/ru/learn/arrays/">07 Массивы</a></li><li><a href="/ru/learn/trits/">08 Триты: минус один, ноль, один</a></li><li class="on" aria-current="page"><a href="/ru/learn/expressions/">09 Выражения</a></li></ol></li><li><h3>Модуль 4 · Тесты и функции</h3><p>Блок test, несколько маленьких тестов в одной спеке и функция с типизированными входами.</p><ol><li><a href="/ru/learn/a-test-block/">10 Блок test</a></li><li><a href="/ru/learn/many-tests/">11 Много тестов в одной спеке</a></li><li><a href="/ru/learn/functions/">12 Функции</a></li></ol></li><li><h3>Модуль 5 · Внутри функции</h3><p>Имена через let и var, выбор через if и switch и циклы.</p><ol><li><a href="/ru/learn/local-names/">13 Локальные имена: let и var</a></li><li><a href="/ru/learn/if-and-else/">14 Выбор: if, else и switch</a></li><li><a href="/ru/learn/loops/">15 Циклы: while и for</a></li></ol></li><li><h3>Модуль 6 · Видимость и свои типы</h3><p>Что pub открывает другим модулям, и структуры и перечисления, которые вы объявляете.</p><ol><li><a href="/ru/learn/pub-or-private/">16 pub или закрытое имя</a></li><li><a href="/ru/learn/structs/">17 Структуры</a></li><li><a href="/ru/learn/enums/">18 Перечисления</a></li></ol></li><li><h3>Модуль 7 · Модули, правила и gen-ts</h3><p>Чтение другого модуля, правило, которое выполняется всегда, и спека, превращённая в TypeScript.</p><ol><li><a href="/ru/learn/use-other-modules/">19 Другие модули</a></li><li><a href="/ru/learn/invariants/">20 Инварианты и замеры</a></li><li><a href="/ru/learn/gen-ts/">21 gen-ts: спека превращается в TypeScript</a></li></ol></li><li><h3>Модуль 8 · Компилятор и tri</h3><p>Бэкенды, которые пишет t27c, стадии, в которых он читает спеку, и команда tri.</p><ol><li><a href="/ru/learn/seven-backends/">22 Семь бэкендов</a></li><li><a href="/ru/learn/stages-of-t27c/">23 Стадии t27c</a></li><li><a href="/ru/learn/tri-commands/">24 tri: командная строка</a></li></ol></li><li><h3>Модуль 9 · Ошибки, программа и что дальше</h3><p>Как читать ошибку компилятора, одна маленькая законченная программа и какой курс пройти дальше.</p><ol><li><a href="/ru/learn/reading-errors/">25 Как читать ошибки компилятора</a></li><li><a href="/ru/learn/a-small-program/">26 Маленькая законченная программа</a></li><li><a href="/ru/learn/where-next/">27 Куда дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/false-path-and-max-delay/index.html b/apps/website/public/ru/learn/false-path-and-max-delay/index.html
index 7de806951c..d0e369cc29 100644
--- a/apps/website/public/ru/learn/false-path-and-max-delay/index.html
+++ b/apps/website/public/ru/learn/false-path-and-max-delay/index.html
@@ -448,5 +448,6 @@ <h1>False path и max delay</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое тактовый сигнал</h3><p>Один фронт, один мир: всё, что делит фронт тактового сигнала, делит и мир; период и джиттер настоящего фронта и место, где тактовый сигнал входит в плату.</p><ol><li><a href="/ru/learn/one-edge-one-world/">01 Один фронт, один мир</a></li><li><a href="/ru/learn/period-and-jitter/">02 Период, фронт и джиттер</a></li><li><a href="/ru/learn/where-the-clock-enters/">03 Где входит тактовый сигнал</a></li></ol></li><li><h3>Модуль 2 · Деревья тактового сигнала</h3><p>Перекос и задержка прохождения, глобальная сеть буферов и ловушка стробирования клока логикой.</p><ol><li><a href="/ru/learn/skew-and-insertion/">04 Перекос и задержка прохождения</a></li><li><a href="/ru/learn/global-buffers/">05 Глобальный буфер</a></li><li><a href="/ru/learn/gated-clock-trap/">06 Ловушка стробированного клока</a></li></ol></li><li><h3>Модуль 3 · PLL и MMCM</h3><p>Умножить и поделить один тактовый сигнал в другой, сдвинуть его фазу шагами VCO и какие тактовые сигналы анализатор считает родственными.</p><ol><li><a href="/ru/learn/multiply-and-divide/">07 Умножить и поделить</a></li><li><a href="/ru/learn/phase-and-lock/">08 Фаза и захват</a></li><li><a href="/ru/learn/related-or-not/">09 Родственные или чужие</a></li></ol></li><li><h3>Модуль 4 · Сбросы</h3><p>Ассертировать асинхронно, снимать синхронно: три вида сброса, конвейер снятия и дерево, которое растит сброс.</p><ol><li><a href="/ru/learn/sync-or-async/">10 Синхронный или асинхронный</a></li><li><a href="/ru/learn/reset-release/">11 Конвейер снятия</a></li><li><a href="/ru/learn/reset-fanout/">12 Разветвление сброса</a></li></ol></li><li><h3>Модуль 5 · Метастабильность</h3><p>Окно setup-hold, среднее время между отказами в целочисленной арифметике и два триггера, которые это чинят.</p><ol><li><a href="/ru/learn/setup-hold-window/">13 Окно setup-hold</a></li><li><a href="/ru/learn/mtbf/">14 Среднее время между отказами</a></li><li><a href="/ru/learn/two-flop-sync/">15 Двухтриггерный синхронизатор</a></li></ol></li><li><h3>Модуль 6 · Пересечение многих бит</h3><p>Почему двоичная шина рвётся, почему код Грея нет, и рукопожатие, переносящее импульс между мирами.</p><ol><li><a href="/ru/learn/why-binary-fails/">16 Почему двоичный код рвётся</a></li><li><a href="/ru/learn/gray-code/">17 Код Грея</a></li><li><a href="/ru/learn/handshakes-and-pulses/">18 Рукопожатия и импульсы</a></li></ol></li><li><h3>Модуль 7 · Асинхронный FIFO</h3><p>Указатели, флаги и глубина: буфер, переносящий поток между двумя тактовыми сигналами.</p><ol><li><a href="/ru/learn/pointers-and-flags/">19 Указатели и флаги</a></li><li><a href="/ru/learn/depth-sizing/">20 Выбор глубины</a></li><li><a href="/ru/learn/fifo-testbench/">21 Тестбенч FIFO</a></li></ol></li><li><h3>Модуль 8 · Ограничения</h3><p>Строки, сообщающие анализатору, что такое тактовый сигнал, какие пути не проверять и чему должны соответствовать выводы.</p><ol><li><a href="/ru/learn/create-clock/">22 Строка create_clock</a></li><li class="on" aria-current="page"><a href="/ru/learn/false-path-and-max-delay/">23 False path и max delay</a></li><li><a href="/ru/learn/io-timing/">24 Тайминг выводов</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Отчёт CDC, одно пересечение, захваченное на триггерах, и дифф битстрима, замыкающий курс.</p><ol><li><a href="/ru/learn/cdc-report/">25 Отчёт CDC</a></li><li><a href="/ru/learn/capture-a-crossing/">26 Захватить пересечение</a></li><li><a href="/ru/learn/capstone-diff/">27 Капстоун: поменяй и сравни</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/fasm-to-frames/index.html b/apps/website/public/ru/learn/fasm-to-frames/index.html
index 7fc06b6eaf..167a4fc36d 100644
--- a/apps/website/public/ru/learn/fasm-to-frames/index.html
+++ b/apps/website/public/ru/learn/fasm-to-frames/index.html
@@ -110,5 +110,6 @@ <h1>От FASM к фреймам</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Чип</h3><p>Что такое FPGA, какой чип у нас и как его выводы встречаются с платой.</p><ol><li><a href="/ru/learn/fpga-is-a-grid/">01 FPGA — это сетка, которую вы настраиваете</a></li><li><a href="/ru/learn/which-chip/">02 Какой это чип?</a></li><li><a href="/ru/learn/pins-and-package/">03 Выводы, шарики и корпус</a></li></ol></li><li><h3>Модуль 2 · Числа в железе</h3><p>Биты, триты и форматы чисел, и сколько стоит арифметика в логике.</p><ol><li><a href="/ru/learn/bits-and-trits/">04 Биты, триты и GF16</a></li><li><a href="/ru/learn/what-arithmetic-costs/">05 Сколько стоит арифметика</a></li><li><a href="/ru/learn/how-much-fits/">06 Сколько помещается на чип</a></li></ol></li><li><h3>Модуль 3 · Ваша программа на t27</h3><p>Напишите спеку, проверьте её и увидьте, почему «компилируется» не значит «верно».</p><ol><li><a href="/ru/learn/hello-t27/">07 Привет, t27</a></li><li><a href="/ru/learn/tests-are-the-spec/">08 Тесты — часть спеки</a></li><li><a href="/ru/learn/compiles-is-not-correct/">09 Компилируется — не значит верно</a></li></ol></li><li><h3>Модуль 4 · Внутри t27c</h3><p>Как компилятор читает спеку и что он пишет, включая нативный t27b.</p><ol><li><a href="/ru/learn/compiler-stages/">10 Стадии t27c</a></li><li><a href="/ru/learn/t27c-on-a-real-machine/">11 t27c на настоящей машине</a></li><li><a href="/ru/learn/seven-backends-and-t27b/">12 Семь бэкендов и t27b</a></li></ol></li><li><h3>Модуль 5 · От спеки к железу</h3><p>Verilog, который пишет t27c, ячейки, в которые он превращается, и что делает одна LUT.</p><ol><li><a href="/ru/learn/spec-to-verilog/">13 От спеки к Verilog</a></li><li><a href="/ru/learn/count-the-cells/">14 Посчитайте ячейки</a></li><li><a href="/ru/learn/one-lut-up-close/">15 Одна LUT вблизи</a></li></ol></li><li><h3>Модуль 6 · Читаем синтез</h3><p>Что yosys сообщает о вашем дизайне и какие предупреждения важны.</p><ol><li><a href="/ru/learn/synthesis-warnings/">16 Важные предупреждения синтеза</a></li><li><a href="/ru/learn/state-machines/">17 Конечные автоматы</a></li><li><a href="/ru/learn/where-the-luts-go/">18 Куда уходят LUT</a></li></ol></li><li><h3>Модуль 7 · Размещение, трассировка, тайминг</h3><p>Где ячейки оказываются на кристалле и успевает ли тактовая частота.</p><ol><li><a href="/ru/learn/placement/">19 Размещение: где оказываются ячейки</a></li><li><a href="/ru/learn/logic-depth/">20 Глубина логики</a></li><li><a href="/ru/learn/timing-slack/">21 Тайминг и запас</a></li></ol></li><li><h3>Модуль 8 · Битстрим</h3><p>Как оттрассированный дизайн становится битами, которые загружает чип, без Vivado.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/fasm-to-frames/">22 От FASM к фреймам</a></li><li><a href="/ru/learn/a-build-receipt/">23 Сборка по пунктам</a></li><li><a href="/ru/learn/read-a-bitstream/">24 Читаем битстрим</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Спросите чип, кто он, загрузите биты и проверьте их.</p><ol><li><a href="/ru/learn/jtag-asks-the-chip/">25 JTAG спрашивает чип</a></li><li><a href="/ru/learn/same-bits-on-the-board/">26 Те же биты — на плате</a></li><li><a href="/ru/learn/capstone/">27 Итог: переугадайте yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/fifo-testbench/index.html b/apps/website/public/ru/learn/fifo-testbench/index.html
index cd6a8de546..0aa3914bc2 100644
--- a/apps/website/public/ru/learn/fifo-testbench/index.html
+++ b/apps/website/public/ru/learn/fifo-testbench/index.html
@@ -434,5 +434,6 @@ <h1>Тестбенч FIFO</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое тактовый сигнал</h3><p>Один фронт, один мир: всё, что делит фронт тактового сигнала, делит и мир; период и джиттер настоящего фронта и место, где тактовый сигнал входит в плату.</p><ol><li><a href="/ru/learn/one-edge-one-world/">01 Один фронт, один мир</a></li><li><a href="/ru/learn/period-and-jitter/">02 Период, фронт и джиттер</a></li><li><a href="/ru/learn/where-the-clock-enters/">03 Где входит тактовый сигнал</a></li></ol></li><li><h3>Модуль 2 · Деревья тактового сигнала</h3><p>Перекос и задержка прохождения, глобальная сеть буферов и ловушка стробирования клока логикой.</p><ol><li><a href="/ru/learn/skew-and-insertion/">04 Перекос и задержка прохождения</a></li><li><a href="/ru/learn/global-buffers/">05 Глобальный буфер</a></li><li><a href="/ru/learn/gated-clock-trap/">06 Ловушка стробированного клока</a></li></ol></li><li><h3>Модуль 3 · PLL и MMCM</h3><p>Умножить и поделить один тактовый сигнал в другой, сдвинуть его фазу шагами VCO и какие тактовые сигналы анализатор считает родственными.</p><ol><li><a href="/ru/learn/multiply-and-divide/">07 Умножить и поделить</a></li><li><a href="/ru/learn/phase-and-lock/">08 Фаза и захват</a></li><li><a href="/ru/learn/related-or-not/">09 Родственные или чужие</a></li></ol></li><li><h3>Модуль 4 · Сбросы</h3><p>Ассертировать асинхронно, снимать синхронно: три вида сброса, конвейер снятия и дерево, которое растит сброс.</p><ol><li><a href="/ru/learn/sync-or-async/">10 Синхронный или асинхронный</a></li><li><a href="/ru/learn/reset-release/">11 Конвейер снятия</a></li><li><a href="/ru/learn/reset-fanout/">12 Разветвление сброса</a></li></ol></li><li><h3>Модуль 5 · Метастабильность</h3><p>Окно setup-hold, среднее время между отказами в целочисленной арифметике и два триггера, которые это чинят.</p><ol><li><a href="/ru/learn/setup-hold-window/">13 Окно setup-hold</a></li><li><a href="/ru/learn/mtbf/">14 Среднее время между отказами</a></li><li><a href="/ru/learn/two-flop-sync/">15 Двухтриггерный синхронизатор</a></li></ol></li><li><h3>Модуль 6 · Пересечение многих бит</h3><p>Почему двоичная шина рвётся, почему код Грея нет, и рукопожатие, переносящее импульс между мирами.</p><ol><li><a href="/ru/learn/why-binary-fails/">16 Почему двоичный код рвётся</a></li><li><a href="/ru/learn/gray-code/">17 Код Грея</a></li><li><a href="/ru/learn/handshakes-and-pulses/">18 Рукопожатия и импульсы</a></li></ol></li><li><h3>Модуль 7 · Асинхронный FIFO</h3><p>Указатели, флаги и глубина: буфер, переносящий поток между двумя тактовыми сигналами.</p><ol><li><a href="/ru/learn/pointers-and-flags/">19 Указатели и флаги</a></li><li><a href="/ru/learn/depth-sizing/">20 Выбор глубины</a></li><li class="on" aria-current="page"><a href="/ru/learn/fifo-testbench/">21 Тестбенч FIFO</a></li></ol></li><li><h3>Модуль 8 · Ограничения</h3><p>Строки, сообщающие анализатору, что такое тактовый сигнал, какие пути не проверять и чему должны соответствовать выводы.</p><ol><li><a href="/ru/learn/create-clock/">22 Строка create_clock</a></li><li><a href="/ru/learn/false-path-and-max-delay/">23 False path и max delay</a></li><li><a href="/ru/learn/io-timing/">24 Тайминг выводов</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Отчёт CDC, одно пересечение, захваченное на триггерах, и дифф битстрима, замыкающий курс.</p><ol><li><a href="/ru/learn/cdc-report/">25 Отчёт CDC</a></li><li><a href="/ru/learn/capture-a-crossing/">26 Захватить пересечение</a></li><li><a href="/ru/learn/capstone-diff/">27 Капстоун: поменяй и сравни</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/five-channels/index.html b/apps/website/public/ru/learn/five-channels/index.html
index 52eda98b7b..8136bcc1ab 100644
--- a/apps/website/public/ru/learn/five-channels/index.html
+++ b/apps/website/public/ru/learn/five-channels/index.html
@@ -463,5 +463,6 @@ <h1>Пять каналов</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое шина</h3><p>Зачем вообще существует шина: разговор по проводам, с кадрами и адресами, и кому позволено говорить.</p><ol><li><a href="/ru/learn/wires-that-talk/">01 Провода, которые говорят</a></li><li><a href="/ru/learn/framing-a-conversation/">02 Кадрирование разговора</a></li><li><a href="/ru/learn/master-and-servant/">03 Хозяин и слуга</a></li></ol></li><li><h3>Модуль 2 · Разговор по UART</h3><p>Двухпроводная шина без такта: кадр, делитель, задающий скорость, и статус, который опрашивает драйвер.</p><ol><li><a href="/ru/learn/the-frame/">04 Кадр: старт, данные, стоп</a></li><li><a href="/ru/learn/baud-and-the-divisor/">05 Бод и делитель</a></li><li><a href="/ru/learn/status-and-fifos/">06 Статус и FIFO</a></li></ol></li><li><h3>Модуль 3 · SPI под тактом</h3><p>Разговор под тактом: четыре режима, лестница предделителей и выбор кристалла на каждого подчинённого.</p><ol><li><a href="/ru/learn/modes-cpol-cpha/">07 Режимы: CPOL и CPHA</a></li><li><a href="/ru/learn/the-prescaler-ladder/">08 Лестница предделителей</a></li><li><a href="/ru/learn/chip-select-and-widths/">09 Выбор кристалла и ширины</a></li></ol></li><li><h3>Модуль 4 · Регистровая шина APB</h3><p>Шина регистров: PSEL и PENABLE, строобы и ожидания, и сколько адресных бит стоит количество периферии.</p><ol><li><a href="/ru/learn/psel-and-penable/">10 PSEL и PENABLE</a></li><li><a href="/ru/learn/strobes-and-waits/">11 Строобы и ожидания</a></li><li><a href="/ru/learn/address-decode/">12 Дешифрация адреса</a></li></ol></li><li><h3>Модуль 5 · Пять каналов AXI4</h3><p>Пять каналов: адрес, данные и отклик в обе стороны, lite или full, пакеты и идентификаторы.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/five-channels/">13 Пять каналов</a></li><li><a href="/ru/learn/lite-or-full/">14 Lite или full</a></li><li><a href="/ru/learn/bursts-and-ids/">15 Пакеты и идентификаторы</a></li></ol></li><li><h3>Модуль 6 · Память</h3><p>Что находится на дальнем конце каждой шины: карты памяти, типы портов и задержка, которую должно покрывать ожидание.</p><ol><li><a href="/ru/learn/memory-maps/">16 Карты памяти</a></li><li><a href="/ru/learn/ports-and-kinds/">17 Порты и типы</a></li><li><a href="/ru/learn/latency/">18 Задержка</a></li></ol></li><li><h3>Модуль 7 · Мосты</h3><p>Почему в дизайне появляется больше одной шины и пакетный мост, переносящий работу между ними.</p><ol><li><a href="/ru/learn/why-bridges/">19 Зачем мосты</a></li><li><a href="/ru/learn/the-packet-bridge/">20 Пакетный мост</a></li><li><a href="/ru/learn/everything-in-one-bench/">21 Всё в одном стенде</a></li></ol></li><li><h3>Модуль 8 · Ethernet: кадры и тайминг</h3><p>Кадры, контрольная сумма кадра, тайминг RGMII и заранее зарегистрированные шаги настоящего запуска.</p><ol><li><a href="/ru/learn/frames-and-the-fcs/">22 Кадры и контрольная сумма</a></li><li><a href="/ru/learn/rgmii-timing/">23 Тайминг RGMII</a></li><li><a href="/ru/learn/bring-up-on-the-bench/">24 Запуск на стенде</a></li></ol></li><li><h3>Модуль 9 · Стенд</h3><p>Дисциплина, оберегающая настоящее железо: кто держит IO, как взять и вернуть захват, и что запускается дальше.</p><ol><li><a href="/ru/learn/who-holds-the-io/">25 Кто держит IO</a></li><li><a href="/ru/learn/claim-and-release/">26 Захват и возврат</a></li><li><a href="/ru/learn/capstone-what-runs-next/">27 Капстоун: что запускается дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/five-rules-for-weights/index.html b/apps/website/public/ru/learn/five-rules-for-weights/index.html
index 48841cef7e..63a6da4bc3 100644
--- a/apps/website/public/ru/learn/five-rules-for-weights/index.html
+++ b/apps/website/public/ru/learn/five-rules-for-weights/index.html
@@ -117,5 +117,6 @@ <h1>Пять правил для алфавита весов</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Лаборатория: наши исследования</h3><p>Свой формат чисел, честная таблица результатов и таблицы модели, перемноженные на плате.</p><ol><li><a href="/ru/learn/t27-number-format/">01 T27: наш собственный формат чисел</a></li><li><a href="/ru/learn/an-honest-scoreboard/">02 Честная таблица результатов</a></li><li><a href="/ru/learn/a-model-on-the-board/">03 Таблицы модели на плате</a></li></ol></li><li><h3>Модуль 2 · ИИ-числа: блок MX</h3><p>Как ИИ-чипы хранят веса в нескольких битах: один общий масштаб на блок, сам байт масштаба и что один выброс делает с соседями.</p><ol><li><a href="/ru/learn/one-scale-per-block/">04 Один масштаб на блок весов</a></li><li><a href="/ru/learn/scale-byte-native/">05 Байт масштаба на настоящей машине</a></li><li><a href="/ru/learn/one-outlier/">06 Один выброс, много нулей</a></li></ol></li><li><h3>Модуль 3 · Тернарные веса</h3><p>Веса, которые бывают только минус масштаб, ноль или плюс масштаб, пять правил, которые должен пройти тернарный алфавит, и прогон тестов, который ничего не проверил.</p><ol><li><a href="/ru/learn/phi-scaled-ternary/">07 Веса минус фи, ноль и плюс фи</a></li><li><a href="/ru/learn/a-pass-that-checked-nothing/">08 Прогон, который ничего не проверил</a></li><li class="on" aria-current="page"><a href="/ru/learn/five-rules-for-weights/">09 Пять правил для алфавита весов</a></li></ol></li><li><h3>Модуль 4 · Тернарный формат Ternary Network Float</h3><p>Правило, которое компилятор проверяет до запуска любого теста, и 17-битное число с плавающей точкой, порядок которого — четыре сбалансированных трита.</p><ol><li><a href="/ru/learn/blocked-is-not-failed/">10 Заблокирован — не значит провален</a></li><li><a href="/ru/learn/four-trit-exponent/">11 Число, у которого порядок — четыре трита</a></li><li><a href="/ru/learn/negate-twice/">12 Дважды сменить знак — получить число обратно</a></li></ol></li><li><h3>Модуль 5 · Арифметика чисел со знаком</h3><p>Умножить два числа со знаком, сложить их, когда знаки разные, и сделать то и другое сразу в умножении с накоплением.</p><ol><li><a href="/ru/learn/a-sign-is-one-bit/">13 Знак — это один бит</a></li><li><a href="/ru/learn/opposite-signs-subtract/">14 Разные знаки — вычитание</a></li><li><a href="/ru/learn/multiply-then-add/">15 Умножить, потом сложить</a></li></ol></li><li><h3>Модуль 6 · Части нейрона</h3><p>ReLU с изломом в нуле, степень двойки для softmax и argmax, который называет ответ.</p><ol><li><a href="/ru/learn/a-bend-at-zero/">16 Излом в нуле</a></li><li><a href="/ru/learn/two-to-the-x/">17 Два в степени x</a></li><li><a href="/ru/learn/pick-the-largest/">18 Выбрать наибольшее</a></li></ol></li><li><h3>Модуль 7 · Обучение на ошибке</h3><p>Потеря, которая оценивает неверную догадку в битах, один шаг, который сдвигает вес против градиента, и скрытый слой, который нужен XOR.</p><ol><li><a href="/ru/learn/a-loss-in-bits/">19 Потеря в битах</a></li><li><a href="/ru/learn/one-step-downhill/">20 Один шаг вниз по склону</a></li><li><a href="/ru/learn/xor-needs-a-bend/">21 XOR нужен излом</a></li></ol></li><li><h3>Модуль 8 · BitNet: тернарные сети</h3><p>Порог, который сжимает сумму обратно до трёх значений, один нейрон, который становится другой функцией при смене весов, и нейрон, который читает входы по 27 тритов за раз.</p><ol><li><a href="/ru/learn/back-to-three-values/">22 Обратно в три значения</a></li><li><a href="/ru/learn/same-neuron-new-weights/">23 Тот же нейрон, другие веса</a></li><li><a href="/ru/learn/a-neuron-in-chunks/">24 Нейрон по кускам</a></li></ol></li><li><h3>Модуль 9 · Тернарный MAC как чип</h3><p>Скалярное произведение 27 тритов как провода без регистра, та же сумма, которую регистр копит на каждом такте, и небольшая целая сеть в конце курса.</p><ol><li><a href="/ru/learn/a-dot-product-in-wires/">25 Скалярное произведение из проводов</a></li><li><a href="/ru/learn/add-it-up-every-clock/">26 Складывать на каждом такте</a></li><li><a href="/ru/learn/a-whole-network/">27 Целая сеть</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/five-trits-a-byte/index.html b/apps/website/public/ru/learn/five-trits-a-byte/index.html
index a02e660cf8..d19b64d7ea 100644
--- a/apps/website/public/ru/learn/five-trits-a-byte/index.html
+++ b/apps/website/public/ru/learn/five-trits-a-byte/index.html
@@ -769,5 +769,6 @@ <h1>Пять тритов в байте</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Три значения</h3><p>Почему три, как сбалансированная троичная система пишет любое число без знака и что дают переворот и отсечение тритов.</p><ol><li><a href="/ru/learn/why-three/">01 Почему три</a></li><li><a href="/ru/learn/balanced-ternary/">02 Сбалансированная троичная</a></li><li><a href="/ru/learn/flip-and-cut/">03 Переворот вместо минуса, отсечение вместо округления</a></li></ol></li><li><h3>Модуль 2 · Логика с «не знаю»</h3><p>Трёхзначные вентили Клини, два вентиля без двоичного двойника и сложение как пара таблиц.</p><ol><li><a href="/ru/learn/kleene-logic/">04 Трёхзначная логика</a></li><li><a href="/ru/learn/consensus-and-any/">05 Консенсус и «любой»</a></li><li><a href="/ru/learn/sum-and-carry/">06 Сумма и перенос</a></li></ol></li><li><h3>Модуль 3 · Складываем триты</h3><p>Полусумматор, полный сумматор и перенос, бегущий влево разряд за разрядом.</p><ol><li><a href="/ru/learn/half-adder/">07 Полусумматор</a></li><li><a href="/ru/learn/full-adder/">08 Полный сумматор</a></li><li><a href="/ru/learn/ripple-carry/">09 Сквозной перенос</a></li></ol></li><li><h3>Модуль 4 · Умножение без умножителя</h3><p>Копия, ноль или переворот: произведение на один трит, умножение столбиком и MAC, который только складывает.</p><ol><li><a href="/ru/learn/times-a-trit/">10 Умножить на один трит</a></li><li><a href="/ru/learn/long-multiplication/">11 Умножение столбиком</a></li><li><a href="/ru/learn/ternary-mac/">12 Троичный MAC</a></li></ol></li><li><h3>Модуль 5 · Триты в двоичной памяти</h3><p>Два бита на трит, пять тритов на байт и сколько бит нужно слову из 27 тритов.</p><ol><li><a href="/ru/learn/two-bits-per-trit/">13 Два бита на трит</a></li><li class="on" aria-current="page"><a href="/ru/learn/five-trits-a-byte/">14 Пять тритов в байте</a></li><li><a href="/ru/learn/a-word-of-27/">15 Слово из 27 тритов</a></li></ol></li><li><h3>Модуль 6 · Командное слово TRI-27</h3><p>32-битное слово, которое декодирует эмулятор Trinity: его поля, 47 опкодов и 15-битное непосредственное значение.</p><ol><li><a href="/ru/learn/the-instruction-word/">16 Командное слово</a></li><li><a href="/ru/learn/the-opcode-map/">17 Карта опкодов</a></li><li><a href="/ru/learn/the-immediate/">18 Непосредственное значение</a></li></ol></li><li><h3>Модуль 7 · Троичная машина</h3><p>АЛУ из сумматоров этого курса, переход на три стороны и программа, которую можно пройти по шагам.</p><ol><li><a href="/ru/learn/a-ternary-alu/">19 Троичное АЛУ</a></li><li><a href="/ru/learn/three-way-jump/">20 Переход на три стороны</a></li><li><a href="/ru/learn/run-a-program/">21 Запустить программу</a></li></ol></li><li><h3>Модуль 8 · Три ответа</h3><p>Сравнение по старшему различному триту, поиск числа по третям и сортировка трёхзначными сравнениями.</p><ol><li><a href="/ru/learn/compare-at-the-top/">22 Сравнение сверху</a></li><li><a href="/ru/learn/questions-in-thirds/">23 Вопросы по третям</a></li><li><a href="/ru/learn/sort-by-three/">24 Сортировка трёхзначными сравнениями</a></li></ol></li><li><h3>Модуль 9 · Троичные нейроны</h3><p>Веса -1, 0, +1: один нейрон, детектор и маленький слой, и нигде нет умножителя.</p><ol><li><a href="/ru/learn/one-ternary-neuron/">25 Один троичный нейрон</a></li><li><a href="/ru/learn/a-bar-detector/">26 Детектор полосы</a></li><li><a href="/ru/learn/a-tiny-layer/">27 Маленький слой</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/flip-and-cut/index.html b/apps/website/public/ru/learn/flip-and-cut/index.html
index bb34ebf674..6cb2532041 100644
--- a/apps/website/public/ru/learn/flip-and-cut/index.html
+++ b/apps/website/public/ru/learn/flip-and-cut/index.html
@@ -769,5 +769,6 @@ <h1>Переворот вместо минуса, отсечение вмест
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Три значения</h3><p>Почему три, как сбалансированная троичная система пишет любое число без знака и что дают переворот и отсечение тритов.</p><ol><li><a href="/ru/learn/why-three/">01 Почему три</a></li><li><a href="/ru/learn/balanced-ternary/">02 Сбалансированная троичная</a></li><li class="on" aria-current="page"><a href="/ru/learn/flip-and-cut/">03 Переворот вместо минуса, отсечение вместо округления</a></li></ol></li><li><h3>Модуль 2 · Логика с «не знаю»</h3><p>Трёхзначные вентили Клини, два вентиля без двоичного двойника и сложение как пара таблиц.</p><ol><li><a href="/ru/learn/kleene-logic/">04 Трёхзначная логика</a></li><li><a href="/ru/learn/consensus-and-any/">05 Консенсус и «любой»</a></li><li><a href="/ru/learn/sum-and-carry/">06 Сумма и перенос</a></li></ol></li><li><h3>Модуль 3 · Складываем триты</h3><p>Полусумматор, полный сумматор и перенос, бегущий влево разряд за разрядом.</p><ol><li><a href="/ru/learn/half-adder/">07 Полусумматор</a></li><li><a href="/ru/learn/full-adder/">08 Полный сумматор</a></li><li><a href="/ru/learn/ripple-carry/">09 Сквозной перенос</a></li></ol></li><li><h3>Модуль 4 · Умножение без умножителя</h3><p>Копия, ноль или переворот: произведение на один трит, умножение столбиком и MAC, который только складывает.</p><ol><li><a href="/ru/learn/times-a-trit/">10 Умножить на один трит</a></li><li><a href="/ru/learn/long-multiplication/">11 Умножение столбиком</a></li><li><a href="/ru/learn/ternary-mac/">12 Троичный MAC</a></li></ol></li><li><h3>Модуль 5 · Триты в двоичной памяти</h3><p>Два бита на трит, пять тритов на байт и сколько бит нужно слову из 27 тритов.</p><ol><li><a href="/ru/learn/two-bits-per-trit/">13 Два бита на трит</a></li><li><a href="/ru/learn/five-trits-a-byte/">14 Пять тритов в байте</a></li><li><a href="/ru/learn/a-word-of-27/">15 Слово из 27 тритов</a></li></ol></li><li><h3>Модуль 6 · Командное слово TRI-27</h3><p>32-битное слово, которое декодирует эмулятор Trinity: его поля, 47 опкодов и 15-битное непосредственное значение.</p><ol><li><a href="/ru/learn/the-instruction-word/">16 Командное слово</a></li><li><a href="/ru/learn/the-opcode-map/">17 Карта опкодов</a></li><li><a href="/ru/learn/the-immediate/">18 Непосредственное значение</a></li></ol></li><li><h3>Модуль 7 · Троичная машина</h3><p>АЛУ из сумматоров этого курса, переход на три стороны и программа, которую можно пройти по шагам.</p><ol><li><a href="/ru/learn/a-ternary-alu/">19 Троичное АЛУ</a></li><li><a href="/ru/learn/three-way-jump/">20 Переход на три стороны</a></li><li><a href="/ru/learn/run-a-program/">21 Запустить программу</a></li></ol></li><li><h3>Модуль 8 · Три ответа</h3><p>Сравнение по старшему различному триту, поиск числа по третям и сортировка трёхзначными сравнениями.</p><ol><li><a href="/ru/learn/compare-at-the-top/">22 Сравнение сверху</a></li><li><a href="/ru/learn/questions-in-thirds/">23 Вопросы по третям</a></li><li><a href="/ru/learn/sort-by-three/">24 Сортировка трёхзначными сравнениями</a></li></ol></li><li><h3>Модуль 9 · Троичные нейроны</h3><p>Веса -1, 0, +1: один нейрон, детектор и маленький слой, и нигде нет умножителя.</p><ol><li><a href="/ru/learn/one-ternary-neuron/">25 Один троичный нейрон</a></li><li><a href="/ru/learn/a-bar-detector/">26 Детектор полосы</a></li><li><a href="/ru/learn/a-tiny-layer/">27 Маленький слой</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/folding-taps-onto-slices/index.html b/apps/website/public/ru/learn/folding-taps-onto-slices/index.html
index 77b6f22178..dade06a53e 100644
--- a/apps/website/public/ru/learn/folding-taps-onto-slices/index.html
+++ b/apps/website/public/ru/learn/folding-taps-onto-slices/index.html
@@ -1122,5 +1122,6 @@ <h1>Раскладка отводов по слайсам</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Сложение</h3><p>Почему сложение медленное: перенос, который идёт вверх по слову, выделенная цепь в каждом слайсе, префиксные сети и деревья сохранения переноса.</p><ol><li><a href="/ru/learn/the-carry-chain/">01 Цепь переноса</a></li><li><a href="/ru/learn/prefix-adders/">02 Взгляд вперёд: префиксные сумматоры</a></li><li><a href="/ru/learn/carry-save/">03 Три числа сразу</a></li></ol></li><li><h3>Модуль 2 · Умножение</h3><p>Произведение — сумма сдвинутых копий; перекодирование Бута вдвое сокращает их, а слайс DSP48E1 делает остальное одним блоком.</p><ol><li><a href="/ru/learn/shift-and-add/">04 Сдвиг и сложение</a></li><li><a href="/ru/learn/booths-trick/">05 Приём Бута</a></li><li><a href="/ru/learn/the-dsp48e1-slice/">06 Слайс DSP48E1</a></li></ol></li><li><h3>Модуль 3 · Фиксированная точка</h3><p>Где стоит двоичная точка, что округление делает со значением и с его средним и что даёт каждый бит квантователя.</p><ol><li><a href="/ru/learn/q-formats/">07 Форматы Q</a></li><li><a href="/ru/learn/rounding-and-saturation/">08 Округление и насыщение</a></li><li><a href="/ru/learn/quantization-noise/">09 Шум квантования</a></li></ol></li><li><h3>Модуль 4 · Функции в железе</h3><p>Синус, косинус, угол и длина из сдвигов и сложений — и когда таблица оказывается выгоднее.</p><ol><li><a href="/ru/learn/rotating-by-shifts/">10 Поворот сдвигами</a></li><li><a href="/ru/learn/angle-and-magnitude/">11 Угол и модуль</a></li><li><a href="/ru/learn/tables-and-interpolation/">12 Таблицы и интерполяция</a></li></ol></li><li><h3>Модуль 5 · Сигналы и дискретизация</h3><p>Что дискретизация делает с частотой, как генератор строится из сумматора и что измеряет бин ДПФ.</p><ol><li><a href="/ru/learn/sampling-and-aliasing/">13 Дискретизация и наложение спектров</a></li><li><a href="/ru/learn/the-nco/">14 Генератор с числовым управлением</a></li><li><a href="/ru/learn/bins-of-a-spectrum/">15 Бины спектра</a></li></ol></li><li><h3>Модуль 6 · Фильтры FIR</h3><p>Скользящее среднее, расчёт окном sinc с целыми коэффициентами и раскладка отводов по слайсам DSP.</p><ol><li><a href="/ru/learn/the-moving-average/">16 Скользящее среднее</a></li><li><a href="/ru/learn/designing-taps/">17 Расчёт коэффициентов</a></li><li class="on" aria-current="page"><a href="/ru/learn/folding-taps-onto-slices/">18 Раскладка отводов по слайсам</a></li></ol></li><li><h3>Модуль 7 · Многоскоростная обработка</h3><p>Понижение частоты дискретизации без заворота шума: децимация, фильтр CIC и полифазная форма.</p><ol><li><a href="/ru/learn/lowering-the-rate/">19 Понижение частоты</a></li><li><a href="/ru/learn/the-cic-filter/">20 Фильтр CIC</a></li><li><a href="/ru/learn/polyphase/">21 Полифазная форма</a></li></ol></li><li><h3>Модуль 8 · БПФ</h3><p>N log N вместо N^2: бабочки, бит-реверсный порядок входа и биты, которые добавляет каждый этап.</p><ol><li><a href="/ru/learn/the-butterfly/">22 Бабочка</a></li><li><a href="/ru/learn/twiddles-and-bit-reversal/">23 Поворачивающие множители и бит-реверс</a></li><li><a href="/ru/learn/growth-through-the-stages/">24 Рост через этапы</a></li></ol></li><li><h3>Модуль 9 · На стенде</h3><p>От арифметики к плате: коррелятор из спеки модема, бюджет слайсов и тайминга и итоговый фильтр.</p><ol><li><a href="/ru/learn/a-correlator-finds-the-frame/">25 Коррелятор находит кадр</a></li><li><a href="/ru/learn/does-it-fit-and-close-timing/">26 Помещается ли и сходится ли тайминг?</a></li><li><a href="/ru/learn/capstone-a-filter-on-the-board/">27 Итог: фильтр на плате</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/four-trit-exponent/index.html b/apps/website/public/ru/learn/four-trit-exponent/index.html
index b408bf578b..99b1c02cae 100644
--- a/apps/website/public/ru/learn/four-trit-exponent/index.html
+++ b/apps/website/public/ru/learn/four-trit-exponent/index.html
@@ -142,5 +142,6 @@ <h1>Число, у которого порядок — четыре трита</
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Лаборатория: наши исследования</h3><p>Свой формат чисел, честная таблица результатов и таблицы модели, перемноженные на плате.</p><ol><li><a href="/ru/learn/t27-number-format/">01 T27: наш собственный формат чисел</a></li><li><a href="/ru/learn/an-honest-scoreboard/">02 Честная таблица результатов</a></li><li><a href="/ru/learn/a-model-on-the-board/">03 Таблицы модели на плате</a></li></ol></li><li><h3>Модуль 2 · ИИ-числа: блок MX</h3><p>Как ИИ-чипы хранят веса в нескольких битах: один общий масштаб на блок, сам байт масштаба и что один выброс делает с соседями.</p><ol><li><a href="/ru/learn/one-scale-per-block/">04 Один масштаб на блок весов</a></li><li><a href="/ru/learn/scale-byte-native/">05 Байт масштаба на настоящей машине</a></li><li><a href="/ru/learn/one-outlier/">06 Один выброс, много нулей</a></li></ol></li><li><h3>Модуль 3 · Тернарные веса</h3><p>Веса, которые бывают только минус масштаб, ноль или плюс масштаб, пять правил, которые должен пройти тернарный алфавит, и прогон тестов, который ничего не проверил.</p><ol><li><a href="/ru/learn/phi-scaled-ternary/">07 Веса минус фи, ноль и плюс фи</a></li><li><a href="/ru/learn/a-pass-that-checked-nothing/">08 Прогон, который ничего не проверил</a></li><li><a href="/ru/learn/five-rules-for-weights/">09 Пять правил для алфавита весов</a></li></ol></li><li><h3>Модуль 4 · Тернарный формат Ternary Network Float</h3><p>Правило, которое компилятор проверяет до запуска любого теста, и 17-битное число с плавающей точкой, порядок которого — четыре сбалансированных трита.</p><ol><li><a href="/ru/learn/blocked-is-not-failed/">10 Заблокирован — не значит провален</a></li><li class="on" aria-current="page"><a href="/ru/learn/four-trit-exponent/">11 Число, у которого порядок — четыре трита</a></li><li><a href="/ru/learn/negate-twice/">12 Дважды сменить знак — получить число обратно</a></li></ol></li><li><h3>Модуль 5 · Арифметика чисел со знаком</h3><p>Умножить два числа со знаком, сложить их, когда знаки разные, и сделать то и другое сразу в умножении с накоплением.</p><ol><li><a href="/ru/learn/a-sign-is-one-bit/">13 Знак — это один бит</a></li><li><a href="/ru/learn/opposite-signs-subtract/">14 Разные знаки — вычитание</a></li><li><a href="/ru/learn/multiply-then-add/">15 Умножить, потом сложить</a></li></ol></li><li><h3>Модуль 6 · Части нейрона</h3><p>ReLU с изломом в нуле, степень двойки для softmax и argmax, который называет ответ.</p><ol><li><a href="/ru/learn/a-bend-at-zero/">16 Излом в нуле</a></li><li><a href="/ru/learn/two-to-the-x/">17 Два в степени x</a></li><li><a href="/ru/learn/pick-the-largest/">18 Выбрать наибольшее</a></li></ol></li><li><h3>Модуль 7 · Обучение на ошибке</h3><p>Потеря, которая оценивает неверную догадку в битах, один шаг, который сдвигает вес против градиента, и скрытый слой, который нужен XOR.</p><ol><li><a href="/ru/learn/a-loss-in-bits/">19 Потеря в битах</a></li><li><a href="/ru/learn/one-step-downhill/">20 Один шаг вниз по склону</a></li><li><a href="/ru/learn/xor-needs-a-bend/">21 XOR нужен излом</a></li></ol></li><li><h3>Модуль 8 · BitNet: тернарные сети</h3><p>Порог, который сжимает сумму обратно до трёх значений, один нейрон, который становится другой функцией при смене весов, и нейрон, который читает входы по 27 тритов за раз.</p><ol><li><a href="/ru/learn/back-to-three-values/">22 Обратно в три значения</a></li><li><a href="/ru/learn/same-neuron-new-weights/">23 Тот же нейрон, другие веса</a></li><li><a href="/ru/learn/a-neuron-in-chunks/">24 Нейрон по кускам</a></li></ol></li><li><h3>Модуль 9 · Тернарный MAC как чип</h3><p>Скалярное произведение 27 тритов как провода без регистра, та же сумма, которую регистр копит на каждом такте, и небольшая целая сеть в конце курса.</p><ol><li><a href="/ru/learn/a-dot-product-in-wires/">25 Скалярное произведение из проводов</a></li><li><a href="/ru/learn/add-it-up-every-clock/">26 Складывать на каждом такте</a></li><li><a href="/ru/learn/a-whole-network/">27 Целая сеть</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/fpga-is-a-grid/index.html b/apps/website/public/ru/learn/fpga-is-a-grid/index.html
index 789b6341d2..8d4a938e6e 100644
--- a/apps/website/public/ru/learn/fpga-is-a-grid/index.html
+++ b/apps/website/public/ru/learn/fpga-is-a-grid/index.html
@@ -367,5 +367,6 @@ <h1>FPGA — это сетка, которую вы настраиваете</h1
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Чип</h3><p>Что такое FPGA, какой чип у нас и как его выводы встречаются с платой.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/fpga-is-a-grid/">01 FPGA — это сетка, которую вы настраиваете</a></li><li><a href="/ru/learn/which-chip/">02 Какой это чип?</a></li><li><a href="/ru/learn/pins-and-package/">03 Выводы, шарики и корпус</a></li></ol></li><li><h3>Модуль 2 · Числа в железе</h3><p>Биты, триты и форматы чисел, и сколько стоит арифметика в логике.</p><ol><li><a href="/ru/learn/bits-and-trits/">04 Биты, триты и GF16</a></li><li><a href="/ru/learn/what-arithmetic-costs/">05 Сколько стоит арифметика</a></li><li><a href="/ru/learn/how-much-fits/">06 Сколько помещается на чип</a></li></ol></li><li><h3>Модуль 3 · Ваша программа на t27</h3><p>Напишите спеку, проверьте её и увидьте, почему «компилируется» не значит «верно».</p><ol><li><a href="/ru/learn/hello-t27/">07 Привет, t27</a></li><li><a href="/ru/learn/tests-are-the-spec/">08 Тесты — часть спеки</a></li><li><a href="/ru/learn/compiles-is-not-correct/">09 Компилируется — не значит верно</a></li></ol></li><li><h3>Модуль 4 · Внутри t27c</h3><p>Как компилятор читает спеку и что он пишет, включая нативный t27b.</p><ol><li><a href="/ru/learn/compiler-stages/">10 Стадии t27c</a></li><li><a href="/ru/learn/t27c-on-a-real-machine/">11 t27c на настоящей машине</a></li><li><a href="/ru/learn/seven-backends-and-t27b/">12 Семь бэкендов и t27b</a></li></ol></li><li><h3>Модуль 5 · От спеки к железу</h3><p>Verilog, который пишет t27c, ячейки, в которые он превращается, и что делает одна LUT.</p><ol><li><a href="/ru/learn/spec-to-verilog/">13 От спеки к Verilog</a></li><li><a href="/ru/learn/count-the-cells/">14 Посчитайте ячейки</a></li><li><a href="/ru/learn/one-lut-up-close/">15 Одна LUT вблизи</a></li></ol></li><li><h3>Модуль 6 · Читаем синтез</h3><p>Что yosys сообщает о вашем дизайне и какие предупреждения важны.</p><ol><li><a href="/ru/learn/synthesis-warnings/">16 Важные предупреждения синтеза</a></li><li><a href="/ru/learn/state-machines/">17 Конечные автоматы</a></li><li><a href="/ru/learn/where-the-luts-go/">18 Куда уходят LUT</a></li></ol></li><li><h3>Модуль 7 · Размещение, трассировка, тайминг</h3><p>Где ячейки оказываются на кристалле и успевает ли тактовая частота.</p><ol><li><a href="/ru/learn/placement/">19 Размещение: где оказываются ячейки</a></li><li><a href="/ru/learn/logic-depth/">20 Глубина логики</a></li><li><a href="/ru/learn/timing-slack/">21 Тайминг и запас</a></li></ol></li><li><h3>Модуль 8 · Битстрим</h3><p>Как оттрассированный дизайн становится битами, которые загружает чип, без Vivado.</p><ol><li><a href="/ru/learn/fasm-to-frames/">22 От FASM к фреймам</a></li><li><a href="/ru/learn/a-build-receipt/">23 Сборка по пунктам</a></li><li><a href="/ru/learn/read-a-bitstream/">24 Читаем битстрим</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Спросите чип, кто он, загрузите биты и проверьте их.</p><ol><li><a href="/ru/learn/jtag-asks-the-chip/">25 JTAG спрашивает чип</a></li><li><a href="/ru/learn/same-bits-on-the-board/">26 Те же биты — на плате</a></li><li><a href="/ru/learn/capstone/">27 Итог: переугадайте yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/frames-and-the-fcs/index.html b/apps/website/public/ru/learn/frames-and-the-fcs/index.html
index 6ec07c1e55..e1734a96f8 100644
--- a/apps/website/public/ru/learn/frames-and-the-fcs/index.html
+++ b/apps/website/public/ru/learn/frames-and-the-fcs/index.html
@@ -299,5 +299,6 @@ <h1>Кадры и контрольная сумма</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое шина</h3><p>Зачем вообще существует шина: разговор по проводам, с кадрами и адресами, и кому позволено говорить.</p><ol><li><a href="/ru/learn/wires-that-talk/">01 Провода, которые говорят</a></li><li><a href="/ru/learn/framing-a-conversation/">02 Кадрирование разговора</a></li><li><a href="/ru/learn/master-and-servant/">03 Хозяин и слуга</a></li></ol></li><li><h3>Модуль 2 · Разговор по UART</h3><p>Двухпроводная шина без такта: кадр, делитель, задающий скорость, и статус, который опрашивает драйвер.</p><ol><li><a href="/ru/learn/the-frame/">04 Кадр: старт, данные, стоп</a></li><li><a href="/ru/learn/baud-and-the-divisor/">05 Бод и делитель</a></li><li><a href="/ru/learn/status-and-fifos/">06 Статус и FIFO</a></li></ol></li><li><h3>Модуль 3 · SPI под тактом</h3><p>Разговор под тактом: четыре режима, лестница предделителей и выбор кристалла на каждого подчинённого.</p><ol><li><a href="/ru/learn/modes-cpol-cpha/">07 Режимы: CPOL и CPHA</a></li><li><a href="/ru/learn/the-prescaler-ladder/">08 Лестница предделителей</a></li><li><a href="/ru/learn/chip-select-and-widths/">09 Выбор кристалла и ширины</a></li></ol></li><li><h3>Модуль 4 · Регистровая шина APB</h3><p>Шина регистров: PSEL и PENABLE, строобы и ожидания, и сколько адресных бит стоит количество периферии.</p><ol><li><a href="/ru/learn/psel-and-penable/">10 PSEL и PENABLE</a></li><li><a href="/ru/learn/strobes-and-waits/">11 Строобы и ожидания</a></li><li><a href="/ru/learn/address-decode/">12 Дешифрация адреса</a></li></ol></li><li><h3>Модуль 5 · Пять каналов AXI4</h3><p>Пять каналов: адрес, данные и отклик в обе стороны, lite или full, пакеты и идентификаторы.</p><ol><li><a href="/ru/learn/five-channels/">13 Пять каналов</a></li><li><a href="/ru/learn/lite-or-full/">14 Lite или full</a></li><li><a href="/ru/learn/bursts-and-ids/">15 Пакеты и идентификаторы</a></li></ol></li><li><h3>Модуль 6 · Память</h3><p>Что находится на дальнем конце каждой шины: карты памяти, типы портов и задержка, которую должно покрывать ожидание.</p><ol><li><a href="/ru/learn/memory-maps/">16 Карты памяти</a></li><li><a href="/ru/learn/ports-and-kinds/">17 Порты и типы</a></li><li><a href="/ru/learn/latency/">18 Задержка</a></li></ol></li><li><h3>Модуль 7 · Мосты</h3><p>Почему в дизайне появляется больше одной шины и пакетный мост, переносящий работу между ними.</p><ol><li><a href="/ru/learn/why-bridges/">19 Зачем мосты</a></li><li><a href="/ru/learn/the-packet-bridge/">20 Пакетный мост</a></li><li><a href="/ru/learn/everything-in-one-bench/">21 Всё в одном стенде</a></li></ol></li><li><h3>Модуль 8 · Ethernet: кадры и тайминг</h3><p>Кадры, контрольная сумма кадра, тайминг RGMII и заранее зарегистрированные шаги настоящего запуска.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/frames-and-the-fcs/">22 Кадры и контрольная сумма</a></li><li><a href="/ru/learn/rgmii-timing/">23 Тайминг RGMII</a></li><li><a href="/ru/learn/bring-up-on-the-bench/">24 Запуск на стенде</a></li></ol></li><li><h3>Модуль 9 · Стенд</h3><p>Дисциплина, оберегающая настоящее железо: кто держит IO, как взять и вернуть захват, и что запускается дальше.</p><ol><li><a href="/ru/learn/who-holds-the-io/">25 Кто держит IO</a></li><li><a href="/ru/learn/claim-and-release/">26 Захват и возврат</a></li><li><a href="/ru/learn/capstone-what-runs-next/">27 Капстоун: что запускается дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/framing-a-conversation/index.html b/apps/website/public/ru/learn/framing-a-conversation/index.html
index 4fb136da1f..132cb66685 100644
--- a/apps/website/public/ru/learn/framing-a-conversation/index.html
+++ b/apps/website/public/ru/learn/framing-a-conversation/index.html
@@ -476,5 +476,6 @@ <h1>Кадрирование разговора</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое шина</h3><p>Зачем вообще существует шина: разговор по проводам, с кадрами и адресами, и кому позволено говорить.</p><ol><li><a href="/ru/learn/wires-that-talk/">01 Провода, которые говорят</a></li><li class="on" aria-current="page"><a href="/ru/learn/framing-a-conversation/">02 Кадрирование разговора</a></li><li><a href="/ru/learn/master-and-servant/">03 Хозяин и слуга</a></li></ol></li><li><h3>Модуль 2 · Разговор по UART</h3><p>Двухпроводная шина без такта: кадр, делитель, задающий скорость, и статус, который опрашивает драйвер.</p><ol><li><a href="/ru/learn/the-frame/">04 Кадр: старт, данные, стоп</a></li><li><a href="/ru/learn/baud-and-the-divisor/">05 Бод и делитель</a></li><li><a href="/ru/learn/status-and-fifos/">06 Статус и FIFO</a></li></ol></li><li><h3>Модуль 3 · SPI под тактом</h3><p>Разговор под тактом: четыре режима, лестница предделителей и выбор кристалла на каждого подчинённого.</p><ol><li><a href="/ru/learn/modes-cpol-cpha/">07 Режимы: CPOL и CPHA</a></li><li><a href="/ru/learn/the-prescaler-ladder/">08 Лестница предделителей</a></li><li><a href="/ru/learn/chip-select-and-widths/">09 Выбор кристалла и ширины</a></li></ol></li><li><h3>Модуль 4 · Регистровая шина APB</h3><p>Шина регистров: PSEL и PENABLE, строобы и ожидания, и сколько адресных бит стоит количество периферии.</p><ol><li><a href="/ru/learn/psel-and-penable/">10 PSEL и PENABLE</a></li><li><a href="/ru/learn/strobes-and-waits/">11 Строобы и ожидания</a></li><li><a href="/ru/learn/address-decode/">12 Дешифрация адреса</a></li></ol></li><li><h3>Модуль 5 · Пять каналов AXI4</h3><p>Пять каналов: адрес, данные и отклик в обе стороны, lite или full, пакеты и идентификаторы.</p><ol><li><a href="/ru/learn/five-channels/">13 Пять каналов</a></li><li><a href="/ru/learn/lite-or-full/">14 Lite или full</a></li><li><a href="/ru/learn/bursts-and-ids/">15 Пакеты и идентификаторы</a></li></ol></li><li><h3>Модуль 6 · Память</h3><p>Что находится на дальнем конце каждой шины: карты памяти, типы портов и задержка, которую должно покрывать ожидание.</p><ol><li><a href="/ru/learn/memory-maps/">16 Карты памяти</a></li><li><a href="/ru/learn/ports-and-kinds/">17 Порты и типы</a></li><li><a href="/ru/learn/latency/">18 Задержка</a></li></ol></li><li><h3>Модуль 7 · Мосты</h3><p>Почему в дизайне появляется больше одной шины и пакетный мост, переносящий работу между ними.</p><ol><li><a href="/ru/learn/why-bridges/">19 Зачем мосты</a></li><li><a href="/ru/learn/the-packet-bridge/">20 Пакетный мост</a></li><li><a href="/ru/learn/everything-in-one-bench/">21 Всё в одном стенде</a></li></ol></li><li><h3>Модуль 8 · Ethernet: кадры и тайминг</h3><p>Кадры, контрольная сумма кадра, тайминг RGMII и заранее зарегистрированные шаги настоящего запуска.</p><ol><li><a href="/ru/learn/frames-and-the-fcs/">22 Кадры и контрольная сумма</a></li><li><a href="/ru/learn/rgmii-timing/">23 Тайминг RGMII</a></li><li><a href="/ru/learn/bring-up-on-the-bench/">24 Запуск на стенде</a></li></ol></li><li><h3>Модуль 9 · Стенд</h3><p>Дисциплина, оберегающая настоящее железо: кто держит IO, как взять и вернуть захват, и что запускается дальше.</p><ol><li><a href="/ru/learn/who-holds-the-io/">25 Кто держит IO</a></li><li><a href="/ru/learn/claim-and-release/">26 Захват и возврат</a></li><li><a href="/ru/learn/capstone-what-runs-next/">27 Капстоун: что запускается дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/fsm-coverage/index.html b/apps/website/public/ru/learn/fsm-coverage/index.html
index da7342d13c..941aad9564 100644
--- a/apps/website/public/ru/learn/fsm-coverage/index.html
+++ b/apps/website/public/ru/learn/fsm-coverage/index.html
@@ -576,5 +576,6 @@ <h1>Покрытие автоматов</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Зачем проверять</h3><p>Дизайны, которые компилируются и ошибаются; модель, которая выносит вердикт; план, записанный до кода.</p><ol><li><a href="/ru/learn/bugs-that-compile/">01 Баги, которые компилируются</a></li><li><a href="/ru/learn/golden-model/">02 Золотая модель</a></li><li><a href="/ru/learn/plan-before-code/">03 Сначала план</a></li></ol></li><li><h3>Модуль 2 · Тестбенчи</h3><p>Стимулы, проверки и вердикт, записанные как один spec рядом с дизайном, который они судят.</p><ol><li><a href="/ru/learn/anatomy-of-a-tb/">04 Анатомия тестбенча</a></li><li><a href="/ru/learn/self-checking/">05 Самопроверка вместо всматривания</a></li><li><a href="/ru/learn/directed-vs-random/">06 Направленные против случайных</a></li></ol></li><li><h3>Модуль 3 · Временные диаграммы</h3><p>Трасса каждого сигнала, прочитанная так, как её читает инженер по железу, и два прогона, сравнённые между собой.</p><ol><li><a href="/ru/learn/vcd-format/">07 Формат VCD</a></li><li><a href="/ru/learn/debug-with-waves/">08 Отладка по волнам</a></li><li><a href="/ru/learn/compare-traces/">09 Сравнение двух трасс</a></li></ol></li><li><h3>Модуль 4 · Векторы соответствия</h3><p>Случаи с ответом, записанным рядом, там, откуда их достанет компилятор.</p><ol><li><a href="/ru/learn/what-a-vector-is/">10 Что такое вектор</a></li><li><a href="/ru/learn/golden-ruler-vectors/">11 Векторы как эталонная линейка</a></li><li><a href="/ru/learn/end-to-end-scenarios/">12 Сквозные сценарии</a></li></ol></li><li><h3>Модуль 5 · Косимуляция</h3><p>Spec, симулятор и плата сходятся в одном ответе на стенде Artix-7 XC7A200T, и что делать, когда не сходятся.</p><ol><li><a href="/ru/learn/three-way-agreement/">13 Тройное согласие</a></li><li><a href="/ru/learn/simulator-internals/">14 Внутри симулятора</a></li><li><a href="/ru/learn/when-they-disagree/">15 Когда они не сходятся</a></li></ol></li><li><h3>Модуль 6 · Покрытие</h3><p>Чего коснулись тесты: строки, переключения, состояния — и что прячет это число.</p><ol><li><a href="/ru/learn/line-and-toggle/">16 Строки и переключения</a></li><li class="on" aria-current="page"><a href="/ru/learn/fsm-coverage/">17 Покрытие автоматов</a></li><li><a href="/ru/learn/coverage-lies/">18 Покрытие лжёт</a></li></ol></li><li><h3>Модуль 7 · Формальные методы</h3><p>Ассерты, верные каждый такт; ограниченный поиск контрпримера; и почему доказательству нужна индукция.</p><ol><li><a href="/ru/learn/assertions/">19 Ассерты</a></li><li><a href="/ru/learn/bounded-model-checking/">20 Ограниченная проверка моделей</a></li><li><a href="/ru/learn/induction/">21 Индукция</a></li></ol></li><li><h3>Модуль 8 · Мутации</h3><p>Ломайте дизайн нарочно и считайте, что заметили тесты.</p><ol><li><a href="/ru/learn/kill-the-mutant/">22 Убить мутанта</a></li><li><a href="/ru/learn/surviving-mutants/">23 Выжившие мутанты</a></li><li><a href="/ru/learn/test-strength/">24 Сила тестов</a></li></ol></li><li><h3>Модуль 9 · Приёмка</h3><p>Одна команда, все квитанции, чистый вердикт, который можно показать.</p><ol><li><a href="/ru/learn/build-verify/">25 Сборка и проверка</a></li><li><a href="/ru/learn/board-selftest/">26 Стенд проверяет себя</a></li><li><a href="/ru/learn/verify-capstone/">27 Капстоун: проверить модуль целиком</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">Этот курс сам является spec t27: прочитайте его</a></p><p>Виджеты сохраняют свои английские слова: каждый живёт в собственном spec t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/full-adder/index.html b/apps/website/public/ru/learn/full-adder/index.html
index 5d381c42ee..43ea5298dc 100644
--- a/apps/website/public/ru/learn/full-adder/index.html
+++ b/apps/website/public/ru/learn/full-adder/index.html
@@ -148,5 +148,6 @@ <h1>Полный сумматор</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Три значения</h3><p>Почему три, как сбалансированная троичная система пишет любое число без знака и что дают переворот и отсечение тритов.</p><ol><li><a href="/ru/learn/why-three/">01 Почему три</a></li><li><a href="/ru/learn/balanced-ternary/">02 Сбалансированная троичная</a></li><li><a href="/ru/learn/flip-and-cut/">03 Переворот вместо минуса, отсечение вместо округления</a></li></ol></li><li><h3>Модуль 2 · Логика с «не знаю»</h3><p>Трёхзначные вентили Клини, два вентиля без двоичного двойника и сложение как пара таблиц.</p><ol><li><a href="/ru/learn/kleene-logic/">04 Трёхзначная логика</a></li><li><a href="/ru/learn/consensus-and-any/">05 Консенсус и «любой»</a></li><li><a href="/ru/learn/sum-and-carry/">06 Сумма и перенос</a></li></ol></li><li><h3>Модуль 3 · Складываем триты</h3><p>Полусумматор, полный сумматор и перенос, бегущий влево разряд за разрядом.</p><ol><li><a href="/ru/learn/half-adder/">07 Полусумматор</a></li><li class="on" aria-current="page"><a href="/ru/learn/full-adder/">08 Полный сумматор</a></li><li><a href="/ru/learn/ripple-carry/">09 Сквозной перенос</a></li></ol></li><li><h3>Модуль 4 · Умножение без умножителя</h3><p>Копия, ноль или переворот: произведение на один трит, умножение столбиком и MAC, который только складывает.</p><ol><li><a href="/ru/learn/times-a-trit/">10 Умножить на один трит</a></li><li><a href="/ru/learn/long-multiplication/">11 Умножение столбиком</a></li><li><a href="/ru/learn/ternary-mac/">12 Троичный MAC</a></li></ol></li><li><h3>Модуль 5 · Триты в двоичной памяти</h3><p>Два бита на трит, пять тритов на байт и сколько бит нужно слову из 27 тритов.</p><ol><li><a href="/ru/learn/two-bits-per-trit/">13 Два бита на трит</a></li><li><a href="/ru/learn/five-trits-a-byte/">14 Пять тритов в байте</a></li><li><a href="/ru/learn/a-word-of-27/">15 Слово из 27 тритов</a></li></ol></li><li><h3>Модуль 6 · Командное слово TRI-27</h3><p>32-битное слово, которое декодирует эмулятор Trinity: его поля, 47 опкодов и 15-битное непосредственное значение.</p><ol><li><a href="/ru/learn/the-instruction-word/">16 Командное слово</a></li><li><a href="/ru/learn/the-opcode-map/">17 Карта опкодов</a></li><li><a href="/ru/learn/the-immediate/">18 Непосредственное значение</a></li></ol></li><li><h3>Модуль 7 · Троичная машина</h3><p>АЛУ из сумматоров этого курса, переход на три стороны и программа, которую можно пройти по шагам.</p><ol><li><a href="/ru/learn/a-ternary-alu/">19 Троичное АЛУ</a></li><li><a href="/ru/learn/three-way-jump/">20 Переход на три стороны</a></li><li><a href="/ru/learn/run-a-program/">21 Запустить программу</a></li></ol></li><li><h3>Модуль 8 · Три ответа</h3><p>Сравнение по старшему различному триту, поиск числа по третям и сортировка трёхзначными сравнениями.</p><ol><li><a href="/ru/learn/compare-at-the-top/">22 Сравнение сверху</a></li><li><a href="/ru/learn/questions-in-thirds/">23 Вопросы по третям</a></li><li><a href="/ru/learn/sort-by-three/">24 Сортировка трёхзначными сравнениями</a></li></ol></li><li><h3>Модуль 9 · Троичные нейроны</h3><p>Веса -1, 0, +1: один нейрон, детектор и маленький слой, и нигде нет умножителя.</p><ol><li><a href="/ru/learn/one-ternary-neuron/">25 Один троичный нейрон</a></li><li><a href="/ru/learn/a-bar-detector/">26 Детектор полосы</a></li><li><a href="/ru/learn/a-tiny-layer/">27 Маленький слой</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/functions/index.html b/apps/website/public/ru/learn/functions/index.html
index 759cc033e3..3469b4538b 100644
--- a/apps/website/public/ru/learn/functions/index.html
+++ b/apps/website/public/ru/learn/functions/index.html
@@ -89,5 +89,6 @@ <h1>Функции</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Файл спеки</h3><p>Что лежит в файле .t27: строка module, строки прозы и комментарии.</p><ol><li><a href="/ru/learn/what-a-spec-is/">01 Что такое спека</a></li><li><a href="/ru/learn/the-module-line/">02 Строка module</a></li><li><a href="/ru/learn/comments-and-prose/">03 Комментарии и строки прозы</a></li></ol></li><li><h3>Модуль 2 · Значения и типы</h3><p>Константы, ширина целого числа, а также true, false и текст.</p><ol><li><a href="/ru/learn/constants/">04 Константы</a></li><li><a href="/ru/learn/whole-numbers/">05 Целые числа и их ширина</a></li><li><a href="/ru/learn/true-false-and-text/">06 Истина, ложь и текст</a></li></ol></li><li><h3>Модуль 3 · Массивы, триты и выражения</h3><p>Списки значений, три значения трита и что вычисляют операторы.</p><ol><li><a href="/ru/learn/arrays/">07 Массивы</a></li><li><a href="/ru/learn/trits/">08 Триты: минус один, ноль, один</a></li><li><a href="/ru/learn/expressions/">09 Выражения</a></li></ol></li><li><h3>Модуль 4 · Тесты и функции</h3><p>Блок test, несколько маленьких тестов в одной спеке и функция с типизированными входами.</p><ol><li><a href="/ru/learn/a-test-block/">10 Блок test</a></li><li><a href="/ru/learn/many-tests/">11 Много тестов в одной спеке</a></li><li class="on" aria-current="page"><a href="/ru/learn/functions/">12 Функции</a></li></ol></li><li><h3>Модуль 5 · Внутри функции</h3><p>Имена через let и var, выбор через if и switch и циклы.</p><ol><li><a href="/ru/learn/local-names/">13 Локальные имена: let и var</a></li><li><a href="/ru/learn/if-and-else/">14 Выбор: if, else и switch</a></li><li><a href="/ru/learn/loops/">15 Циклы: while и for</a></li></ol></li><li><h3>Модуль 6 · Видимость и свои типы</h3><p>Что pub открывает другим модулям, и структуры и перечисления, которые вы объявляете.</p><ol><li><a href="/ru/learn/pub-or-private/">16 pub или закрытое имя</a></li><li><a href="/ru/learn/structs/">17 Структуры</a></li><li><a href="/ru/learn/enums/">18 Перечисления</a></li></ol></li><li><h3>Модуль 7 · Модули, правила и gen-ts</h3><p>Чтение другого модуля, правило, которое выполняется всегда, и спека, превращённая в TypeScript.</p><ol><li><a href="/ru/learn/use-other-modules/">19 Другие модули</a></li><li><a href="/ru/learn/invariants/">20 Инварианты и замеры</a></li><li><a href="/ru/learn/gen-ts/">21 gen-ts: спека превращается в TypeScript</a></li></ol></li><li><h3>Модуль 8 · Компилятор и tri</h3><p>Бэкенды, которые пишет t27c, стадии, в которых он читает спеку, и команда tri.</p><ol><li><a href="/ru/learn/seven-backends/">22 Семь бэкендов</a></li><li><a href="/ru/learn/stages-of-t27c/">23 Стадии t27c</a></li><li><a href="/ru/learn/tri-commands/">24 tri: командная строка</a></li></ol></li><li><h3>Модуль 9 · Ошибки, программа и что дальше</h3><p>Как читать ошибку компилятора, одна маленькая законченная программа и какой курс пройти дальше.</p><ol><li><a href="/ru/learn/reading-errors/">25 Как читать ошибки компилятора</a></li><li><a href="/ru/learn/a-small-program/">26 Маленькая законченная программа</a></li><li><a href="/ru/learn/where-next/">27 Куда дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/gated-clock-trap/index.html b/apps/website/public/ru/learn/gated-clock-trap/index.html
index 25205ba55c..11bce009f7 100644
--- a/apps/website/public/ru/learn/gated-clock-trap/index.html
+++ b/apps/website/public/ru/learn/gated-clock-trap/index.html
@@ -300,5 +300,6 @@ <h1>Ловушка стробированного клока</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое тактовый сигнал</h3><p>Один фронт, один мир: всё, что делит фронт тактового сигнала, делит и мир; период и джиттер настоящего фронта и место, где тактовый сигнал входит в плату.</p><ol><li><a href="/ru/learn/one-edge-one-world/">01 Один фронт, один мир</a></li><li><a href="/ru/learn/period-and-jitter/">02 Период, фронт и джиттер</a></li><li><a href="/ru/learn/where-the-clock-enters/">03 Где входит тактовый сигнал</a></li></ol></li><li><h3>Модуль 2 · Деревья тактового сигнала</h3><p>Перекос и задержка прохождения, глобальная сеть буферов и ловушка стробирования клока логикой.</p><ol><li><a href="/ru/learn/skew-and-insertion/">04 Перекос и задержка прохождения</a></li><li><a href="/ru/learn/global-buffers/">05 Глобальный буфер</a></li><li class="on" aria-current="page"><a href="/ru/learn/gated-clock-trap/">06 Ловушка стробированного клока</a></li></ol></li><li><h3>Модуль 3 · PLL и MMCM</h3><p>Умножить и поделить один тактовый сигнал в другой, сдвинуть его фазу шагами VCO и какие тактовые сигналы анализатор считает родственными.</p><ol><li><a href="/ru/learn/multiply-and-divide/">07 Умножить и поделить</a></li><li><a href="/ru/learn/phase-and-lock/">08 Фаза и захват</a></li><li><a href="/ru/learn/related-or-not/">09 Родственные или чужие</a></li></ol></li><li><h3>Модуль 4 · Сбросы</h3><p>Ассертировать асинхронно, снимать синхронно: три вида сброса, конвейер снятия и дерево, которое растит сброс.</p><ol><li><a href="/ru/learn/sync-or-async/">10 Синхронный или асинхронный</a></li><li><a href="/ru/learn/reset-release/">11 Конвейер снятия</a></li><li><a href="/ru/learn/reset-fanout/">12 Разветвление сброса</a></li></ol></li><li><h3>Модуль 5 · Метастабильность</h3><p>Окно setup-hold, среднее время между отказами в целочисленной арифметике и два триггера, которые это чинят.</p><ol><li><a href="/ru/learn/setup-hold-window/">13 Окно setup-hold</a></li><li><a href="/ru/learn/mtbf/">14 Среднее время между отказами</a></li><li><a href="/ru/learn/two-flop-sync/">15 Двухтриггерный синхронизатор</a></li></ol></li><li><h3>Модуль 6 · Пересечение многих бит</h3><p>Почему двоичная шина рвётся, почему код Грея нет, и рукопожатие, переносящее импульс между мирами.</p><ol><li><a href="/ru/learn/why-binary-fails/">16 Почему двоичный код рвётся</a></li><li><a href="/ru/learn/gray-code/">17 Код Грея</a></li><li><a href="/ru/learn/handshakes-and-pulses/">18 Рукопожатия и импульсы</a></li></ol></li><li><h3>Модуль 7 · Асинхронный FIFO</h3><p>Указатели, флаги и глубина: буфер, переносящий поток между двумя тактовыми сигналами.</p><ol><li><a href="/ru/learn/pointers-and-flags/">19 Указатели и флаги</a></li><li><a href="/ru/learn/depth-sizing/">20 Выбор глубины</a></li><li><a href="/ru/learn/fifo-testbench/">21 Тестбенч FIFO</a></li></ol></li><li><h3>Модуль 8 · Ограничения</h3><p>Строки, сообщающие анализатору, что такое тактовый сигнал, какие пути не проверять и чему должны соответствовать выводы.</p><ol><li><a href="/ru/learn/create-clock/">22 Строка create_clock</a></li><li><a href="/ru/learn/false-path-and-max-delay/">23 False path и max delay</a></li><li><a href="/ru/learn/io-timing/">24 Тайминг выводов</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Отчёт CDC, одно пересечение, захваченное на триггерах, и дифф битстрима, замыкающий курс.</p><ol><li><a href="/ru/learn/cdc-report/">25 Отчёт CDC</a></li><li><a href="/ru/learn/capture-a-crossing/">26 Захватить пересечение</a></li><li><a href="/ru/learn/capstone-diff/">27 Капстоун: поменяй и сравни</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/gen-ts/index.html b/apps/website/public/ru/learn/gen-ts/index.html
index f1251f7072..5b16bd1b4a 100644
--- a/apps/website/public/ru/learn/gen-ts/index.html
+++ b/apps/website/public/ru/learn/gen-ts/index.html
@@ -86,5 +86,6 @@ <h1>gen-ts: спека превращается в TypeScript</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Файл спеки</h3><p>Что лежит в файле .t27: строка module, строки прозы и комментарии.</p><ol><li><a href="/ru/learn/what-a-spec-is/">01 Что такое спека</a></li><li><a href="/ru/learn/the-module-line/">02 Строка module</a></li><li><a href="/ru/learn/comments-and-prose/">03 Комментарии и строки прозы</a></li></ol></li><li><h3>Модуль 2 · Значения и типы</h3><p>Константы, ширина целого числа, а также true, false и текст.</p><ol><li><a href="/ru/learn/constants/">04 Константы</a></li><li><a href="/ru/learn/whole-numbers/">05 Целые числа и их ширина</a></li><li><a href="/ru/learn/true-false-and-text/">06 Истина, ложь и текст</a></li></ol></li><li><h3>Модуль 3 · Массивы, триты и выражения</h3><p>Списки значений, три значения трита и что вычисляют операторы.</p><ol><li><a href="/ru/learn/arrays/">07 Массивы</a></li><li><a href="/ru/learn/trits/">08 Триты: минус один, ноль, один</a></li><li><a href="/ru/learn/expressions/">09 Выражения</a></li></ol></li><li><h3>Модуль 4 · Тесты и функции</h3><p>Блок test, несколько маленьких тестов в одной спеке и функция с типизированными входами.</p><ol><li><a href="/ru/learn/a-test-block/">10 Блок test</a></li><li><a href="/ru/learn/many-tests/">11 Много тестов в одной спеке</a></li><li><a href="/ru/learn/functions/">12 Функции</a></li></ol></li><li><h3>Модуль 5 · Внутри функции</h3><p>Имена через let и var, выбор через if и switch и циклы.</p><ol><li><a href="/ru/learn/local-names/">13 Локальные имена: let и var</a></li><li><a href="/ru/learn/if-and-else/">14 Выбор: if, else и switch</a></li><li><a href="/ru/learn/loops/">15 Циклы: while и for</a></li></ol></li><li><h3>Модуль 6 · Видимость и свои типы</h3><p>Что pub открывает другим модулям, и структуры и перечисления, которые вы объявляете.</p><ol><li><a href="/ru/learn/pub-or-private/">16 pub или закрытое имя</a></li><li><a href="/ru/learn/structs/">17 Структуры</a></li><li><a href="/ru/learn/enums/">18 Перечисления</a></li></ol></li><li><h3>Модуль 7 · Модули, правила и gen-ts</h3><p>Чтение другого модуля, правило, которое выполняется всегда, и спека, превращённая в TypeScript.</p><ol><li><a href="/ru/learn/use-other-modules/">19 Другие модули</a></li><li><a href="/ru/learn/invariants/">20 Инварианты и замеры</a></li><li class="on" aria-current="page"><a href="/ru/learn/gen-ts/">21 gen-ts: спека превращается в TypeScript</a></li></ol></li><li><h3>Модуль 8 · Компилятор и tri</h3><p>Бэкенды, которые пишет t27c, стадии, в которых он читает спеку, и команда tri.</p><ol><li><a href="/ru/learn/seven-backends/">22 Семь бэкендов</a></li><li><a href="/ru/learn/stages-of-t27c/">23 Стадии t27c</a></li><li><a href="/ru/learn/tri-commands/">24 tri: командная строка</a></li></ol></li><li><h3>Модуль 9 · Ошибки, программа и что дальше</h3><p>Как читать ошибку компилятора, одна маленькая законченная программа и какой курс пройти дальше.</p><ol><li><a href="/ru/learn/reading-errors/">25 Как читать ошибки компилятора</a></li><li><a href="/ru/learn/a-small-program/">26 Маленькая законченная программа</a></li><li><a href="/ru/learn/where-next/">27 Куда дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/gf10-ten-bits/index.html b/apps/website/public/ru/learn/gf10-ten-bits/index.html
index 6176aee451..1c081ca0e6 100644
--- a/apps/website/public/ru/learn/gf10-ten-bits/index.html
+++ b/apps/website/public/ru/learn/gf10-ten-bits/index.html
@@ -154,5 +154,6 @@ <h1>GF10: десять битов</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Правило и его числа</h3><p>Одно правило делит каждую ширину, отношение, к которому оно стремится, и числа Люка за тройкой 3.</p><ol><li><a href="/ru/learn/a-float-cut-by-phi/">01 Число с плавающей точкой, разрезанное по phi</a></li><li><a href="/ru/learn/phi-as-a-ratio/">02 Phi как отношение</a></li><li><a href="/ru/learn/lucas-numbers-stay-whole/">03 Числа Люка остаются целыми</a></li></ol></li><li><h3>Модуль 2 · Почему phi, почему три</h3><p>Почему деление идёт по phi, почему основание три и как спека проверяет, что GF16 хранит phi.</p><ol><li><a href="/ru/learn/why-the-split-is-phi/">04 Почему деление по phi</a></li><li><a href="/ru/learn/why-base-three/">05 Почему основание три</a></li><li><a href="/ru/learn/phi-in-sixteen-bits/">06 Phi в шестнадцати битах</a></li></ol></li><li><h3>Модуль 3 · Малые ступени: от GF4 до GF8</h3><p>GF4, GF6 и GF8 — меньше всего битов, и округление до целых битов стоит здесь дороже всего.</p><ol><li><a href="/ru/learn/gf4-four-bits/">07 GF4: четыре бита</a></li><li><a href="/ru/learn/gf6-six-bits/">08 GF6: шесть битов</a></li><li><a href="/ru/learn/gf8-one-byte/">09 GF8: один байт</a></li></ol></li><li><h3>Модуль 4 · От десяти до четырнадцати битов</h3><p>GF10, GF12 и GF14 и то, как расстояние до 1 / phi меняется с ростом слова.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/gf10-ten-bits/">10 GF10: десять битов</a></li><li><a href="/ru/learn/gf12-twelve-bits/">11 GF12: двенадцать битов</a></li><li><a href="/ru/learn/gf14-fourteen-bits/">12 GF14: четырнадцать битов</a></li></ol></li><li><h3>Модуль 5 · GF16 в работе</h3><p>Основной 16-битный формат, скалярное произведение из двух слагаемых в GF-T16, затем GF20 и GF24.</p><ol><li><a href="/ru/learn/gf16-the-primary/">13 GF16: основной формат</a></li><li><a href="/ru/learn/two-products-one-sum/">14 Два произведения, одна сумма</a></li><li><a href="/ru/learn/gf24-twenty-four-bits/">15 GF20 и GF24</a></li></ol></li><li><h3>Модуль 6 · От GF32 до GF64</h3><p>GF32 рядом с IEEE single, GF48 без пары в IEEE, GF64 рядом с IEEE double.</p><ol><li><a href="/ru/learn/gf32-a-single/">16 GF32: одинарная точность</a></li><li><a href="/ru/learn/gf48-forty-eight-bits/">17 GF48: сорок восемь битов</a></li><li><a href="/ru/learn/gf64-a-double/">18 GF64: двойная точность</a></li></ol></li><li><h3>Модуль 7 · От GF96 до GF256</h3><p>GF96, GF128 и GF256, где спеки держат раскладку инвариантами.</p><ol><li><a href="/ru/learn/gf96-ninety-six-bits/">19 GF96: девяносто шесть битов</a></li><li><a href="/ru/learn/gf128-a-quad/">20 GF128: четверная точность</a></li><li><a href="/ru/learn/gf256-two-five-six/">21 GF256: двести пятьдесят шесть битов</a></li></ol></li><li><h3>Модуль 8 · Самые широкие ступени, затем триты</h3><p>GF512 и GF1024, две самые широкие ступени, затем GF-T8, где порядок уходит в триты.</p><ol><li><a href="/ru/learn/gf512-five-twelve/">22 GF512: пятьсот двенадцать битов</a></li><li><a href="/ru/learn/gf1024-one-kilobit/">23 GF1024: один килобит</a></li><li><a href="/ru/learn/gft8-exponent-in-trits/">24 GF-T8: порядок в тритах</a></li></ol></li><li><h3>Модуль 9 · Ещё триты, затем декодирование</h3><p>GF-T16 и GF-T32, затем почему фиксированные поля декодируются параллельно, а posit — нет.</p><ol><li><a href="/ru/learn/gft16-six-trits/">25 GF-T16: шесть тритов</a></li><li><a href="/ru/learn/gft32-twelve-trits/">26 GF-T32: двенадцать тритов</a></li><li><a href="/ru/learn/decode-in-one-step/">27 Раскодировать за один шаг</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/gf1024-one-kilobit/index.html b/apps/website/public/ru/learn/gf1024-one-kilobit/index.html
index 20127f99f8..70377342c0 100644
--- a/apps/website/public/ru/learn/gf1024-one-kilobit/index.html
+++ b/apps/website/public/ru/learn/gf1024-one-kilobit/index.html
@@ -152,5 +152,6 @@ <h1>GF1024: один килобит</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Правило и его числа</h3><p>Одно правило делит каждую ширину, отношение, к которому оно стремится, и числа Люка за тройкой 3.</p><ol><li><a href="/ru/learn/a-float-cut-by-phi/">01 Число с плавающей точкой, разрезанное по phi</a></li><li><a href="/ru/learn/phi-as-a-ratio/">02 Phi как отношение</a></li><li><a href="/ru/learn/lucas-numbers-stay-whole/">03 Числа Люка остаются целыми</a></li></ol></li><li><h3>Модуль 2 · Почему phi, почему три</h3><p>Почему деление идёт по phi, почему основание три и как спека проверяет, что GF16 хранит phi.</p><ol><li><a href="/ru/learn/why-the-split-is-phi/">04 Почему деление по phi</a></li><li><a href="/ru/learn/why-base-three/">05 Почему основание три</a></li><li><a href="/ru/learn/phi-in-sixteen-bits/">06 Phi в шестнадцати битах</a></li></ol></li><li><h3>Модуль 3 · Малые ступени: от GF4 до GF8</h3><p>GF4, GF6 и GF8 — меньше всего битов, и округление до целых битов стоит здесь дороже всего.</p><ol><li><a href="/ru/learn/gf4-four-bits/">07 GF4: четыре бита</a></li><li><a href="/ru/learn/gf6-six-bits/">08 GF6: шесть битов</a></li><li><a href="/ru/learn/gf8-one-byte/">09 GF8: один байт</a></li></ol></li><li><h3>Модуль 4 · От десяти до четырнадцати битов</h3><p>GF10, GF12 и GF14 и то, как расстояние до 1 / phi меняется с ростом слова.</p><ol><li><a href="/ru/learn/gf10-ten-bits/">10 GF10: десять битов</a></li><li><a href="/ru/learn/gf12-twelve-bits/">11 GF12: двенадцать битов</a></li><li><a href="/ru/learn/gf14-fourteen-bits/">12 GF14: четырнадцать битов</a></li></ol></li><li><h3>Модуль 5 · GF16 в работе</h3><p>Основной 16-битный формат, скалярное произведение из двух слагаемых в GF-T16, затем GF20 и GF24.</p><ol><li><a href="/ru/learn/gf16-the-primary/">13 GF16: основной формат</a></li><li><a href="/ru/learn/two-products-one-sum/">14 Два произведения, одна сумма</a></li><li><a href="/ru/learn/gf24-twenty-four-bits/">15 GF20 и GF24</a></li></ol></li><li><h3>Модуль 6 · От GF32 до GF64</h3><p>GF32 рядом с IEEE single, GF48 без пары в IEEE, GF64 рядом с IEEE double.</p><ol><li><a href="/ru/learn/gf32-a-single/">16 GF32: одинарная точность</a></li><li><a href="/ru/learn/gf48-forty-eight-bits/">17 GF48: сорок восемь битов</a></li><li><a href="/ru/learn/gf64-a-double/">18 GF64: двойная точность</a></li></ol></li><li><h3>Модуль 7 · От GF96 до GF256</h3><p>GF96, GF128 и GF256, где спеки держат раскладку инвариантами.</p><ol><li><a href="/ru/learn/gf96-ninety-six-bits/">19 GF96: девяносто шесть битов</a></li><li><a href="/ru/learn/gf128-a-quad/">20 GF128: четверная точность</a></li><li><a href="/ru/learn/gf256-two-five-six/">21 GF256: двести пятьдесят шесть битов</a></li></ol></li><li><h3>Модуль 8 · Самые широкие ступени, затем триты</h3><p>GF512 и GF1024, две самые широкие ступени, затем GF-T8, где порядок уходит в триты.</p><ol><li><a href="/ru/learn/gf512-five-twelve/">22 GF512: пятьсот двенадцать битов</a></li><li class="on" aria-current="page"><a href="/ru/learn/gf1024-one-kilobit/">23 GF1024: один килобит</a></li><li><a href="/ru/learn/gft8-exponent-in-trits/">24 GF-T8: порядок в тритах</a></li></ol></li><li><h3>Модуль 9 · Ещё триты, затем декодирование</h3><p>GF-T16 и GF-T32, затем почему фиксированные поля декодируются параллельно, а posit — нет.</p><ol><li><a href="/ru/learn/gft16-six-trits/">25 GF-T16: шесть тритов</a></li><li><a href="/ru/learn/gft32-twelve-trits/">26 GF-T32: двенадцать тритов</a></li><li><a href="/ru/learn/decode-in-one-step/">27 Раскодировать за один шаг</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/gf12-twelve-bits/index.html b/apps/website/public/ru/learn/gf12-twelve-bits/index.html
index 4c8aff1426..9d71209c3b 100644
--- a/apps/website/public/ru/learn/gf12-twelve-bits/index.html
+++ b/apps/website/public/ru/learn/gf12-twelve-bits/index.html
@@ -562,5 +562,6 @@ <h1>GF12: двенадцать битов</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Правило и его числа</h3><p>Одно правило делит каждую ширину, отношение, к которому оно стремится, и числа Люка за тройкой 3.</p><ol><li><a href="/ru/learn/a-float-cut-by-phi/">01 Число с плавающей точкой, разрезанное по phi</a></li><li><a href="/ru/learn/phi-as-a-ratio/">02 Phi как отношение</a></li><li><a href="/ru/learn/lucas-numbers-stay-whole/">03 Числа Люка остаются целыми</a></li></ol></li><li><h3>Модуль 2 · Почему phi, почему три</h3><p>Почему деление идёт по phi, почему основание три и как спека проверяет, что GF16 хранит phi.</p><ol><li><a href="/ru/learn/why-the-split-is-phi/">04 Почему деление по phi</a></li><li><a href="/ru/learn/why-base-three/">05 Почему основание три</a></li><li><a href="/ru/learn/phi-in-sixteen-bits/">06 Phi в шестнадцати битах</a></li></ol></li><li><h3>Модуль 3 · Малые ступени: от GF4 до GF8</h3><p>GF4, GF6 и GF8 — меньше всего битов, и округление до целых битов стоит здесь дороже всего.</p><ol><li><a href="/ru/learn/gf4-four-bits/">07 GF4: четыре бита</a></li><li><a href="/ru/learn/gf6-six-bits/">08 GF6: шесть битов</a></li><li><a href="/ru/learn/gf8-one-byte/">09 GF8: один байт</a></li></ol></li><li><h3>Модуль 4 · От десяти до четырнадцати битов</h3><p>GF10, GF12 и GF14 и то, как расстояние до 1 / phi меняется с ростом слова.</p><ol><li><a href="/ru/learn/gf10-ten-bits/">10 GF10: десять битов</a></li><li class="on" aria-current="page"><a href="/ru/learn/gf12-twelve-bits/">11 GF12: двенадцать битов</a></li><li><a href="/ru/learn/gf14-fourteen-bits/">12 GF14: четырнадцать битов</a></li></ol></li><li><h3>Модуль 5 · GF16 в работе</h3><p>Основной 16-битный формат, скалярное произведение из двух слагаемых в GF-T16, затем GF20 и GF24.</p><ol><li><a href="/ru/learn/gf16-the-primary/">13 GF16: основной формат</a></li><li><a href="/ru/learn/two-products-one-sum/">14 Два произведения, одна сумма</a></li><li><a href="/ru/learn/gf24-twenty-four-bits/">15 GF20 и GF24</a></li></ol></li><li><h3>Модуль 6 · От GF32 до GF64</h3><p>GF32 рядом с IEEE single, GF48 без пары в IEEE, GF64 рядом с IEEE double.</p><ol><li><a href="/ru/learn/gf32-a-single/">16 GF32: одинарная точность</a></li><li><a href="/ru/learn/gf48-forty-eight-bits/">17 GF48: сорок восемь битов</a></li><li><a href="/ru/learn/gf64-a-double/">18 GF64: двойная точность</a></li></ol></li><li><h3>Модуль 7 · От GF96 до GF256</h3><p>GF96, GF128 и GF256, где спеки держат раскладку инвариантами.</p><ol><li><a href="/ru/learn/gf96-ninety-six-bits/">19 GF96: девяносто шесть битов</a></li><li><a href="/ru/learn/gf128-a-quad/">20 GF128: четверная точность</a></li><li><a href="/ru/learn/gf256-two-five-six/">21 GF256: двести пятьдесят шесть битов</a></li></ol></li><li><h3>Модуль 8 · Самые широкие ступени, затем триты</h3><p>GF512 и GF1024, две самые широкие ступени, затем GF-T8, где порядок уходит в триты.</p><ol><li><a href="/ru/learn/gf512-five-twelve/">22 GF512: пятьсот двенадцать битов</a></li><li><a href="/ru/learn/gf1024-one-kilobit/">23 GF1024: один килобит</a></li><li><a href="/ru/learn/gft8-exponent-in-trits/">24 GF-T8: порядок в тритах</a></li></ol></li><li><h3>Модуль 9 · Ещё триты, затем декодирование</h3><p>GF-T16 и GF-T32, затем почему фиксированные поля декодируются параллельно, а posit — нет.</p><ol><li><a href="/ru/learn/gft16-six-trits/">25 GF-T16: шесть тритов</a></li><li><a href="/ru/learn/gft32-twelve-trits/">26 GF-T32: двенадцать тритов</a></li><li><a href="/ru/learn/decode-in-one-step/">27 Раскодировать за один шаг</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/gf128-a-quad/index.html b/apps/website/public/ru/learn/gf128-a-quad/index.html
index be2bfccafb..ad6e99a17b 100644
--- a/apps/website/public/ru/learn/gf128-a-quad/index.html
+++ b/apps/website/public/ru/learn/gf128-a-quad/index.html
@@ -146,5 +146,6 @@ <h1>GF128: четверная точность</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Правило и его числа</h3><p>Одно правило делит каждую ширину, отношение, к которому оно стремится, и числа Люка за тройкой 3.</p><ol><li><a href="/ru/learn/a-float-cut-by-phi/">01 Число с плавающей точкой, разрезанное по phi</a></li><li><a href="/ru/learn/phi-as-a-ratio/">02 Phi как отношение</a></li><li><a href="/ru/learn/lucas-numbers-stay-whole/">03 Числа Люка остаются целыми</a></li></ol></li><li><h3>Модуль 2 · Почему phi, почему три</h3><p>Почему деление идёт по phi, почему основание три и как спека проверяет, что GF16 хранит phi.</p><ol><li><a href="/ru/learn/why-the-split-is-phi/">04 Почему деление по phi</a></li><li><a href="/ru/learn/why-base-three/">05 Почему основание три</a></li><li><a href="/ru/learn/phi-in-sixteen-bits/">06 Phi в шестнадцати битах</a></li></ol></li><li><h3>Модуль 3 · Малые ступени: от GF4 до GF8</h3><p>GF4, GF6 и GF8 — меньше всего битов, и округление до целых битов стоит здесь дороже всего.</p><ol><li><a href="/ru/learn/gf4-four-bits/">07 GF4: четыре бита</a></li><li><a href="/ru/learn/gf6-six-bits/">08 GF6: шесть битов</a></li><li><a href="/ru/learn/gf8-one-byte/">09 GF8: один байт</a></li></ol></li><li><h3>Модуль 4 · От десяти до четырнадцати битов</h3><p>GF10, GF12 и GF14 и то, как расстояние до 1 / phi меняется с ростом слова.</p><ol><li><a href="/ru/learn/gf10-ten-bits/">10 GF10: десять битов</a></li><li><a href="/ru/learn/gf12-twelve-bits/">11 GF12: двенадцать битов</a></li><li><a href="/ru/learn/gf14-fourteen-bits/">12 GF14: четырнадцать битов</a></li></ol></li><li><h3>Модуль 5 · GF16 в работе</h3><p>Основной 16-битный формат, скалярное произведение из двух слагаемых в GF-T16, затем GF20 и GF24.</p><ol><li><a href="/ru/learn/gf16-the-primary/">13 GF16: основной формат</a></li><li><a href="/ru/learn/two-products-one-sum/">14 Два произведения, одна сумма</a></li><li><a href="/ru/learn/gf24-twenty-four-bits/">15 GF20 и GF24</a></li></ol></li><li><h3>Модуль 6 · От GF32 до GF64</h3><p>GF32 рядом с IEEE single, GF48 без пары в IEEE, GF64 рядом с IEEE double.</p><ol><li><a href="/ru/learn/gf32-a-single/">16 GF32: одинарная точность</a></li><li><a href="/ru/learn/gf48-forty-eight-bits/">17 GF48: сорок восемь битов</a></li><li><a href="/ru/learn/gf64-a-double/">18 GF64: двойная точность</a></li></ol></li><li><h3>Модуль 7 · От GF96 до GF256</h3><p>GF96, GF128 и GF256, где спеки держат раскладку инвариантами.</p><ol><li><a href="/ru/learn/gf96-ninety-six-bits/">19 GF96: девяносто шесть битов</a></li><li class="on" aria-current="page"><a href="/ru/learn/gf128-a-quad/">20 GF128: четверная точность</a></li><li><a href="/ru/learn/gf256-two-five-six/">21 GF256: двести пятьдесят шесть битов</a></li></ol></li><li><h3>Модуль 8 · Самые широкие ступени, затем триты</h3><p>GF512 и GF1024, две самые широкие ступени, затем GF-T8, где порядок уходит в триты.</p><ol><li><a href="/ru/learn/gf512-five-twelve/">22 GF512: пятьсот двенадцать битов</a></li><li><a href="/ru/learn/gf1024-one-kilobit/">23 GF1024: один килобит</a></li><li><a href="/ru/learn/gft8-exponent-in-trits/">24 GF-T8: порядок в тритах</a></li></ol></li><li><h3>Модуль 9 · Ещё триты, затем декодирование</h3><p>GF-T16 и GF-T32, затем почему фиксированные поля декодируются параллельно, а posit — нет.</p><ol><li><a href="/ru/learn/gft16-six-trits/">25 GF-T16: шесть тритов</a></li><li><a href="/ru/learn/gft32-twelve-trits/">26 GF-T32: двенадцать тритов</a></li><li><a href="/ru/learn/decode-in-one-step/">27 Раскодировать за один шаг</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/gf14-fourteen-bits/index.html b/apps/website/public/ru/learn/gf14-fourteen-bits/index.html
index f4b6fc793e..b0443dd623 100644
--- a/apps/website/public/ru/learn/gf14-fourteen-bits/index.html
+++ b/apps/website/public/ru/learn/gf14-fourteen-bits/index.html
@@ -154,5 +154,6 @@ <h1>GF14: четырнадцать битов</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Правило и его числа</h3><p>Одно правило делит каждую ширину, отношение, к которому оно стремится, и числа Люка за тройкой 3.</p><ol><li><a href="/ru/learn/a-float-cut-by-phi/">01 Число с плавающей точкой, разрезанное по phi</a></li><li><a href="/ru/learn/phi-as-a-ratio/">02 Phi как отношение</a></li><li><a href="/ru/learn/lucas-numbers-stay-whole/">03 Числа Люка остаются целыми</a></li></ol></li><li><h3>Модуль 2 · Почему phi, почему три</h3><p>Почему деление идёт по phi, почему основание три и как спека проверяет, что GF16 хранит phi.</p><ol><li><a href="/ru/learn/why-the-split-is-phi/">04 Почему деление по phi</a></li><li><a href="/ru/learn/why-base-three/">05 Почему основание три</a></li><li><a href="/ru/learn/phi-in-sixteen-bits/">06 Phi в шестнадцати битах</a></li></ol></li><li><h3>Модуль 3 · Малые ступени: от GF4 до GF8</h3><p>GF4, GF6 и GF8 — меньше всего битов, и округление до целых битов стоит здесь дороже всего.</p><ol><li><a href="/ru/learn/gf4-four-bits/">07 GF4: четыре бита</a></li><li><a href="/ru/learn/gf6-six-bits/">08 GF6: шесть битов</a></li><li><a href="/ru/learn/gf8-one-byte/">09 GF8: один байт</a></li></ol></li><li><h3>Модуль 4 · От десяти до четырнадцати битов</h3><p>GF10, GF12 и GF14 и то, как расстояние до 1 / phi меняется с ростом слова.</p><ol><li><a href="/ru/learn/gf10-ten-bits/">10 GF10: десять битов</a></li><li><a href="/ru/learn/gf12-twelve-bits/">11 GF12: двенадцать битов</a></li><li class="on" aria-current="page"><a href="/ru/learn/gf14-fourteen-bits/">12 GF14: четырнадцать битов</a></li></ol></li><li><h3>Модуль 5 · GF16 в работе</h3><p>Основной 16-битный формат, скалярное произведение из двух слагаемых в GF-T16, затем GF20 и GF24.</p><ol><li><a href="/ru/learn/gf16-the-primary/">13 GF16: основной формат</a></li><li><a href="/ru/learn/two-products-one-sum/">14 Два произведения, одна сумма</a></li><li><a href="/ru/learn/gf24-twenty-four-bits/">15 GF20 и GF24</a></li></ol></li><li><h3>Модуль 6 · От GF32 до GF64</h3><p>GF32 рядом с IEEE single, GF48 без пары в IEEE, GF64 рядом с IEEE double.</p><ol><li><a href="/ru/learn/gf32-a-single/">16 GF32: одинарная точность</a></li><li><a href="/ru/learn/gf48-forty-eight-bits/">17 GF48: сорок восемь битов</a></li><li><a href="/ru/learn/gf64-a-double/">18 GF64: двойная точность</a></li></ol></li><li><h3>Модуль 7 · От GF96 до GF256</h3><p>GF96, GF128 и GF256, где спеки держат раскладку инвариантами.</p><ol><li><a href="/ru/learn/gf96-ninety-six-bits/">19 GF96: девяносто шесть битов</a></li><li><a href="/ru/learn/gf128-a-quad/">20 GF128: четверная точность</a></li><li><a href="/ru/learn/gf256-two-five-six/">21 GF256: двести пятьдесят шесть битов</a></li></ol></li><li><h3>Модуль 8 · Самые широкие ступени, затем триты</h3><p>GF512 и GF1024, две самые широкие ступени, затем GF-T8, где порядок уходит в триты.</p><ol><li><a href="/ru/learn/gf512-five-twelve/">22 GF512: пятьсот двенадцать битов</a></li><li><a href="/ru/learn/gf1024-one-kilobit/">23 GF1024: один килобит</a></li><li><a href="/ru/learn/gft8-exponent-in-trits/">24 GF-T8: порядок в тритах</a></li></ol></li><li><h3>Модуль 9 · Ещё триты, затем декодирование</h3><p>GF-T16 и GF-T32, затем почему фиксированные поля декодируются параллельно, а posit — нет.</p><ol><li><a href="/ru/learn/gft16-six-trits/">25 GF-T16: шесть тритов</a></li><li><a href="/ru/learn/gft32-twelve-trits/">26 GF-T32: двенадцать тритов</a></li><li><a href="/ru/learn/decode-in-one-step/">27 Раскодировать за один шаг</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/gf16-the-primary/index.html b/apps/website/public/ru/learn/gf16-the-primary/index.html
index ccdb430d4b..0662dca1a0 100644
--- a/apps/website/public/ru/learn/gf16-the-primary/index.html
+++ b/apps/website/public/ru/learn/gf16-the-primary/index.html
@@ -3863,5 +3863,6 @@ <h1>GF16: основной формат</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Правило и его числа</h3><p>Одно правило делит каждую ширину, отношение, к которому оно стремится, и числа Люка за тройкой 3.</p><ol><li><a href="/ru/learn/a-float-cut-by-phi/">01 Число с плавающей точкой, разрезанное по phi</a></li><li><a href="/ru/learn/phi-as-a-ratio/">02 Phi как отношение</a></li><li><a href="/ru/learn/lucas-numbers-stay-whole/">03 Числа Люка остаются целыми</a></li></ol></li><li><h3>Модуль 2 · Почему phi, почему три</h3><p>Почему деление идёт по phi, почему основание три и как спека проверяет, что GF16 хранит phi.</p><ol><li><a href="/ru/learn/why-the-split-is-phi/">04 Почему деление по phi</a></li><li><a href="/ru/learn/why-base-three/">05 Почему основание три</a></li><li><a href="/ru/learn/phi-in-sixteen-bits/">06 Phi в шестнадцати битах</a></li></ol></li><li><h3>Модуль 3 · Малые ступени: от GF4 до GF8</h3><p>GF4, GF6 и GF8 — меньше всего битов, и округление до целых битов стоит здесь дороже всего.</p><ol><li><a href="/ru/learn/gf4-four-bits/">07 GF4: четыре бита</a></li><li><a href="/ru/learn/gf6-six-bits/">08 GF6: шесть битов</a></li><li><a href="/ru/learn/gf8-one-byte/">09 GF8: один байт</a></li></ol></li><li><h3>Модуль 4 · От десяти до четырнадцати битов</h3><p>GF10, GF12 и GF14 и то, как расстояние до 1 / phi меняется с ростом слова.</p><ol><li><a href="/ru/learn/gf10-ten-bits/">10 GF10: десять битов</a></li><li><a href="/ru/learn/gf12-twelve-bits/">11 GF12: двенадцать битов</a></li><li><a href="/ru/learn/gf14-fourteen-bits/">12 GF14: четырнадцать битов</a></li></ol></li><li><h3>Модуль 5 · GF16 в работе</h3><p>Основной 16-битный формат, скалярное произведение из двух слагаемых в GF-T16, затем GF20 и GF24.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/gf16-the-primary/">13 GF16: основной формат</a></li><li><a href="/ru/learn/two-products-one-sum/">14 Два произведения, одна сумма</a></li><li><a href="/ru/learn/gf24-twenty-four-bits/">15 GF20 и GF24</a></li></ol></li><li><h3>Модуль 6 · От GF32 до GF64</h3><p>GF32 рядом с IEEE single, GF48 без пары в IEEE, GF64 рядом с IEEE double.</p><ol><li><a href="/ru/learn/gf32-a-single/">16 GF32: одинарная точность</a></li><li><a href="/ru/learn/gf48-forty-eight-bits/">17 GF48: сорок восемь битов</a></li><li><a href="/ru/learn/gf64-a-double/">18 GF64: двойная точность</a></li></ol></li><li><h3>Модуль 7 · От GF96 до GF256</h3><p>GF96, GF128 и GF256, где спеки держат раскладку инвариантами.</p><ol><li><a href="/ru/learn/gf96-ninety-six-bits/">19 GF96: девяносто шесть битов</a></li><li><a href="/ru/learn/gf128-a-quad/">20 GF128: четверная точность</a></li><li><a href="/ru/learn/gf256-two-five-six/">21 GF256: двести пятьдесят шесть битов</a></li></ol></li><li><h3>Модуль 8 · Самые широкие ступени, затем триты</h3><p>GF512 и GF1024, две самые широкие ступени, затем GF-T8, где порядок уходит в триты.</p><ol><li><a href="/ru/learn/gf512-five-twelve/">22 GF512: пятьсот двенадцать битов</a></li><li><a href="/ru/learn/gf1024-one-kilobit/">23 GF1024: один килобит</a></li><li><a href="/ru/learn/gft8-exponent-in-trits/">24 GF-T8: порядок в тритах</a></li></ol></li><li><h3>Модуль 9 · Ещё триты, затем декодирование</h3><p>GF-T16 и GF-T32, затем почему фиксированные поля декодируются параллельно, а posit — нет.</p><ol><li><a href="/ru/learn/gft16-six-trits/">25 GF-T16: шесть тритов</a></li><li><a href="/ru/learn/gft32-twelve-trits/">26 GF-T32: двенадцать тритов</a></li><li><a href="/ru/learn/decode-in-one-step/">27 Раскодировать за один шаг</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/gf24-twenty-four-bits/index.html b/apps/website/public/ru/learn/gf24-twenty-four-bits/index.html
index f3b3da250f..fcf01d5541 100644
--- a/apps/website/public/ru/learn/gf24-twenty-four-bits/index.html
+++ b/apps/website/public/ru/learn/gf24-twenty-four-bits/index.html
@@ -549,5 +549,6 @@ <h1>GF20 и GF24</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Правило и его числа</h3><p>Одно правило делит каждую ширину, отношение, к которому оно стремится, и числа Люка за тройкой 3.</p><ol><li><a href="/ru/learn/a-float-cut-by-phi/">01 Число с плавающей точкой, разрезанное по phi</a></li><li><a href="/ru/learn/phi-as-a-ratio/">02 Phi как отношение</a></li><li><a href="/ru/learn/lucas-numbers-stay-whole/">03 Числа Люка остаются целыми</a></li></ol></li><li><h3>Модуль 2 · Почему phi, почему три</h3><p>Почему деление идёт по phi, почему основание три и как спека проверяет, что GF16 хранит phi.</p><ol><li><a href="/ru/learn/why-the-split-is-phi/">04 Почему деление по phi</a></li><li><a href="/ru/learn/why-base-three/">05 Почему основание три</a></li><li><a href="/ru/learn/phi-in-sixteen-bits/">06 Phi в шестнадцати битах</a></li></ol></li><li><h3>Модуль 3 · Малые ступени: от GF4 до GF8</h3><p>GF4, GF6 и GF8 — меньше всего битов, и округление до целых битов стоит здесь дороже всего.</p><ol><li><a href="/ru/learn/gf4-four-bits/">07 GF4: четыре бита</a></li><li><a href="/ru/learn/gf6-six-bits/">08 GF6: шесть битов</a></li><li><a href="/ru/learn/gf8-one-byte/">09 GF8: один байт</a></li></ol></li><li><h3>Модуль 4 · От десяти до четырнадцати битов</h3><p>GF10, GF12 и GF14 и то, как расстояние до 1 / phi меняется с ростом слова.</p><ol><li><a href="/ru/learn/gf10-ten-bits/">10 GF10: десять битов</a></li><li><a href="/ru/learn/gf12-twelve-bits/">11 GF12: двенадцать битов</a></li><li><a href="/ru/learn/gf14-fourteen-bits/">12 GF14: четырнадцать битов</a></li></ol></li><li><h3>Модуль 5 · GF16 в работе</h3><p>Основной 16-битный формат, скалярное произведение из двух слагаемых в GF-T16, затем GF20 и GF24.</p><ol><li><a href="/ru/learn/gf16-the-primary/">13 GF16: основной формат</a></li><li><a href="/ru/learn/two-products-one-sum/">14 Два произведения, одна сумма</a></li><li class="on" aria-current="page"><a href="/ru/learn/gf24-twenty-four-bits/">15 GF20 и GF24</a></li></ol></li><li><h3>Модуль 6 · От GF32 до GF64</h3><p>GF32 рядом с IEEE single, GF48 без пары в IEEE, GF64 рядом с IEEE double.</p><ol><li><a href="/ru/learn/gf32-a-single/">16 GF32: одинарная точность</a></li><li><a href="/ru/learn/gf48-forty-eight-bits/">17 GF48: сорок восемь битов</a></li><li><a href="/ru/learn/gf64-a-double/">18 GF64: двойная точность</a></li></ol></li><li><h3>Модуль 7 · От GF96 до GF256</h3><p>GF96, GF128 и GF256, где спеки держат раскладку инвариантами.</p><ol><li><a href="/ru/learn/gf96-ninety-six-bits/">19 GF96: девяносто шесть битов</a></li><li><a href="/ru/learn/gf128-a-quad/">20 GF128: четверная точность</a></li><li><a href="/ru/learn/gf256-two-five-six/">21 GF256: двести пятьдесят шесть битов</a></li></ol></li><li><h3>Модуль 8 · Самые широкие ступени, затем триты</h3><p>GF512 и GF1024, две самые широкие ступени, затем GF-T8, где порядок уходит в триты.</p><ol><li><a href="/ru/learn/gf512-five-twelve/">22 GF512: пятьсот двенадцать битов</a></li><li><a href="/ru/learn/gf1024-one-kilobit/">23 GF1024: один килобит</a></li><li><a href="/ru/learn/gft8-exponent-in-trits/">24 GF-T8: порядок в тритах</a></li></ol></li><li><h3>Модуль 9 · Ещё триты, затем декодирование</h3><p>GF-T16 и GF-T32, затем почему фиксированные поля декодируются параллельно, а posit — нет.</p><ol><li><a href="/ru/learn/gft16-six-trits/">25 GF-T16: шесть тритов</a></li><li><a href="/ru/learn/gft32-twelve-trits/">26 GF-T32: двенадцать тритов</a></li><li><a href="/ru/learn/decode-in-one-step/">27 Раскодировать за один шаг</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/gf256-two-five-six/index.html b/apps/website/public/ru/learn/gf256-two-five-six/index.html
index a57b309363..2b6125cab0 100644
--- a/apps/website/public/ru/learn/gf256-two-five-six/index.html
+++ b/apps/website/public/ru/learn/gf256-two-five-six/index.html
@@ -146,5 +146,6 @@ <h1>GF256: двести пятьдесят шесть битов</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Правило и его числа</h3><p>Одно правило делит каждую ширину, отношение, к которому оно стремится, и числа Люка за тройкой 3.</p><ol><li><a href="/ru/learn/a-float-cut-by-phi/">01 Число с плавающей точкой, разрезанное по phi</a></li><li><a href="/ru/learn/phi-as-a-ratio/">02 Phi как отношение</a></li><li><a href="/ru/learn/lucas-numbers-stay-whole/">03 Числа Люка остаются целыми</a></li></ol></li><li><h3>Модуль 2 · Почему phi, почему три</h3><p>Почему деление идёт по phi, почему основание три и как спека проверяет, что GF16 хранит phi.</p><ol><li><a href="/ru/learn/why-the-split-is-phi/">04 Почему деление по phi</a></li><li><a href="/ru/learn/why-base-three/">05 Почему основание три</a></li><li><a href="/ru/learn/phi-in-sixteen-bits/">06 Phi в шестнадцати битах</a></li></ol></li><li><h3>Модуль 3 · Малые ступени: от GF4 до GF8</h3><p>GF4, GF6 и GF8 — меньше всего битов, и округление до целых битов стоит здесь дороже всего.</p><ol><li><a href="/ru/learn/gf4-four-bits/">07 GF4: четыре бита</a></li><li><a href="/ru/learn/gf6-six-bits/">08 GF6: шесть битов</a></li><li><a href="/ru/learn/gf8-one-byte/">09 GF8: один байт</a></li></ol></li><li><h3>Модуль 4 · От десяти до четырнадцати битов</h3><p>GF10, GF12 и GF14 и то, как расстояние до 1 / phi меняется с ростом слова.</p><ol><li><a href="/ru/learn/gf10-ten-bits/">10 GF10: десять битов</a></li><li><a href="/ru/learn/gf12-twelve-bits/">11 GF12: двенадцать битов</a></li><li><a href="/ru/learn/gf14-fourteen-bits/">12 GF14: четырнадцать битов</a></li></ol></li><li><h3>Модуль 5 · GF16 в работе</h3><p>Основной 16-битный формат, скалярное произведение из двух слагаемых в GF-T16, затем GF20 и GF24.</p><ol><li><a href="/ru/learn/gf16-the-primary/">13 GF16: основной формат</a></li><li><a href="/ru/learn/two-products-one-sum/">14 Два произведения, одна сумма</a></li><li><a href="/ru/learn/gf24-twenty-four-bits/">15 GF20 и GF24</a></li></ol></li><li><h3>Модуль 6 · От GF32 до GF64</h3><p>GF32 рядом с IEEE single, GF48 без пары в IEEE, GF64 рядом с IEEE double.</p><ol><li><a href="/ru/learn/gf32-a-single/">16 GF32: одинарная точность</a></li><li><a href="/ru/learn/gf48-forty-eight-bits/">17 GF48: сорок восемь битов</a></li><li><a href="/ru/learn/gf64-a-double/">18 GF64: двойная точность</a></li></ol></li><li><h3>Модуль 7 · От GF96 до GF256</h3><p>GF96, GF128 и GF256, где спеки держат раскладку инвариантами.</p><ol><li><a href="/ru/learn/gf96-ninety-six-bits/">19 GF96: девяносто шесть битов</a></li><li><a href="/ru/learn/gf128-a-quad/">20 GF128: четверная точность</a></li><li class="on" aria-current="page"><a href="/ru/learn/gf256-two-five-six/">21 GF256: двести пятьдесят шесть битов</a></li></ol></li><li><h3>Модуль 8 · Самые широкие ступени, затем триты</h3><p>GF512 и GF1024, две самые широкие ступени, затем GF-T8, где порядок уходит в триты.</p><ol><li><a href="/ru/learn/gf512-five-twelve/">22 GF512: пятьсот двенадцать битов</a></li><li><a href="/ru/learn/gf1024-one-kilobit/">23 GF1024: один килобит</a></li><li><a href="/ru/learn/gft8-exponent-in-trits/">24 GF-T8: порядок в тритах</a></li></ol></li><li><h3>Модуль 9 · Ещё триты, затем декодирование</h3><p>GF-T16 и GF-T32, затем почему фиксированные поля декодируются параллельно, а posit — нет.</p><ol><li><a href="/ru/learn/gft16-six-trits/">25 GF-T16: шесть тритов</a></li><li><a href="/ru/learn/gft32-twelve-trits/">26 GF-T32: двенадцать тритов</a></li><li><a href="/ru/learn/decode-in-one-step/">27 Раскодировать за один шаг</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/gf32-a-single/index.html b/apps/website/public/ru/learn/gf32-a-single/index.html
index 4b49ea1c51..4375ab39ef 100644
--- a/apps/website/public/ru/learn/gf32-a-single/index.html
+++ b/apps/website/public/ru/learn/gf32-a-single/index.html
@@ -560,5 +560,6 @@ <h1>GF32: одинарная точность</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Правило и его числа</h3><p>Одно правило делит каждую ширину, отношение, к которому оно стремится, и числа Люка за тройкой 3.</p><ol><li><a href="/ru/learn/a-float-cut-by-phi/">01 Число с плавающей точкой, разрезанное по phi</a></li><li><a href="/ru/learn/phi-as-a-ratio/">02 Phi как отношение</a></li><li><a href="/ru/learn/lucas-numbers-stay-whole/">03 Числа Люка остаются целыми</a></li></ol></li><li><h3>Модуль 2 · Почему phi, почему три</h3><p>Почему деление идёт по phi, почему основание три и как спека проверяет, что GF16 хранит phi.</p><ol><li><a href="/ru/learn/why-the-split-is-phi/">04 Почему деление по phi</a></li><li><a href="/ru/learn/why-base-three/">05 Почему основание три</a></li><li><a href="/ru/learn/phi-in-sixteen-bits/">06 Phi в шестнадцати битах</a></li></ol></li><li><h3>Модуль 3 · Малые ступени: от GF4 до GF8</h3><p>GF4, GF6 и GF8 — меньше всего битов, и округление до целых битов стоит здесь дороже всего.</p><ol><li><a href="/ru/learn/gf4-four-bits/">07 GF4: четыре бита</a></li><li><a href="/ru/learn/gf6-six-bits/">08 GF6: шесть битов</a></li><li><a href="/ru/learn/gf8-one-byte/">09 GF8: один байт</a></li></ol></li><li><h3>Модуль 4 · От десяти до четырнадцати битов</h3><p>GF10, GF12 и GF14 и то, как расстояние до 1 / phi меняется с ростом слова.</p><ol><li><a href="/ru/learn/gf10-ten-bits/">10 GF10: десять битов</a></li><li><a href="/ru/learn/gf12-twelve-bits/">11 GF12: двенадцать битов</a></li><li><a href="/ru/learn/gf14-fourteen-bits/">12 GF14: четырнадцать битов</a></li></ol></li><li><h3>Модуль 5 · GF16 в работе</h3><p>Основной 16-битный формат, скалярное произведение из двух слагаемых в GF-T16, затем GF20 и GF24.</p><ol><li><a href="/ru/learn/gf16-the-primary/">13 GF16: основной формат</a></li><li><a href="/ru/learn/two-products-one-sum/">14 Два произведения, одна сумма</a></li><li><a href="/ru/learn/gf24-twenty-four-bits/">15 GF20 и GF24</a></li></ol></li><li><h3>Модуль 6 · От GF32 до GF64</h3><p>GF32 рядом с IEEE single, GF48 без пары в IEEE, GF64 рядом с IEEE double.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/gf32-a-single/">16 GF32: одинарная точность</a></li><li><a href="/ru/learn/gf48-forty-eight-bits/">17 GF48: сорок восемь битов</a></li><li><a href="/ru/learn/gf64-a-double/">18 GF64: двойная точность</a></li></ol></li><li><h3>Модуль 7 · От GF96 до GF256</h3><p>GF96, GF128 и GF256, где спеки держат раскладку инвариантами.</p><ol><li><a href="/ru/learn/gf96-ninety-six-bits/">19 GF96: девяносто шесть битов</a></li><li><a href="/ru/learn/gf128-a-quad/">20 GF128: четверная точность</a></li><li><a href="/ru/learn/gf256-two-five-six/">21 GF256: двести пятьдесят шесть битов</a></li></ol></li><li><h3>Модуль 8 · Самые широкие ступени, затем триты</h3><p>GF512 и GF1024, две самые широкие ступени, затем GF-T8, где порядок уходит в триты.</p><ol><li><a href="/ru/learn/gf512-five-twelve/">22 GF512: пятьсот двенадцать битов</a></li><li><a href="/ru/learn/gf1024-one-kilobit/">23 GF1024: один килобит</a></li><li><a href="/ru/learn/gft8-exponent-in-trits/">24 GF-T8: порядок в тритах</a></li></ol></li><li><h3>Модуль 9 · Ещё триты, затем декодирование</h3><p>GF-T16 и GF-T32, затем почему фиксированные поля декодируются параллельно, а posit — нет.</p><ol><li><a href="/ru/learn/gft16-six-trits/">25 GF-T16: шесть тритов</a></li><li><a href="/ru/learn/gft32-twelve-trits/">26 GF-T32: двенадцать тритов</a></li><li><a href="/ru/learn/decode-in-one-step/">27 Раскодировать за один шаг</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/gf4-four-bits/index.html b/apps/website/public/ru/learn/gf4-four-bits/index.html
index ed426fb6a0..0be23877e6 100644
--- a/apps/website/public/ru/learn/gf4-four-bits/index.html
+++ b/apps/website/public/ru/learn/gf4-four-bits/index.html
@@ -375,5 +375,6 @@ <h1>GF4: четыре бита</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Правило и его числа</h3><p>Одно правило делит каждую ширину, отношение, к которому оно стремится, и числа Люка за тройкой 3.</p><ol><li><a href="/ru/learn/a-float-cut-by-phi/">01 Число с плавающей точкой, разрезанное по phi</a></li><li><a href="/ru/learn/phi-as-a-ratio/">02 Phi как отношение</a></li><li><a href="/ru/learn/lucas-numbers-stay-whole/">03 Числа Люка остаются целыми</a></li></ol></li><li><h3>Модуль 2 · Почему phi, почему три</h3><p>Почему деление идёт по phi, почему основание три и как спека проверяет, что GF16 хранит phi.</p><ol><li><a href="/ru/learn/why-the-split-is-phi/">04 Почему деление по phi</a></li><li><a href="/ru/learn/why-base-three/">05 Почему основание три</a></li><li><a href="/ru/learn/phi-in-sixteen-bits/">06 Phi в шестнадцати битах</a></li></ol></li><li><h3>Модуль 3 · Малые ступени: от GF4 до GF8</h3><p>GF4, GF6 и GF8 — меньше всего битов, и округление до целых битов стоит здесь дороже всего.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/gf4-four-bits/">07 GF4: четыре бита</a></li><li><a href="/ru/learn/gf6-six-bits/">08 GF6: шесть битов</a></li><li><a href="/ru/learn/gf8-one-byte/">09 GF8: один байт</a></li></ol></li><li><h3>Модуль 4 · От десяти до четырнадцати битов</h3><p>GF10, GF12 и GF14 и то, как расстояние до 1 / phi меняется с ростом слова.</p><ol><li><a href="/ru/learn/gf10-ten-bits/">10 GF10: десять битов</a></li><li><a href="/ru/learn/gf12-twelve-bits/">11 GF12: двенадцать битов</a></li><li><a href="/ru/learn/gf14-fourteen-bits/">12 GF14: четырнадцать битов</a></li></ol></li><li><h3>Модуль 5 · GF16 в работе</h3><p>Основной 16-битный формат, скалярное произведение из двух слагаемых в GF-T16, затем GF20 и GF24.</p><ol><li><a href="/ru/learn/gf16-the-primary/">13 GF16: основной формат</a></li><li><a href="/ru/learn/two-products-one-sum/">14 Два произведения, одна сумма</a></li><li><a href="/ru/learn/gf24-twenty-four-bits/">15 GF20 и GF24</a></li></ol></li><li><h3>Модуль 6 · От GF32 до GF64</h3><p>GF32 рядом с IEEE single, GF48 без пары в IEEE, GF64 рядом с IEEE double.</p><ol><li><a href="/ru/learn/gf32-a-single/">16 GF32: одинарная точность</a></li><li><a href="/ru/learn/gf48-forty-eight-bits/">17 GF48: сорок восемь битов</a></li><li><a href="/ru/learn/gf64-a-double/">18 GF64: двойная точность</a></li></ol></li><li><h3>Модуль 7 · От GF96 до GF256</h3><p>GF96, GF128 и GF256, где спеки держат раскладку инвариантами.</p><ol><li><a href="/ru/learn/gf96-ninety-six-bits/">19 GF96: девяносто шесть битов</a></li><li><a href="/ru/learn/gf128-a-quad/">20 GF128: четверная точность</a></li><li><a href="/ru/learn/gf256-two-five-six/">21 GF256: двести пятьдесят шесть битов</a></li></ol></li><li><h3>Модуль 8 · Самые широкие ступени, затем триты</h3><p>GF512 и GF1024, две самые широкие ступени, затем GF-T8, где порядок уходит в триты.</p><ol><li><a href="/ru/learn/gf512-five-twelve/">22 GF512: пятьсот двенадцать битов</a></li><li><a href="/ru/learn/gf1024-one-kilobit/">23 GF1024: один килобит</a></li><li><a href="/ru/learn/gft8-exponent-in-trits/">24 GF-T8: порядок в тритах</a></li></ol></li><li><h3>Модуль 9 · Ещё триты, затем декодирование</h3><p>GF-T16 и GF-T32, затем почему фиксированные поля декодируются параллельно, а posit — нет.</p><ol><li><a href="/ru/learn/gft16-six-trits/">25 GF-T16: шесть тритов</a></li><li><a href="/ru/learn/gft32-twelve-trits/">26 GF-T32: двенадцать тритов</a></li><li><a href="/ru/learn/decode-in-one-step/">27 Раскодировать за один шаг</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/gf48-forty-eight-bits/index.html b/apps/website/public/ru/learn/gf48-forty-eight-bits/index.html
index a26bc63952..476af4d056 100644
--- a/apps/website/public/ru/learn/gf48-forty-eight-bits/index.html
+++ b/apps/website/public/ru/learn/gf48-forty-eight-bits/index.html
@@ -154,5 +154,6 @@ <h1>GF48: сорок восемь битов</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Правило и его числа</h3><p>Одно правило делит каждую ширину, отношение, к которому оно стремится, и числа Люка за тройкой 3.</p><ol><li><a href="/ru/learn/a-float-cut-by-phi/">01 Число с плавающей точкой, разрезанное по phi</a></li><li><a href="/ru/learn/phi-as-a-ratio/">02 Phi как отношение</a></li><li><a href="/ru/learn/lucas-numbers-stay-whole/">03 Числа Люка остаются целыми</a></li></ol></li><li><h3>Модуль 2 · Почему phi, почему три</h3><p>Почему деление идёт по phi, почему основание три и как спека проверяет, что GF16 хранит phi.</p><ol><li><a href="/ru/learn/why-the-split-is-phi/">04 Почему деление по phi</a></li><li><a href="/ru/learn/why-base-three/">05 Почему основание три</a></li><li><a href="/ru/learn/phi-in-sixteen-bits/">06 Phi в шестнадцати битах</a></li></ol></li><li><h3>Модуль 3 · Малые ступени: от GF4 до GF8</h3><p>GF4, GF6 и GF8 — меньше всего битов, и округление до целых битов стоит здесь дороже всего.</p><ol><li><a href="/ru/learn/gf4-four-bits/">07 GF4: четыре бита</a></li><li><a href="/ru/learn/gf6-six-bits/">08 GF6: шесть битов</a></li><li><a href="/ru/learn/gf8-one-byte/">09 GF8: один байт</a></li></ol></li><li><h3>Модуль 4 · От десяти до четырнадцати битов</h3><p>GF10, GF12 и GF14 и то, как расстояние до 1 / phi меняется с ростом слова.</p><ol><li><a href="/ru/learn/gf10-ten-bits/">10 GF10: десять битов</a></li><li><a href="/ru/learn/gf12-twelve-bits/">11 GF12: двенадцать битов</a></li><li><a href="/ru/learn/gf14-fourteen-bits/">12 GF14: четырнадцать битов</a></li></ol></li><li><h3>Модуль 5 · GF16 в работе</h3><p>Основной 16-битный формат, скалярное произведение из двух слагаемых в GF-T16, затем GF20 и GF24.</p><ol><li><a href="/ru/learn/gf16-the-primary/">13 GF16: основной формат</a></li><li><a href="/ru/learn/two-products-one-sum/">14 Два произведения, одна сумма</a></li><li><a href="/ru/learn/gf24-twenty-four-bits/">15 GF20 и GF24</a></li></ol></li><li><h3>Модуль 6 · От GF32 до GF64</h3><p>GF32 рядом с IEEE single, GF48 без пары в IEEE, GF64 рядом с IEEE double.</p><ol><li><a href="/ru/learn/gf32-a-single/">16 GF32: одинарная точность</a></li><li class="on" aria-current="page"><a href="/ru/learn/gf48-forty-eight-bits/">17 GF48: сорок восемь битов</a></li><li><a href="/ru/learn/gf64-a-double/">18 GF64: двойная точность</a></li></ol></li><li><h3>Модуль 7 · От GF96 до GF256</h3><p>GF96, GF128 и GF256, где спеки держат раскладку инвариантами.</p><ol><li><a href="/ru/learn/gf96-ninety-six-bits/">19 GF96: девяносто шесть битов</a></li><li><a href="/ru/learn/gf128-a-quad/">20 GF128: четверная точность</a></li><li><a href="/ru/learn/gf256-two-five-six/">21 GF256: двести пятьдесят шесть битов</a></li></ol></li><li><h3>Модуль 8 · Самые широкие ступени, затем триты</h3><p>GF512 и GF1024, две самые широкие ступени, затем GF-T8, где порядок уходит в триты.</p><ol><li><a href="/ru/learn/gf512-five-twelve/">22 GF512: пятьсот двенадцать битов</a></li><li><a href="/ru/learn/gf1024-one-kilobit/">23 GF1024: один килобит</a></li><li><a href="/ru/learn/gft8-exponent-in-trits/">24 GF-T8: порядок в тритах</a></li></ol></li><li><h3>Модуль 9 · Ещё триты, затем декодирование</h3><p>GF-T16 и GF-T32, затем почему фиксированные поля декодируются параллельно, а posit — нет.</p><ol><li><a href="/ru/learn/gft16-six-trits/">25 GF-T16: шесть тритов</a></li><li><a href="/ru/learn/gft32-twelve-trits/">26 GF-T32: двенадцать тритов</a></li><li><a href="/ru/learn/decode-in-one-step/">27 Раскодировать за один шаг</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/gf512-five-twelve/index.html b/apps/website/public/ru/learn/gf512-five-twelve/index.html
index f4cb779385..57238e3adb 100644
--- a/apps/website/public/ru/learn/gf512-five-twelve/index.html
+++ b/apps/website/public/ru/learn/gf512-five-twelve/index.html
@@ -147,5 +147,6 @@ <h1>GF512: пятьсот двенадцать битов</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Правило и его числа</h3><p>Одно правило делит каждую ширину, отношение, к которому оно стремится, и числа Люка за тройкой 3.</p><ol><li><a href="/ru/learn/a-float-cut-by-phi/">01 Число с плавающей точкой, разрезанное по phi</a></li><li><a href="/ru/learn/phi-as-a-ratio/">02 Phi как отношение</a></li><li><a href="/ru/learn/lucas-numbers-stay-whole/">03 Числа Люка остаются целыми</a></li></ol></li><li><h3>Модуль 2 · Почему phi, почему три</h3><p>Почему деление идёт по phi, почему основание три и как спека проверяет, что GF16 хранит phi.</p><ol><li><a href="/ru/learn/why-the-split-is-phi/">04 Почему деление по phi</a></li><li><a href="/ru/learn/why-base-three/">05 Почему основание три</a></li><li><a href="/ru/learn/phi-in-sixteen-bits/">06 Phi в шестнадцати битах</a></li></ol></li><li><h3>Модуль 3 · Малые ступени: от GF4 до GF8</h3><p>GF4, GF6 и GF8 — меньше всего битов, и округление до целых битов стоит здесь дороже всего.</p><ol><li><a href="/ru/learn/gf4-four-bits/">07 GF4: четыре бита</a></li><li><a href="/ru/learn/gf6-six-bits/">08 GF6: шесть битов</a></li><li><a href="/ru/learn/gf8-one-byte/">09 GF8: один байт</a></li></ol></li><li><h3>Модуль 4 · От десяти до четырнадцати битов</h3><p>GF10, GF12 и GF14 и то, как расстояние до 1 / phi меняется с ростом слова.</p><ol><li><a href="/ru/learn/gf10-ten-bits/">10 GF10: десять битов</a></li><li><a href="/ru/learn/gf12-twelve-bits/">11 GF12: двенадцать битов</a></li><li><a href="/ru/learn/gf14-fourteen-bits/">12 GF14: четырнадцать битов</a></li></ol></li><li><h3>Модуль 5 · GF16 в работе</h3><p>Основной 16-битный формат, скалярное произведение из двух слагаемых в GF-T16, затем GF20 и GF24.</p><ol><li><a href="/ru/learn/gf16-the-primary/">13 GF16: основной формат</a></li><li><a href="/ru/learn/two-products-one-sum/">14 Два произведения, одна сумма</a></li><li><a href="/ru/learn/gf24-twenty-four-bits/">15 GF20 и GF24</a></li></ol></li><li><h3>Модуль 6 · От GF32 до GF64</h3><p>GF32 рядом с IEEE single, GF48 без пары в IEEE, GF64 рядом с IEEE double.</p><ol><li><a href="/ru/learn/gf32-a-single/">16 GF32: одинарная точность</a></li><li><a href="/ru/learn/gf48-forty-eight-bits/">17 GF48: сорок восемь битов</a></li><li><a href="/ru/learn/gf64-a-double/">18 GF64: двойная точность</a></li></ol></li><li><h3>Модуль 7 · От GF96 до GF256</h3><p>GF96, GF128 и GF256, где спеки держат раскладку инвариантами.</p><ol><li><a href="/ru/learn/gf96-ninety-six-bits/">19 GF96: девяносто шесть битов</a></li><li><a href="/ru/learn/gf128-a-quad/">20 GF128: четверная точность</a></li><li><a href="/ru/learn/gf256-two-five-six/">21 GF256: двести пятьдесят шесть битов</a></li></ol></li><li><h3>Модуль 8 · Самые широкие ступени, затем триты</h3><p>GF512 и GF1024, две самые широкие ступени, затем GF-T8, где порядок уходит в триты.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/gf512-five-twelve/">22 GF512: пятьсот двенадцать битов</a></li><li><a href="/ru/learn/gf1024-one-kilobit/">23 GF1024: один килобит</a></li><li><a href="/ru/learn/gft8-exponent-in-trits/">24 GF-T8: порядок в тритах</a></li></ol></li><li><h3>Модуль 9 · Ещё триты, затем декодирование</h3><p>GF-T16 и GF-T32, затем почему фиксированные поля декодируются параллельно, а posit — нет.</p><ol><li><a href="/ru/learn/gft16-six-trits/">25 GF-T16: шесть тритов</a></li><li><a href="/ru/learn/gft32-twelve-trits/">26 GF-T32: двенадцать тритов</a></li><li><a href="/ru/learn/decode-in-one-step/">27 Раскодировать за один шаг</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/gf6-six-bits/index.html b/apps/website/public/ru/learn/gf6-six-bits/index.html
index 1e0faaa540..2b88bc46d1 100644
--- a/apps/website/public/ru/learn/gf6-six-bits/index.html
+++ b/apps/website/public/ru/learn/gf6-six-bits/index.html
@@ -154,5 +154,6 @@ <h1>GF6: шесть битов</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Правило и его числа</h3><p>Одно правило делит каждую ширину, отношение, к которому оно стремится, и числа Люка за тройкой 3.</p><ol><li><a href="/ru/learn/a-float-cut-by-phi/">01 Число с плавающей точкой, разрезанное по phi</a></li><li><a href="/ru/learn/phi-as-a-ratio/">02 Phi как отношение</a></li><li><a href="/ru/learn/lucas-numbers-stay-whole/">03 Числа Люка остаются целыми</a></li></ol></li><li><h3>Модуль 2 · Почему phi, почему три</h3><p>Почему деление идёт по phi, почему основание три и как спека проверяет, что GF16 хранит phi.</p><ol><li><a href="/ru/learn/why-the-split-is-phi/">04 Почему деление по phi</a></li><li><a href="/ru/learn/why-base-three/">05 Почему основание три</a></li><li><a href="/ru/learn/phi-in-sixteen-bits/">06 Phi в шестнадцати битах</a></li></ol></li><li><h3>Модуль 3 · Малые ступени: от GF4 до GF8</h3><p>GF4, GF6 и GF8 — меньше всего битов, и округление до целых битов стоит здесь дороже всего.</p><ol><li><a href="/ru/learn/gf4-four-bits/">07 GF4: четыре бита</a></li><li class="on" aria-current="page"><a href="/ru/learn/gf6-six-bits/">08 GF6: шесть битов</a></li><li><a href="/ru/learn/gf8-one-byte/">09 GF8: один байт</a></li></ol></li><li><h3>Модуль 4 · От десяти до четырнадцати битов</h3><p>GF10, GF12 и GF14 и то, как расстояние до 1 / phi меняется с ростом слова.</p><ol><li><a href="/ru/learn/gf10-ten-bits/">10 GF10: десять битов</a></li><li><a href="/ru/learn/gf12-twelve-bits/">11 GF12: двенадцать битов</a></li><li><a href="/ru/learn/gf14-fourteen-bits/">12 GF14: четырнадцать битов</a></li></ol></li><li><h3>Модуль 5 · GF16 в работе</h3><p>Основной 16-битный формат, скалярное произведение из двух слагаемых в GF-T16, затем GF20 и GF24.</p><ol><li><a href="/ru/learn/gf16-the-primary/">13 GF16: основной формат</a></li><li><a href="/ru/learn/two-products-one-sum/">14 Два произведения, одна сумма</a></li><li><a href="/ru/learn/gf24-twenty-four-bits/">15 GF20 и GF24</a></li></ol></li><li><h3>Модуль 6 · От GF32 до GF64</h3><p>GF32 рядом с IEEE single, GF48 без пары в IEEE, GF64 рядом с IEEE double.</p><ol><li><a href="/ru/learn/gf32-a-single/">16 GF32: одинарная точность</a></li><li><a href="/ru/learn/gf48-forty-eight-bits/">17 GF48: сорок восемь битов</a></li><li><a href="/ru/learn/gf64-a-double/">18 GF64: двойная точность</a></li></ol></li><li><h3>Модуль 7 · От GF96 до GF256</h3><p>GF96, GF128 и GF256, где спеки держат раскладку инвариантами.</p><ol><li><a href="/ru/learn/gf96-ninety-six-bits/">19 GF96: девяносто шесть битов</a></li><li><a href="/ru/learn/gf128-a-quad/">20 GF128: четверная точность</a></li><li><a href="/ru/learn/gf256-two-five-six/">21 GF256: двести пятьдесят шесть битов</a></li></ol></li><li><h3>Модуль 8 · Самые широкие ступени, затем триты</h3><p>GF512 и GF1024, две самые широкие ступени, затем GF-T8, где порядок уходит в триты.</p><ol><li><a href="/ru/learn/gf512-five-twelve/">22 GF512: пятьсот двенадцать битов</a></li><li><a href="/ru/learn/gf1024-one-kilobit/">23 GF1024: один килобит</a></li><li><a href="/ru/learn/gft8-exponent-in-trits/">24 GF-T8: порядок в тритах</a></li></ol></li><li><h3>Модуль 9 · Ещё триты, затем декодирование</h3><p>GF-T16 и GF-T32, затем почему фиксированные поля декодируются параллельно, а posit — нет.</p><ol><li><a href="/ru/learn/gft16-six-trits/">25 GF-T16: шесть тритов</a></li><li><a href="/ru/learn/gft32-twelve-trits/">26 GF-T32: двенадцать тритов</a></li><li><a href="/ru/learn/decode-in-one-step/">27 Раскодировать за один шаг</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/gf64-a-double/index.html b/apps/website/public/ru/learn/gf64-a-double/index.html
index 604046baf4..0e24fa0ad6 100644
--- a/apps/website/public/ru/learn/gf64-a-double/index.html
+++ b/apps/website/public/ru/learn/gf64-a-double/index.html
@@ -298,5 +298,6 @@ <h1>GF64: двойная точность</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Правило и его числа</h3><p>Одно правило делит каждую ширину, отношение, к которому оно стремится, и числа Люка за тройкой 3.</p><ol><li><a href="/ru/learn/a-float-cut-by-phi/">01 Число с плавающей точкой, разрезанное по phi</a></li><li><a href="/ru/learn/phi-as-a-ratio/">02 Phi как отношение</a></li><li><a href="/ru/learn/lucas-numbers-stay-whole/">03 Числа Люка остаются целыми</a></li></ol></li><li><h3>Модуль 2 · Почему phi, почему три</h3><p>Почему деление идёт по phi, почему основание три и как спека проверяет, что GF16 хранит phi.</p><ol><li><a href="/ru/learn/why-the-split-is-phi/">04 Почему деление по phi</a></li><li><a href="/ru/learn/why-base-three/">05 Почему основание три</a></li><li><a href="/ru/learn/phi-in-sixteen-bits/">06 Phi в шестнадцати битах</a></li></ol></li><li><h3>Модуль 3 · Малые ступени: от GF4 до GF8</h3><p>GF4, GF6 и GF8 — меньше всего битов, и округление до целых битов стоит здесь дороже всего.</p><ol><li><a href="/ru/learn/gf4-four-bits/">07 GF4: четыре бита</a></li><li><a href="/ru/learn/gf6-six-bits/">08 GF6: шесть битов</a></li><li><a href="/ru/learn/gf8-one-byte/">09 GF8: один байт</a></li></ol></li><li><h3>Модуль 4 · От десяти до четырнадцати битов</h3><p>GF10, GF12 и GF14 и то, как расстояние до 1 / phi меняется с ростом слова.</p><ol><li><a href="/ru/learn/gf10-ten-bits/">10 GF10: десять битов</a></li><li><a href="/ru/learn/gf12-twelve-bits/">11 GF12: двенадцать битов</a></li><li><a href="/ru/learn/gf14-fourteen-bits/">12 GF14: четырнадцать битов</a></li></ol></li><li><h3>Модуль 5 · GF16 в работе</h3><p>Основной 16-битный формат, скалярное произведение из двух слагаемых в GF-T16, затем GF20 и GF24.</p><ol><li><a href="/ru/learn/gf16-the-primary/">13 GF16: основной формат</a></li><li><a href="/ru/learn/two-products-one-sum/">14 Два произведения, одна сумма</a></li><li><a href="/ru/learn/gf24-twenty-four-bits/">15 GF20 и GF24</a></li></ol></li><li><h3>Модуль 6 · От GF32 до GF64</h3><p>GF32 рядом с IEEE single, GF48 без пары в IEEE, GF64 рядом с IEEE double.</p><ol><li><a href="/ru/learn/gf32-a-single/">16 GF32: одинарная точность</a></li><li><a href="/ru/learn/gf48-forty-eight-bits/">17 GF48: сорок восемь битов</a></li><li class="on" aria-current="page"><a href="/ru/learn/gf64-a-double/">18 GF64: двойная точность</a></li></ol></li><li><h3>Модуль 7 · От GF96 до GF256</h3><p>GF96, GF128 и GF256, где спеки держат раскладку инвариантами.</p><ol><li><a href="/ru/learn/gf96-ninety-six-bits/">19 GF96: девяносто шесть битов</a></li><li><a href="/ru/learn/gf128-a-quad/">20 GF128: четверная точность</a></li><li><a href="/ru/learn/gf256-two-five-six/">21 GF256: двести пятьдесят шесть битов</a></li></ol></li><li><h3>Модуль 8 · Самые широкие ступени, затем триты</h3><p>GF512 и GF1024, две самые широкие ступени, затем GF-T8, где порядок уходит в триты.</p><ol><li><a href="/ru/learn/gf512-five-twelve/">22 GF512: пятьсот двенадцать битов</a></li><li><a href="/ru/learn/gf1024-one-kilobit/">23 GF1024: один килобит</a></li><li><a href="/ru/learn/gft8-exponent-in-trits/">24 GF-T8: порядок в тритах</a></li></ol></li><li><h3>Модуль 9 · Ещё триты, затем декодирование</h3><p>GF-T16 и GF-T32, затем почему фиксированные поля декодируются параллельно, а posit — нет.</p><ol><li><a href="/ru/learn/gft16-six-trits/">25 GF-T16: шесть тритов</a></li><li><a href="/ru/learn/gft32-twelve-trits/">26 GF-T32: двенадцать тритов</a></li><li><a href="/ru/learn/decode-in-one-step/">27 Раскодировать за один шаг</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/gf8-one-byte/index.html b/apps/website/public/ru/learn/gf8-one-byte/index.html
index c5d5c67e20..47897c15cd 100644
--- a/apps/website/public/ru/learn/gf8-one-byte/index.html
+++ b/apps/website/public/ru/learn/gf8-one-byte/index.html
@@ -598,5 +598,6 @@ <h1>GF8: один байт</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Правило и его числа</h3><p>Одно правило делит каждую ширину, отношение, к которому оно стремится, и числа Люка за тройкой 3.</p><ol><li><a href="/ru/learn/a-float-cut-by-phi/">01 Число с плавающей точкой, разрезанное по phi</a></li><li><a href="/ru/learn/phi-as-a-ratio/">02 Phi как отношение</a></li><li><a href="/ru/learn/lucas-numbers-stay-whole/">03 Числа Люка остаются целыми</a></li></ol></li><li><h3>Модуль 2 · Почему phi, почему три</h3><p>Почему деление идёт по phi, почему основание три и как спека проверяет, что GF16 хранит phi.</p><ol><li><a href="/ru/learn/why-the-split-is-phi/">04 Почему деление по phi</a></li><li><a href="/ru/learn/why-base-three/">05 Почему основание три</a></li><li><a href="/ru/learn/phi-in-sixteen-bits/">06 Phi в шестнадцати битах</a></li></ol></li><li><h3>Модуль 3 · Малые ступени: от GF4 до GF8</h3><p>GF4, GF6 и GF8 — меньше всего битов, и округление до целых битов стоит здесь дороже всего.</p><ol><li><a href="/ru/learn/gf4-four-bits/">07 GF4: четыре бита</a></li><li><a href="/ru/learn/gf6-six-bits/">08 GF6: шесть битов</a></li><li class="on" aria-current="page"><a href="/ru/learn/gf8-one-byte/">09 GF8: один байт</a></li></ol></li><li><h3>Модуль 4 · От десяти до четырнадцати битов</h3><p>GF10, GF12 и GF14 и то, как расстояние до 1 / phi меняется с ростом слова.</p><ol><li><a href="/ru/learn/gf10-ten-bits/">10 GF10: десять битов</a></li><li><a href="/ru/learn/gf12-twelve-bits/">11 GF12: двенадцать битов</a></li><li><a href="/ru/learn/gf14-fourteen-bits/">12 GF14: четырнадцать битов</a></li></ol></li><li><h3>Модуль 5 · GF16 в работе</h3><p>Основной 16-битный формат, скалярное произведение из двух слагаемых в GF-T16, затем GF20 и GF24.</p><ol><li><a href="/ru/learn/gf16-the-primary/">13 GF16: основной формат</a></li><li><a href="/ru/learn/two-products-one-sum/">14 Два произведения, одна сумма</a></li><li><a href="/ru/learn/gf24-twenty-four-bits/">15 GF20 и GF24</a></li></ol></li><li><h3>Модуль 6 · От GF32 до GF64</h3><p>GF32 рядом с IEEE single, GF48 без пары в IEEE, GF64 рядом с IEEE double.</p><ol><li><a href="/ru/learn/gf32-a-single/">16 GF32: одинарная точность</a></li><li><a href="/ru/learn/gf48-forty-eight-bits/">17 GF48: сорок восемь битов</a></li><li><a href="/ru/learn/gf64-a-double/">18 GF64: двойная точность</a></li></ol></li><li><h3>Модуль 7 · От GF96 до GF256</h3><p>GF96, GF128 и GF256, где спеки держат раскладку инвариантами.</p><ol><li><a href="/ru/learn/gf96-ninety-six-bits/">19 GF96: девяносто шесть битов</a></li><li><a href="/ru/learn/gf128-a-quad/">20 GF128: четверная точность</a></li><li><a href="/ru/learn/gf256-two-five-six/">21 GF256: двести пятьдесят шесть битов</a></li></ol></li><li><h3>Модуль 8 · Самые широкие ступени, затем триты</h3><p>GF512 и GF1024, две самые широкие ступени, затем GF-T8, где порядок уходит в триты.</p><ol><li><a href="/ru/learn/gf512-five-twelve/">22 GF512: пятьсот двенадцать битов</a></li><li><a href="/ru/learn/gf1024-one-kilobit/">23 GF1024: один килобит</a></li><li><a href="/ru/learn/gft8-exponent-in-trits/">24 GF-T8: порядок в тритах</a></li></ol></li><li><h3>Модуль 9 · Ещё триты, затем декодирование</h3><p>GF-T16 и GF-T32, затем почему фиксированные поля декодируются параллельно, а posit — нет.</p><ol><li><a href="/ru/learn/gft16-six-trits/">25 GF-T16: шесть тритов</a></li><li><a href="/ru/learn/gft32-twelve-trits/">26 GF-T32: двенадцать тритов</a></li><li><a href="/ru/learn/decode-in-one-step/">27 Раскодировать за один шаг</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/gf96-ninety-six-bits/index.html b/apps/website/public/ru/learn/gf96-ninety-six-bits/index.html
index 2e179e8729..c0d686b325 100644
--- a/apps/website/public/ru/learn/gf96-ninety-six-bits/index.html
+++ b/apps/website/public/ru/learn/gf96-ninety-six-bits/index.html
@@ -146,5 +146,6 @@ <h1>GF96: девяносто шесть битов</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Правило и его числа</h3><p>Одно правило делит каждую ширину, отношение, к которому оно стремится, и числа Люка за тройкой 3.</p><ol><li><a href="/ru/learn/a-float-cut-by-phi/">01 Число с плавающей точкой, разрезанное по phi</a></li><li><a href="/ru/learn/phi-as-a-ratio/">02 Phi как отношение</a></li><li><a href="/ru/learn/lucas-numbers-stay-whole/">03 Числа Люка остаются целыми</a></li></ol></li><li><h3>Модуль 2 · Почему phi, почему три</h3><p>Почему деление идёт по phi, почему основание три и как спека проверяет, что GF16 хранит phi.</p><ol><li><a href="/ru/learn/why-the-split-is-phi/">04 Почему деление по phi</a></li><li><a href="/ru/learn/why-base-three/">05 Почему основание три</a></li><li><a href="/ru/learn/phi-in-sixteen-bits/">06 Phi в шестнадцати битах</a></li></ol></li><li><h3>Модуль 3 · Малые ступени: от GF4 до GF8</h3><p>GF4, GF6 и GF8 — меньше всего битов, и округление до целых битов стоит здесь дороже всего.</p><ol><li><a href="/ru/learn/gf4-four-bits/">07 GF4: четыре бита</a></li><li><a href="/ru/learn/gf6-six-bits/">08 GF6: шесть битов</a></li><li><a href="/ru/learn/gf8-one-byte/">09 GF8: один байт</a></li></ol></li><li><h3>Модуль 4 · От десяти до четырнадцати битов</h3><p>GF10, GF12 и GF14 и то, как расстояние до 1 / phi меняется с ростом слова.</p><ol><li><a href="/ru/learn/gf10-ten-bits/">10 GF10: десять битов</a></li><li><a href="/ru/learn/gf12-twelve-bits/">11 GF12: двенадцать битов</a></li><li><a href="/ru/learn/gf14-fourteen-bits/">12 GF14: четырнадцать битов</a></li></ol></li><li><h3>Модуль 5 · GF16 в работе</h3><p>Основной 16-битный формат, скалярное произведение из двух слагаемых в GF-T16, затем GF20 и GF24.</p><ol><li><a href="/ru/learn/gf16-the-primary/">13 GF16: основной формат</a></li><li><a href="/ru/learn/two-products-one-sum/">14 Два произведения, одна сумма</a></li><li><a href="/ru/learn/gf24-twenty-four-bits/">15 GF20 и GF24</a></li></ol></li><li><h3>Модуль 6 · От GF32 до GF64</h3><p>GF32 рядом с IEEE single, GF48 без пары в IEEE, GF64 рядом с IEEE double.</p><ol><li><a href="/ru/learn/gf32-a-single/">16 GF32: одинарная точность</a></li><li><a href="/ru/learn/gf48-forty-eight-bits/">17 GF48: сорок восемь битов</a></li><li><a href="/ru/learn/gf64-a-double/">18 GF64: двойная точность</a></li></ol></li><li><h3>Модуль 7 · От GF96 до GF256</h3><p>GF96, GF128 и GF256, где спеки держат раскладку инвариантами.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/gf96-ninety-six-bits/">19 GF96: девяносто шесть битов</a></li><li><a href="/ru/learn/gf128-a-quad/">20 GF128: четверная точность</a></li><li><a href="/ru/learn/gf256-two-five-six/">21 GF256: двести пятьдесят шесть битов</a></li></ol></li><li><h3>Модуль 8 · Самые широкие ступени, затем триты</h3><p>GF512 и GF1024, две самые широкие ступени, затем GF-T8, где порядок уходит в триты.</p><ol><li><a href="/ru/learn/gf512-five-twelve/">22 GF512: пятьсот двенадцать битов</a></li><li><a href="/ru/learn/gf1024-one-kilobit/">23 GF1024: один килобит</a></li><li><a href="/ru/learn/gft8-exponent-in-trits/">24 GF-T8: порядок в тритах</a></li></ol></li><li><h3>Модуль 9 · Ещё триты, затем декодирование</h3><p>GF-T16 и GF-T32, затем почему фиксированные поля декодируются параллельно, а posit — нет.</p><ol><li><a href="/ru/learn/gft16-six-trits/">25 GF-T16: шесть тритов</a></li><li><a href="/ru/learn/gft32-twelve-trits/">26 GF-T32: двенадцать тритов</a></li><li><a href="/ru/learn/decode-in-one-step/">27 Раскодировать за один шаг</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/gft16-six-trits/index.html b/apps/website/public/ru/learn/gft16-six-trits/index.html
index da1599aeb3..167bfbfe9c 100644
--- a/apps/website/public/ru/learn/gft16-six-trits/index.html
+++ b/apps/website/public/ru/learn/gft16-six-trits/index.html
@@ -125,5 +125,6 @@ <h1>GF-T16: шесть тритов</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Правило и его числа</h3><p>Одно правило делит каждую ширину, отношение, к которому оно стремится, и числа Люка за тройкой 3.</p><ol><li><a href="/ru/learn/a-float-cut-by-phi/">01 Число с плавающей точкой, разрезанное по phi</a></li><li><a href="/ru/learn/phi-as-a-ratio/">02 Phi как отношение</a></li><li><a href="/ru/learn/lucas-numbers-stay-whole/">03 Числа Люка остаются целыми</a></li></ol></li><li><h3>Модуль 2 · Почему phi, почему три</h3><p>Почему деление идёт по phi, почему основание три и как спека проверяет, что GF16 хранит phi.</p><ol><li><a href="/ru/learn/why-the-split-is-phi/">04 Почему деление по phi</a></li><li><a href="/ru/learn/why-base-three/">05 Почему основание три</a></li><li><a href="/ru/learn/phi-in-sixteen-bits/">06 Phi в шестнадцати битах</a></li></ol></li><li><h3>Модуль 3 · Малые ступени: от GF4 до GF8</h3><p>GF4, GF6 и GF8 — меньше всего битов, и округление до целых битов стоит здесь дороже всего.</p><ol><li><a href="/ru/learn/gf4-four-bits/">07 GF4: четыре бита</a></li><li><a href="/ru/learn/gf6-six-bits/">08 GF6: шесть битов</a></li><li><a href="/ru/learn/gf8-one-byte/">09 GF8: один байт</a></li></ol></li><li><h3>Модуль 4 · От десяти до четырнадцати битов</h3><p>GF10, GF12 и GF14 и то, как расстояние до 1 / phi меняется с ростом слова.</p><ol><li><a href="/ru/learn/gf10-ten-bits/">10 GF10: десять битов</a></li><li><a href="/ru/learn/gf12-twelve-bits/">11 GF12: двенадцать битов</a></li><li><a href="/ru/learn/gf14-fourteen-bits/">12 GF14: четырнадцать битов</a></li></ol></li><li><h3>Модуль 5 · GF16 в работе</h3><p>Основной 16-битный формат, скалярное произведение из двух слагаемых в GF-T16, затем GF20 и GF24.</p><ol><li><a href="/ru/learn/gf16-the-primary/">13 GF16: основной формат</a></li><li><a href="/ru/learn/two-products-one-sum/">14 Два произведения, одна сумма</a></li><li><a href="/ru/learn/gf24-twenty-four-bits/">15 GF20 и GF24</a></li></ol></li><li><h3>Модуль 6 · От GF32 до GF64</h3><p>GF32 рядом с IEEE single, GF48 без пары в IEEE, GF64 рядом с IEEE double.</p><ol><li><a href="/ru/learn/gf32-a-single/">16 GF32: одинарная точность</a></li><li><a href="/ru/learn/gf48-forty-eight-bits/">17 GF48: сорок восемь битов</a></li><li><a href="/ru/learn/gf64-a-double/">18 GF64: двойная точность</a></li></ol></li><li><h3>Модуль 7 · От GF96 до GF256</h3><p>GF96, GF128 и GF256, где спеки держат раскладку инвариантами.</p><ol><li><a href="/ru/learn/gf96-ninety-six-bits/">19 GF96: девяносто шесть битов</a></li><li><a href="/ru/learn/gf128-a-quad/">20 GF128: четверная точность</a></li><li><a href="/ru/learn/gf256-two-five-six/">21 GF256: двести пятьдесят шесть битов</a></li></ol></li><li><h3>Модуль 8 · Самые широкие ступени, затем триты</h3><p>GF512 и GF1024, две самые широкие ступени, затем GF-T8, где порядок уходит в триты.</p><ol><li><a href="/ru/learn/gf512-five-twelve/">22 GF512: пятьсот двенадцать битов</a></li><li><a href="/ru/learn/gf1024-one-kilobit/">23 GF1024: один килобит</a></li><li><a href="/ru/learn/gft8-exponent-in-trits/">24 GF-T8: порядок в тритах</a></li></ol></li><li><h3>Модуль 9 · Ещё триты, затем декодирование</h3><p>GF-T16 и GF-T32, затем почему фиксированные поля декодируются параллельно, а posit — нет.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/gft16-six-trits/">25 GF-T16: шесть тритов</a></li><li><a href="/ru/learn/gft32-twelve-trits/">26 GF-T32: двенадцать тритов</a></li><li><a href="/ru/learn/decode-in-one-step/">27 Раскодировать за один шаг</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/gft32-twelve-trits/index.html b/apps/website/public/ru/learn/gft32-twelve-trits/index.html
index 5b92b8e82d..42134717c5 100644
--- a/apps/website/public/ru/learn/gft32-twelve-trits/index.html
+++ b/apps/website/public/ru/learn/gft32-twelve-trits/index.html
@@ -125,5 +125,6 @@ <h1>GF-T32: двенадцать тритов</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Правило и его числа</h3><p>Одно правило делит каждую ширину, отношение, к которому оно стремится, и числа Люка за тройкой 3.</p><ol><li><a href="/ru/learn/a-float-cut-by-phi/">01 Число с плавающей точкой, разрезанное по phi</a></li><li><a href="/ru/learn/phi-as-a-ratio/">02 Phi как отношение</a></li><li><a href="/ru/learn/lucas-numbers-stay-whole/">03 Числа Люка остаются целыми</a></li></ol></li><li><h3>Модуль 2 · Почему phi, почему три</h3><p>Почему деление идёт по phi, почему основание три и как спека проверяет, что GF16 хранит phi.</p><ol><li><a href="/ru/learn/why-the-split-is-phi/">04 Почему деление по phi</a></li><li><a href="/ru/learn/why-base-three/">05 Почему основание три</a></li><li><a href="/ru/learn/phi-in-sixteen-bits/">06 Phi в шестнадцати битах</a></li></ol></li><li><h3>Модуль 3 · Малые ступени: от GF4 до GF8</h3><p>GF4, GF6 и GF8 — меньше всего битов, и округление до целых битов стоит здесь дороже всего.</p><ol><li><a href="/ru/learn/gf4-four-bits/">07 GF4: четыре бита</a></li><li><a href="/ru/learn/gf6-six-bits/">08 GF6: шесть битов</a></li><li><a href="/ru/learn/gf8-one-byte/">09 GF8: один байт</a></li></ol></li><li><h3>Модуль 4 · От десяти до четырнадцати битов</h3><p>GF10, GF12 и GF14 и то, как расстояние до 1 / phi меняется с ростом слова.</p><ol><li><a href="/ru/learn/gf10-ten-bits/">10 GF10: десять битов</a></li><li><a href="/ru/learn/gf12-twelve-bits/">11 GF12: двенадцать битов</a></li><li><a href="/ru/learn/gf14-fourteen-bits/">12 GF14: четырнадцать битов</a></li></ol></li><li><h3>Модуль 5 · GF16 в работе</h3><p>Основной 16-битный формат, скалярное произведение из двух слагаемых в GF-T16, затем GF20 и GF24.</p><ol><li><a href="/ru/learn/gf16-the-primary/">13 GF16: основной формат</a></li><li><a href="/ru/learn/two-products-one-sum/">14 Два произведения, одна сумма</a></li><li><a href="/ru/learn/gf24-twenty-four-bits/">15 GF20 и GF24</a></li></ol></li><li><h3>Модуль 6 · От GF32 до GF64</h3><p>GF32 рядом с IEEE single, GF48 без пары в IEEE, GF64 рядом с IEEE double.</p><ol><li><a href="/ru/learn/gf32-a-single/">16 GF32: одинарная точность</a></li><li><a href="/ru/learn/gf48-forty-eight-bits/">17 GF48: сорок восемь битов</a></li><li><a href="/ru/learn/gf64-a-double/">18 GF64: двойная точность</a></li></ol></li><li><h3>Модуль 7 · От GF96 до GF256</h3><p>GF96, GF128 и GF256, где спеки держат раскладку инвариантами.</p><ol><li><a href="/ru/learn/gf96-ninety-six-bits/">19 GF96: девяносто шесть битов</a></li><li><a href="/ru/learn/gf128-a-quad/">20 GF128: четверная точность</a></li><li><a href="/ru/learn/gf256-two-five-six/">21 GF256: двести пятьдесят шесть битов</a></li></ol></li><li><h3>Модуль 8 · Самые широкие ступени, затем триты</h3><p>GF512 и GF1024, две самые широкие ступени, затем GF-T8, где порядок уходит в триты.</p><ol><li><a href="/ru/learn/gf512-five-twelve/">22 GF512: пятьсот двенадцать битов</a></li><li><a href="/ru/learn/gf1024-one-kilobit/">23 GF1024: один килобит</a></li><li><a href="/ru/learn/gft8-exponent-in-trits/">24 GF-T8: порядок в тритах</a></li></ol></li><li><h3>Модуль 9 · Ещё триты, затем декодирование</h3><p>GF-T16 и GF-T32, затем почему фиксированные поля декодируются параллельно, а posit — нет.</p><ol><li><a href="/ru/learn/gft16-six-trits/">25 GF-T16: шесть тритов</a></li><li class="on" aria-current="page"><a href="/ru/learn/gft32-twelve-trits/">26 GF-T32: двенадцать тритов</a></li><li><a href="/ru/learn/decode-in-one-step/">27 Раскодировать за один шаг</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/gft8-exponent-in-trits/index.html b/apps/website/public/ru/learn/gft8-exponent-in-trits/index.html
index 18f8509c02..bd3945a2fe 100644
--- a/apps/website/public/ru/learn/gft8-exponent-in-trits/index.html
+++ b/apps/website/public/ru/learn/gft8-exponent-in-trits/index.html
@@ -125,5 +125,6 @@ <h1>GF-T8: порядок в тритах</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Правило и его числа</h3><p>Одно правило делит каждую ширину, отношение, к которому оно стремится, и числа Люка за тройкой 3.</p><ol><li><a href="/ru/learn/a-float-cut-by-phi/">01 Число с плавающей точкой, разрезанное по phi</a></li><li><a href="/ru/learn/phi-as-a-ratio/">02 Phi как отношение</a></li><li><a href="/ru/learn/lucas-numbers-stay-whole/">03 Числа Люка остаются целыми</a></li></ol></li><li><h3>Модуль 2 · Почему phi, почему три</h3><p>Почему деление идёт по phi, почему основание три и как спека проверяет, что GF16 хранит phi.</p><ol><li><a href="/ru/learn/why-the-split-is-phi/">04 Почему деление по phi</a></li><li><a href="/ru/learn/why-base-three/">05 Почему основание три</a></li><li><a href="/ru/learn/phi-in-sixteen-bits/">06 Phi в шестнадцати битах</a></li></ol></li><li><h3>Модуль 3 · Малые ступени: от GF4 до GF8</h3><p>GF4, GF6 и GF8 — меньше всего битов, и округление до целых битов стоит здесь дороже всего.</p><ol><li><a href="/ru/learn/gf4-four-bits/">07 GF4: четыре бита</a></li><li><a href="/ru/learn/gf6-six-bits/">08 GF6: шесть битов</a></li><li><a href="/ru/learn/gf8-one-byte/">09 GF8: один байт</a></li></ol></li><li><h3>Модуль 4 · От десяти до четырнадцати битов</h3><p>GF10, GF12 и GF14 и то, как расстояние до 1 / phi меняется с ростом слова.</p><ol><li><a href="/ru/learn/gf10-ten-bits/">10 GF10: десять битов</a></li><li><a href="/ru/learn/gf12-twelve-bits/">11 GF12: двенадцать битов</a></li><li><a href="/ru/learn/gf14-fourteen-bits/">12 GF14: четырнадцать битов</a></li></ol></li><li><h3>Модуль 5 · GF16 в работе</h3><p>Основной 16-битный формат, скалярное произведение из двух слагаемых в GF-T16, затем GF20 и GF24.</p><ol><li><a href="/ru/learn/gf16-the-primary/">13 GF16: основной формат</a></li><li><a href="/ru/learn/two-products-one-sum/">14 Два произведения, одна сумма</a></li><li><a href="/ru/learn/gf24-twenty-four-bits/">15 GF20 и GF24</a></li></ol></li><li><h3>Модуль 6 · От GF32 до GF64</h3><p>GF32 рядом с IEEE single, GF48 без пары в IEEE, GF64 рядом с IEEE double.</p><ol><li><a href="/ru/learn/gf32-a-single/">16 GF32: одинарная точность</a></li><li><a href="/ru/learn/gf48-forty-eight-bits/">17 GF48: сорок восемь битов</a></li><li><a href="/ru/learn/gf64-a-double/">18 GF64: двойная точность</a></li></ol></li><li><h3>Модуль 7 · От GF96 до GF256</h3><p>GF96, GF128 и GF256, где спеки держат раскладку инвариантами.</p><ol><li><a href="/ru/learn/gf96-ninety-six-bits/">19 GF96: девяносто шесть битов</a></li><li><a href="/ru/learn/gf128-a-quad/">20 GF128: четверная точность</a></li><li><a href="/ru/learn/gf256-two-five-six/">21 GF256: двести пятьдесят шесть битов</a></li></ol></li><li><h3>Модуль 8 · Самые широкие ступени, затем триты</h3><p>GF512 и GF1024, две самые широкие ступени, затем GF-T8, где порядок уходит в триты.</p><ol><li><a href="/ru/learn/gf512-five-twelve/">22 GF512: пятьсот двенадцать битов</a></li><li><a href="/ru/learn/gf1024-one-kilobit/">23 GF1024: один килобит</a></li><li class="on" aria-current="page"><a href="/ru/learn/gft8-exponent-in-trits/">24 GF-T8: порядок в тритах</a></li></ol></li><li><h3>Модуль 9 · Ещё триты, затем декодирование</h3><p>GF-T16 и GF-T32, затем почему фиксированные поля декодируются параллельно, а posit — нет.</p><ol><li><a href="/ru/learn/gft16-six-trits/">25 GF-T16: шесть тритов</a></li><li><a href="/ru/learn/gft32-twelve-trits/">26 GF-T32: двенадцать тритов</a></li><li><a href="/ru/learn/decode-in-one-step/">27 Раскодировать за один шаг</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/global-buffers/index.html b/apps/website/public/ru/learn/global-buffers/index.html
index 1bfe01c493..668b754873 100644
--- a/apps/website/public/ru/learn/global-buffers/index.html
+++ b/apps/website/public/ru/learn/global-buffers/index.html
@@ -294,5 +294,6 @@ <h1>Глобальный буфер</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое тактовый сигнал</h3><p>Один фронт, один мир: всё, что делит фронт тактового сигнала, делит и мир; период и джиттер настоящего фронта и место, где тактовый сигнал входит в плату.</p><ol><li><a href="/ru/learn/one-edge-one-world/">01 Один фронт, один мир</a></li><li><a href="/ru/learn/period-and-jitter/">02 Период, фронт и джиттер</a></li><li><a href="/ru/learn/where-the-clock-enters/">03 Где входит тактовый сигнал</a></li></ol></li><li><h3>Модуль 2 · Деревья тактового сигнала</h3><p>Перекос и задержка прохождения, глобальная сеть буферов и ловушка стробирования клока логикой.</p><ol><li><a href="/ru/learn/skew-and-insertion/">04 Перекос и задержка прохождения</a></li><li class="on" aria-current="page"><a href="/ru/learn/global-buffers/">05 Глобальный буфер</a></li><li><a href="/ru/learn/gated-clock-trap/">06 Ловушка стробированного клока</a></li></ol></li><li><h3>Модуль 3 · PLL и MMCM</h3><p>Умножить и поделить один тактовый сигнал в другой, сдвинуть его фазу шагами VCO и какие тактовые сигналы анализатор считает родственными.</p><ol><li><a href="/ru/learn/multiply-and-divide/">07 Умножить и поделить</a></li><li><a href="/ru/learn/phase-and-lock/">08 Фаза и захват</a></li><li><a href="/ru/learn/related-or-not/">09 Родственные или чужие</a></li></ol></li><li><h3>Модуль 4 · Сбросы</h3><p>Ассертировать асинхронно, снимать синхронно: три вида сброса, конвейер снятия и дерево, которое растит сброс.</p><ol><li><a href="/ru/learn/sync-or-async/">10 Синхронный или асинхронный</a></li><li><a href="/ru/learn/reset-release/">11 Конвейер снятия</a></li><li><a href="/ru/learn/reset-fanout/">12 Разветвление сброса</a></li></ol></li><li><h3>Модуль 5 · Метастабильность</h3><p>Окно setup-hold, среднее время между отказами в целочисленной арифметике и два триггера, которые это чинят.</p><ol><li><a href="/ru/learn/setup-hold-window/">13 Окно setup-hold</a></li><li><a href="/ru/learn/mtbf/">14 Среднее время между отказами</a></li><li><a href="/ru/learn/two-flop-sync/">15 Двухтриггерный синхронизатор</a></li></ol></li><li><h3>Модуль 6 · Пересечение многих бит</h3><p>Почему двоичная шина рвётся, почему код Грея нет, и рукопожатие, переносящее импульс между мирами.</p><ol><li><a href="/ru/learn/why-binary-fails/">16 Почему двоичный код рвётся</a></li><li><a href="/ru/learn/gray-code/">17 Код Грея</a></li><li><a href="/ru/learn/handshakes-and-pulses/">18 Рукопожатия и импульсы</a></li></ol></li><li><h3>Модуль 7 · Асинхронный FIFO</h3><p>Указатели, флаги и глубина: буфер, переносящий поток между двумя тактовыми сигналами.</p><ol><li><a href="/ru/learn/pointers-and-flags/">19 Указатели и флаги</a></li><li><a href="/ru/learn/depth-sizing/">20 Выбор глубины</a></li><li><a href="/ru/learn/fifo-testbench/">21 Тестбенч FIFO</a></li></ol></li><li><h3>Модуль 8 · Ограничения</h3><p>Строки, сообщающие анализатору, что такое тактовый сигнал, какие пути не проверять и чему должны соответствовать выводы.</p><ol><li><a href="/ru/learn/create-clock/">22 Строка create_clock</a></li><li><a href="/ru/learn/false-path-and-max-delay/">23 False path и max delay</a></li><li><a href="/ru/learn/io-timing/">24 Тайминг выводов</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Отчёт CDC, одно пересечение, захваченное на триггерах, и дифф битстрима, замыкающий курс.</p><ol><li><a href="/ru/learn/cdc-report/">25 Отчёт CDC</a></li><li><a href="/ru/learn/capture-a-crossing/">26 Захватить пересечение</a></li><li><a href="/ru/learn/capstone-diff/">27 Капстоун: поменяй и сравни</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/golden-model/index.html b/apps/website/public/ru/learn/golden-model/index.html
index 27b4bb54df..14efc6fbe1 100644
--- a/apps/website/public/ru/learn/golden-model/index.html
+++ b/apps/website/public/ru/learn/golden-model/index.html
@@ -360,5 +360,6 @@ <h1>Золотая модель</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Зачем проверять</h3><p>Дизайны, которые компилируются и ошибаются; модель, которая выносит вердикт; план, записанный до кода.</p><ol><li><a href="/ru/learn/bugs-that-compile/">01 Баги, которые компилируются</a></li><li class="on" aria-current="page"><a href="/ru/learn/golden-model/">02 Золотая модель</a></li><li><a href="/ru/learn/plan-before-code/">03 Сначала план</a></li></ol></li><li><h3>Модуль 2 · Тестбенчи</h3><p>Стимулы, проверки и вердикт, записанные как один spec рядом с дизайном, который они судят.</p><ol><li><a href="/ru/learn/anatomy-of-a-tb/">04 Анатомия тестбенча</a></li><li><a href="/ru/learn/self-checking/">05 Самопроверка вместо всматривания</a></li><li><a href="/ru/learn/directed-vs-random/">06 Направленные против случайных</a></li></ol></li><li><h3>Модуль 3 · Временные диаграммы</h3><p>Трасса каждого сигнала, прочитанная так, как её читает инженер по железу, и два прогона, сравнённые между собой.</p><ol><li><a href="/ru/learn/vcd-format/">07 Формат VCD</a></li><li><a href="/ru/learn/debug-with-waves/">08 Отладка по волнам</a></li><li><a href="/ru/learn/compare-traces/">09 Сравнение двух трасс</a></li></ol></li><li><h3>Модуль 4 · Векторы соответствия</h3><p>Случаи с ответом, записанным рядом, там, откуда их достанет компилятор.</p><ol><li><a href="/ru/learn/what-a-vector-is/">10 Что такое вектор</a></li><li><a href="/ru/learn/golden-ruler-vectors/">11 Векторы как эталонная линейка</a></li><li><a href="/ru/learn/end-to-end-scenarios/">12 Сквозные сценарии</a></li></ol></li><li><h3>Модуль 5 · Косимуляция</h3><p>Spec, симулятор и плата сходятся в одном ответе на стенде Artix-7 XC7A200T, и что делать, когда не сходятся.</p><ol><li><a href="/ru/learn/three-way-agreement/">13 Тройное согласие</a></li><li><a href="/ru/learn/simulator-internals/">14 Внутри симулятора</a></li><li><a href="/ru/learn/when-they-disagree/">15 Когда они не сходятся</a></li></ol></li><li><h3>Модуль 6 · Покрытие</h3><p>Чего коснулись тесты: строки, переключения, состояния — и что прячет это число.</p><ol><li><a href="/ru/learn/line-and-toggle/">16 Строки и переключения</a></li><li><a href="/ru/learn/fsm-coverage/">17 Покрытие автоматов</a></li><li><a href="/ru/learn/coverage-lies/">18 Покрытие лжёт</a></li></ol></li><li><h3>Модуль 7 · Формальные методы</h3><p>Ассерты, верные каждый такт; ограниченный поиск контрпримера; и почему доказательству нужна индукция.</p><ol><li><a href="/ru/learn/assertions/">19 Ассерты</a></li><li><a href="/ru/learn/bounded-model-checking/">20 Ограниченная проверка моделей</a></li><li><a href="/ru/learn/induction/">21 Индукция</a></li></ol></li><li><h3>Модуль 8 · Мутации</h3><p>Ломайте дизайн нарочно и считайте, что заметили тесты.</p><ol><li><a href="/ru/learn/kill-the-mutant/">22 Убить мутанта</a></li><li><a href="/ru/learn/surviving-mutants/">23 Выжившие мутанты</a></li><li><a href="/ru/learn/test-strength/">24 Сила тестов</a></li></ol></li><li><h3>Модуль 9 · Приёмка</h3><p>Одна команда, все квитанции, чистый вердикт, который можно показать.</p><ol><li><a href="/ru/learn/build-verify/">25 Сборка и проверка</a></li><li><a href="/ru/learn/board-selftest/">26 Стенд проверяет себя</a></li><li><a href="/ru/learn/verify-capstone/">27 Капстоун: проверить модуль целиком</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">Этот курс сам является spec t27: прочитайте его</a></p><p>Виджеты сохраняют свои английские слова: каждый живёт в собственном spec t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/golden-ruler-vectors/index.html b/apps/website/public/ru/learn/golden-ruler-vectors/index.html
index e9be956454..cb7ba1e012 100644
--- a/apps/website/public/ru/learn/golden-ruler-vectors/index.html
+++ b/apps/website/public/ru/learn/golden-ruler-vectors/index.html
@@ -565,5 +565,6 @@ <h1>Векторы как эталонная линейка</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Зачем проверять</h3><p>Дизайны, которые компилируются и ошибаются; модель, которая выносит вердикт; план, записанный до кода.</p><ol><li><a href="/ru/learn/bugs-that-compile/">01 Баги, которые компилируются</a></li><li><a href="/ru/learn/golden-model/">02 Золотая модель</a></li><li><a href="/ru/learn/plan-before-code/">03 Сначала план</a></li></ol></li><li><h3>Модуль 2 · Тестбенчи</h3><p>Стимулы, проверки и вердикт, записанные как один spec рядом с дизайном, который они судят.</p><ol><li><a href="/ru/learn/anatomy-of-a-tb/">04 Анатомия тестбенча</a></li><li><a href="/ru/learn/self-checking/">05 Самопроверка вместо всматривания</a></li><li><a href="/ru/learn/directed-vs-random/">06 Направленные против случайных</a></li></ol></li><li><h3>Модуль 3 · Временные диаграммы</h3><p>Трасса каждого сигнала, прочитанная так, как её читает инженер по железу, и два прогона, сравнённые между собой.</p><ol><li><a href="/ru/learn/vcd-format/">07 Формат VCD</a></li><li><a href="/ru/learn/debug-with-waves/">08 Отладка по волнам</a></li><li><a href="/ru/learn/compare-traces/">09 Сравнение двух трасс</a></li></ol></li><li><h3>Модуль 4 · Векторы соответствия</h3><p>Случаи с ответом, записанным рядом, там, откуда их достанет компилятор.</p><ol><li><a href="/ru/learn/what-a-vector-is/">10 Что такое вектор</a></li><li class="on" aria-current="page"><a href="/ru/learn/golden-ruler-vectors/">11 Векторы как эталонная линейка</a></li><li><a href="/ru/learn/end-to-end-scenarios/">12 Сквозные сценарии</a></li></ol></li><li><h3>Модуль 5 · Косимуляция</h3><p>Spec, симулятор и плата сходятся в одном ответе на стенде Artix-7 XC7A200T, и что делать, когда не сходятся.</p><ol><li><a href="/ru/learn/three-way-agreement/">13 Тройное согласие</a></li><li><a href="/ru/learn/simulator-internals/">14 Внутри симулятора</a></li><li><a href="/ru/learn/when-they-disagree/">15 Когда они не сходятся</a></li></ol></li><li><h3>Модуль 6 · Покрытие</h3><p>Чего коснулись тесты: строки, переключения, состояния — и что прячет это число.</p><ol><li><a href="/ru/learn/line-and-toggle/">16 Строки и переключения</a></li><li><a href="/ru/learn/fsm-coverage/">17 Покрытие автоматов</a></li><li><a href="/ru/learn/coverage-lies/">18 Покрытие лжёт</a></li></ol></li><li><h3>Модуль 7 · Формальные методы</h3><p>Ассерты, верные каждый такт; ограниченный поиск контрпримера; и почему доказательству нужна индукция.</p><ol><li><a href="/ru/learn/assertions/">19 Ассерты</a></li><li><a href="/ru/learn/bounded-model-checking/">20 Ограниченная проверка моделей</a></li><li><a href="/ru/learn/induction/">21 Индукция</a></li></ol></li><li><h3>Модуль 8 · Мутации</h3><p>Ломайте дизайн нарочно и считайте, что заметили тесты.</p><ol><li><a href="/ru/learn/kill-the-mutant/">22 Убить мутанта</a></li><li><a href="/ru/learn/surviving-mutants/">23 Выжившие мутанты</a></li><li><a href="/ru/learn/test-strength/">24 Сила тестов</a></li></ol></li><li><h3>Модуль 9 · Приёмка</h3><p>Одна команда, все квитанции, чистый вердикт, который можно показать.</p><ol><li><a href="/ru/learn/build-verify/">25 Сборка и проверка</a></li><li><a href="/ru/learn/board-selftest/">26 Стенд проверяет себя</a></li><li><a href="/ru/learn/verify-capstone/">27 Капстоун: проверить модуль целиком</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">Этот курс сам является spec t27: прочитайте его</a></p><p>Виджеты сохраняют свои английские слова: каждый живёт в собственном spec t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/goldenfloat/index.html b/apps/website/public/ru/learn/goldenfloat/index.html
index 249615319c..4e093da75e 100644
--- a/apps/website/public/ru/learn/goldenfloat/index.html
+++ b/apps/website/public/ru/learn/goldenfloat/index.html
@@ -67,5 +67,6 @@ <h1>GoldenFloat на t27: одно правило, семнадцать шири
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Правило и его числа</h3><p>Одно правило делит каждую ширину, отношение, к которому оно стремится, и числа Люка за тройкой 3.</p><ol><li><a href="/ru/learn/a-float-cut-by-phi/">01 Число с плавающей точкой, разрезанное по phi</a></li><li><a href="/ru/learn/phi-as-a-ratio/">02 Phi как отношение</a></li><li><a href="/ru/learn/lucas-numbers-stay-whole/">03 Числа Люка остаются целыми</a></li></ol></li><li><h3>Модуль 2 · Почему phi, почему три</h3><p>Почему деление идёт по phi, почему основание три и как спека проверяет, что GF16 хранит phi.</p><ol><li><a href="/ru/learn/why-the-split-is-phi/">04 Почему деление по phi</a></li><li><a href="/ru/learn/why-base-three/">05 Почему основание три</a></li><li><a href="/ru/learn/phi-in-sixteen-bits/">06 Phi в шестнадцати битах</a></li></ol></li><li><h3>Модуль 3 · Малые ступени: от GF4 до GF8</h3><p>GF4, GF6 и GF8 — меньше всего битов, и округление до целых битов стоит здесь дороже всего.</p><ol><li><a href="/ru/learn/gf4-four-bits/">07 GF4: четыре бита</a></li><li><a href="/ru/learn/gf6-six-bits/">08 GF6: шесть битов</a></li><li><a href="/ru/learn/gf8-one-byte/">09 GF8: один байт</a></li></ol></li><li><h3>Модуль 4 · От десяти до четырнадцати битов</h3><p>GF10, GF12 и GF14 и то, как расстояние до 1 / phi меняется с ростом слова.</p><ol><li><a href="/ru/learn/gf10-ten-bits/">10 GF10: десять битов</a></li><li><a href="/ru/learn/gf12-twelve-bits/">11 GF12: двенадцать битов</a></li><li><a href="/ru/learn/gf14-fourteen-bits/">12 GF14: четырнадцать битов</a></li></ol></li><li><h3>Модуль 5 · GF16 в работе</h3><p>Основной 16-битный формат, скалярное произведение из двух слагаемых в GF-T16, затем GF20 и GF24.</p><ol><li><a href="/ru/learn/gf16-the-primary/">13 GF16: основной формат</a></li><li><a href="/ru/learn/two-products-one-sum/">14 Два произведения, одна сумма</a></li><li><a href="/ru/learn/gf24-twenty-four-bits/">15 GF20 и GF24</a></li></ol></li><li><h3>Модуль 6 · От GF32 до GF64</h3><p>GF32 рядом с IEEE single, GF48 без пары в IEEE, GF64 рядом с IEEE double.</p><ol><li><a href="/ru/learn/gf32-a-single/">16 GF32: одинарная точность</a></li><li><a href="/ru/learn/gf48-forty-eight-bits/">17 GF48: сорок восемь битов</a></li><li><a href="/ru/learn/gf64-a-double/">18 GF64: двойная точность</a></li></ol></li><li><h3>Модуль 7 · От GF96 до GF256</h3><p>GF96, GF128 и GF256, где спеки держат раскладку инвариантами.</p><ol><li><a href="/ru/learn/gf96-ninety-six-bits/">19 GF96: девяносто шесть битов</a></li><li><a href="/ru/learn/gf128-a-quad/">20 GF128: четверная точность</a></li><li><a href="/ru/learn/gf256-two-five-six/">21 GF256: двести пятьдесят шесть битов</a></li></ol></li><li><h3>Модуль 8 · Самые широкие ступени, затем триты</h3><p>GF512 и GF1024, две самые широкие ступени, затем GF-T8, где порядок уходит в триты.</p><ol><li><a href="/ru/learn/gf512-five-twelve/">22 GF512: пятьсот двенадцать битов</a></li><li><a href="/ru/learn/gf1024-one-kilobit/">23 GF1024: один килобит</a></li><li><a href="/ru/learn/gft8-exponent-in-trits/">24 GF-T8: порядок в тритах</a></li></ol></li><li><h3>Модуль 9 · Ещё триты, затем декодирование</h3><p>GF-T16 и GF-T32, затем почему фиксированные поля декодируются параллельно, а posit — нет.</p><ol><li><a href="/ru/learn/gft16-six-trits/">25 GF-T16: шесть тритов</a></li><li><a href="/ru/learn/gft32-twelve-trits/">26 GF-T32: двенадцать тритов</a></li><li><a href="/ru/learn/decode-in-one-step/">27 Раскодировать за один шаг</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/gray-code/index.html b/apps/website/public/ru/learn/gray-code/index.html
index 9e8a0f6df8..fad4365c54 100644
--- a/apps/website/public/ru/learn/gray-code/index.html
+++ b/apps/website/public/ru/learn/gray-code/index.html
@@ -434,5 +434,6 @@ <h1>Код Грея</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое тактовый сигнал</h3><p>Один фронт, один мир: всё, что делит фронт тактового сигнала, делит и мир; период и джиттер настоящего фронта и место, где тактовый сигнал входит в плату.</p><ol><li><a href="/ru/learn/one-edge-one-world/">01 Один фронт, один мир</a></li><li><a href="/ru/learn/period-and-jitter/">02 Период, фронт и джиттер</a></li><li><a href="/ru/learn/where-the-clock-enters/">03 Где входит тактовый сигнал</a></li></ol></li><li><h3>Модуль 2 · Деревья тактового сигнала</h3><p>Перекос и задержка прохождения, глобальная сеть буферов и ловушка стробирования клока логикой.</p><ol><li><a href="/ru/learn/skew-and-insertion/">04 Перекос и задержка прохождения</a></li><li><a href="/ru/learn/global-buffers/">05 Глобальный буфер</a></li><li><a href="/ru/learn/gated-clock-trap/">06 Ловушка стробированного клока</a></li></ol></li><li><h3>Модуль 3 · PLL и MMCM</h3><p>Умножить и поделить один тактовый сигнал в другой, сдвинуть его фазу шагами VCO и какие тактовые сигналы анализатор считает родственными.</p><ol><li><a href="/ru/learn/multiply-and-divide/">07 Умножить и поделить</a></li><li><a href="/ru/learn/phase-and-lock/">08 Фаза и захват</a></li><li><a href="/ru/learn/related-or-not/">09 Родственные или чужие</a></li></ol></li><li><h3>Модуль 4 · Сбросы</h3><p>Ассертировать асинхронно, снимать синхронно: три вида сброса, конвейер снятия и дерево, которое растит сброс.</p><ol><li><a href="/ru/learn/sync-or-async/">10 Синхронный или асинхронный</a></li><li><a href="/ru/learn/reset-release/">11 Конвейер снятия</a></li><li><a href="/ru/learn/reset-fanout/">12 Разветвление сброса</a></li></ol></li><li><h3>Модуль 5 · Метастабильность</h3><p>Окно setup-hold, среднее время между отказами в целочисленной арифметике и два триггера, которые это чинят.</p><ol><li><a href="/ru/learn/setup-hold-window/">13 Окно setup-hold</a></li><li><a href="/ru/learn/mtbf/">14 Среднее время между отказами</a></li><li><a href="/ru/learn/two-flop-sync/">15 Двухтриггерный синхронизатор</a></li></ol></li><li><h3>Модуль 6 · Пересечение многих бит</h3><p>Почему двоичная шина рвётся, почему код Грея нет, и рукопожатие, переносящее импульс между мирами.</p><ol><li><a href="/ru/learn/why-binary-fails/">16 Почему двоичный код рвётся</a></li><li class="on" aria-current="page"><a href="/ru/learn/gray-code/">17 Код Грея</a></li><li><a href="/ru/learn/handshakes-and-pulses/">18 Рукопожатия и импульсы</a></li></ol></li><li><h3>Модуль 7 · Асинхронный FIFO</h3><p>Указатели, флаги и глубина: буфер, переносящий поток между двумя тактовыми сигналами.</p><ol><li><a href="/ru/learn/pointers-and-flags/">19 Указатели и флаги</a></li><li><a href="/ru/learn/depth-sizing/">20 Выбор глубины</a></li><li><a href="/ru/learn/fifo-testbench/">21 Тестбенч FIFO</a></li></ol></li><li><h3>Модуль 8 · Ограничения</h3><p>Строки, сообщающие анализатору, что такое тактовый сигнал, какие пути не проверять и чему должны соответствовать выводы.</p><ol><li><a href="/ru/learn/create-clock/">22 Строка create_clock</a></li><li><a href="/ru/learn/false-path-and-max-delay/">23 False path и max delay</a></li><li><a href="/ru/learn/io-timing/">24 Тайминг выводов</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Отчёт CDC, одно пересечение, захваченное на триггерах, и дифф битстрима, замыкающий курс.</p><ol><li><a href="/ru/learn/cdc-report/">25 Отчёт CDC</a></li><li><a href="/ru/learn/capture-a-crossing/">26 Захватить пересечение</a></li><li><a href="/ru/learn/capstone-diff/">27 Капстоун: поменяй и сравни</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/growth-through-the-stages/index.html b/apps/website/public/ru/learn/growth-through-the-stages/index.html
index ae3fc227d9..217eb58bbb 100644
--- a/apps/website/public/ru/learn/growth-through-the-stages/index.html
+++ b/apps/website/public/ru/learn/growth-through-the-stages/index.html
@@ -1122,5 +1122,6 @@ <h1>Рост через этапы</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Сложение</h3><p>Почему сложение медленное: перенос, который идёт вверх по слову, выделенная цепь в каждом слайсе, префиксные сети и деревья сохранения переноса.</p><ol><li><a href="/ru/learn/the-carry-chain/">01 Цепь переноса</a></li><li><a href="/ru/learn/prefix-adders/">02 Взгляд вперёд: префиксные сумматоры</a></li><li><a href="/ru/learn/carry-save/">03 Три числа сразу</a></li></ol></li><li><h3>Модуль 2 · Умножение</h3><p>Произведение — сумма сдвинутых копий; перекодирование Бута вдвое сокращает их, а слайс DSP48E1 делает остальное одним блоком.</p><ol><li><a href="/ru/learn/shift-and-add/">04 Сдвиг и сложение</a></li><li><a href="/ru/learn/booths-trick/">05 Приём Бута</a></li><li><a href="/ru/learn/the-dsp48e1-slice/">06 Слайс DSP48E1</a></li></ol></li><li><h3>Модуль 3 · Фиксированная точка</h3><p>Где стоит двоичная точка, что округление делает со значением и с его средним и что даёт каждый бит квантователя.</p><ol><li><a href="/ru/learn/q-formats/">07 Форматы Q</a></li><li><a href="/ru/learn/rounding-and-saturation/">08 Округление и насыщение</a></li><li><a href="/ru/learn/quantization-noise/">09 Шум квантования</a></li></ol></li><li><h3>Модуль 4 · Функции в железе</h3><p>Синус, косинус, угол и длина из сдвигов и сложений — и когда таблица оказывается выгоднее.</p><ol><li><a href="/ru/learn/rotating-by-shifts/">10 Поворот сдвигами</a></li><li><a href="/ru/learn/angle-and-magnitude/">11 Угол и модуль</a></li><li><a href="/ru/learn/tables-and-interpolation/">12 Таблицы и интерполяция</a></li></ol></li><li><h3>Модуль 5 · Сигналы и дискретизация</h3><p>Что дискретизация делает с частотой, как генератор строится из сумматора и что измеряет бин ДПФ.</p><ol><li><a href="/ru/learn/sampling-and-aliasing/">13 Дискретизация и наложение спектров</a></li><li><a href="/ru/learn/the-nco/">14 Генератор с числовым управлением</a></li><li><a href="/ru/learn/bins-of-a-spectrum/">15 Бины спектра</a></li></ol></li><li><h3>Модуль 6 · Фильтры FIR</h3><p>Скользящее среднее, расчёт окном sinc с целыми коэффициентами и раскладка отводов по слайсам DSP.</p><ol><li><a href="/ru/learn/the-moving-average/">16 Скользящее среднее</a></li><li><a href="/ru/learn/designing-taps/">17 Расчёт коэффициентов</a></li><li><a href="/ru/learn/folding-taps-onto-slices/">18 Раскладка отводов по слайсам</a></li></ol></li><li><h3>Модуль 7 · Многоскоростная обработка</h3><p>Понижение частоты дискретизации без заворота шума: децимация, фильтр CIC и полифазная форма.</p><ol><li><a href="/ru/learn/lowering-the-rate/">19 Понижение частоты</a></li><li><a href="/ru/learn/the-cic-filter/">20 Фильтр CIC</a></li><li><a href="/ru/learn/polyphase/">21 Полифазная форма</a></li></ol></li><li><h3>Модуль 8 · БПФ</h3><p>N log N вместо N^2: бабочки, бит-реверсный порядок входа и биты, которые добавляет каждый этап.</p><ol><li><a href="/ru/learn/the-butterfly/">22 Бабочка</a></li><li><a href="/ru/learn/twiddles-and-bit-reversal/">23 Поворачивающие множители и бит-реверс</a></li><li class="on" aria-current="page"><a href="/ru/learn/growth-through-the-stages/">24 Рост через этапы</a></li></ol></li><li><h3>Модуль 9 · На стенде</h3><p>От арифметики к плате: коррелятор из спеки модема, бюджет слайсов и тайминга и итоговый фильтр.</p><ol><li><a href="/ru/learn/a-correlator-finds-the-frame/">25 Коррелятор находит кадр</a></li><li><a href="/ru/learn/does-it-fit-and-close-timing/">26 Помещается ли и сходится ли тайминг?</a></li><li><a href="/ru/learn/capstone-a-filter-on-the-board/">27 Итог: фильтр на плате</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/half-adder/index.html b/apps/website/public/ru/learn/half-adder/index.html
index 7448622c40..30f71bdf26 100644
--- a/apps/website/public/ru/learn/half-adder/index.html
+++ b/apps/website/public/ru/learn/half-adder/index.html
@@ -769,5 +769,6 @@ <h1>Полусумматор</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Три значения</h3><p>Почему три, как сбалансированная троичная система пишет любое число без знака и что дают переворот и отсечение тритов.</p><ol><li><a href="/ru/learn/why-three/">01 Почему три</a></li><li><a href="/ru/learn/balanced-ternary/">02 Сбалансированная троичная</a></li><li><a href="/ru/learn/flip-and-cut/">03 Переворот вместо минуса, отсечение вместо округления</a></li></ol></li><li><h3>Модуль 2 · Логика с «не знаю»</h3><p>Трёхзначные вентили Клини, два вентиля без двоичного двойника и сложение как пара таблиц.</p><ol><li><a href="/ru/learn/kleene-logic/">04 Трёхзначная логика</a></li><li><a href="/ru/learn/consensus-and-any/">05 Консенсус и «любой»</a></li><li><a href="/ru/learn/sum-and-carry/">06 Сумма и перенос</a></li></ol></li><li><h3>Модуль 3 · Складываем триты</h3><p>Полусумматор, полный сумматор и перенос, бегущий влево разряд за разрядом.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/half-adder/">07 Полусумматор</a></li><li><a href="/ru/learn/full-adder/">08 Полный сумматор</a></li><li><a href="/ru/learn/ripple-carry/">09 Сквозной перенос</a></li></ol></li><li><h3>Модуль 4 · Умножение без умножителя</h3><p>Копия, ноль или переворот: произведение на один трит, умножение столбиком и MAC, который только складывает.</p><ol><li><a href="/ru/learn/times-a-trit/">10 Умножить на один трит</a></li><li><a href="/ru/learn/long-multiplication/">11 Умножение столбиком</a></li><li><a href="/ru/learn/ternary-mac/">12 Троичный MAC</a></li></ol></li><li><h3>Модуль 5 · Триты в двоичной памяти</h3><p>Два бита на трит, пять тритов на байт и сколько бит нужно слову из 27 тритов.</p><ol><li><a href="/ru/learn/two-bits-per-trit/">13 Два бита на трит</a></li><li><a href="/ru/learn/five-trits-a-byte/">14 Пять тритов в байте</a></li><li><a href="/ru/learn/a-word-of-27/">15 Слово из 27 тритов</a></li></ol></li><li><h3>Модуль 6 · Командное слово TRI-27</h3><p>32-битное слово, которое декодирует эмулятор Trinity: его поля, 47 опкодов и 15-битное непосредственное значение.</p><ol><li><a href="/ru/learn/the-instruction-word/">16 Командное слово</a></li><li><a href="/ru/learn/the-opcode-map/">17 Карта опкодов</a></li><li><a href="/ru/learn/the-immediate/">18 Непосредственное значение</a></li></ol></li><li><h3>Модуль 7 · Троичная машина</h3><p>АЛУ из сумматоров этого курса, переход на три стороны и программа, которую можно пройти по шагам.</p><ol><li><a href="/ru/learn/a-ternary-alu/">19 Троичное АЛУ</a></li><li><a href="/ru/learn/three-way-jump/">20 Переход на три стороны</a></li><li><a href="/ru/learn/run-a-program/">21 Запустить программу</a></li></ol></li><li><h3>Модуль 8 · Три ответа</h3><p>Сравнение по старшему различному триту, поиск числа по третям и сортировка трёхзначными сравнениями.</p><ol><li><a href="/ru/learn/compare-at-the-top/">22 Сравнение сверху</a></li><li><a href="/ru/learn/questions-in-thirds/">23 Вопросы по третям</a></li><li><a href="/ru/learn/sort-by-three/">24 Сортировка трёхзначными сравнениями</a></li></ol></li><li><h3>Модуль 9 · Троичные нейроны</h3><p>Веса -1, 0, +1: один нейрон, детектор и маленький слой, и нигде нет умножителя.</p><ol><li><a href="/ru/learn/one-ternary-neuron/">25 Один троичный нейрон</a></li><li><a href="/ru/learn/a-bar-detector/">26 Детектор полосы</a></li><li><a href="/ru/learn/a-tiny-layer/">27 Маленький слой</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/handshakes-and-pulses/index.html b/apps/website/public/ru/learn/handshakes-and-pulses/index.html
index fc922b9699..e924a46c08 100644
--- a/apps/website/public/ru/learn/handshakes-and-pulses/index.html
+++ b/apps/website/public/ru/learn/handshakes-and-pulses/index.html
@@ -300,5 +300,6 @@ <h1>Рукопожатия и импульсы</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое тактовый сигнал</h3><p>Один фронт, один мир: всё, что делит фронт тактового сигнала, делит и мир; период и джиттер настоящего фронта и место, где тактовый сигнал входит в плату.</p><ol><li><a href="/ru/learn/one-edge-one-world/">01 Один фронт, один мир</a></li><li><a href="/ru/learn/period-and-jitter/">02 Период, фронт и джиттер</a></li><li><a href="/ru/learn/where-the-clock-enters/">03 Где входит тактовый сигнал</a></li></ol></li><li><h3>Модуль 2 · Деревья тактового сигнала</h3><p>Перекос и задержка прохождения, глобальная сеть буферов и ловушка стробирования клока логикой.</p><ol><li><a href="/ru/learn/skew-and-insertion/">04 Перекос и задержка прохождения</a></li><li><a href="/ru/learn/global-buffers/">05 Глобальный буфер</a></li><li><a href="/ru/learn/gated-clock-trap/">06 Ловушка стробированного клока</a></li></ol></li><li><h3>Модуль 3 · PLL и MMCM</h3><p>Умножить и поделить один тактовый сигнал в другой, сдвинуть его фазу шагами VCO и какие тактовые сигналы анализатор считает родственными.</p><ol><li><a href="/ru/learn/multiply-and-divide/">07 Умножить и поделить</a></li><li><a href="/ru/learn/phase-and-lock/">08 Фаза и захват</a></li><li><a href="/ru/learn/related-or-not/">09 Родственные или чужие</a></li></ol></li><li><h3>Модуль 4 · Сбросы</h3><p>Ассертировать асинхронно, снимать синхронно: три вида сброса, конвейер снятия и дерево, которое растит сброс.</p><ol><li><a href="/ru/learn/sync-or-async/">10 Синхронный или асинхронный</a></li><li><a href="/ru/learn/reset-release/">11 Конвейер снятия</a></li><li><a href="/ru/learn/reset-fanout/">12 Разветвление сброса</a></li></ol></li><li><h3>Модуль 5 · Метастабильность</h3><p>Окно setup-hold, среднее время между отказами в целочисленной арифметике и два триггера, которые это чинят.</p><ol><li><a href="/ru/learn/setup-hold-window/">13 Окно setup-hold</a></li><li><a href="/ru/learn/mtbf/">14 Среднее время между отказами</a></li><li><a href="/ru/learn/two-flop-sync/">15 Двухтриггерный синхронизатор</a></li></ol></li><li><h3>Модуль 6 · Пересечение многих бит</h3><p>Почему двоичная шина рвётся, почему код Грея нет, и рукопожатие, переносящее импульс между мирами.</p><ol><li><a href="/ru/learn/why-binary-fails/">16 Почему двоичный код рвётся</a></li><li><a href="/ru/learn/gray-code/">17 Код Грея</a></li><li class="on" aria-current="page"><a href="/ru/learn/handshakes-and-pulses/">18 Рукопожатия и импульсы</a></li></ol></li><li><h3>Модуль 7 · Асинхронный FIFO</h3><p>Указатели, флаги и глубина: буфер, переносящий поток между двумя тактовыми сигналами.</p><ol><li><a href="/ru/learn/pointers-and-flags/">19 Указатели и флаги</a></li><li><a href="/ru/learn/depth-sizing/">20 Выбор глубины</a></li><li><a href="/ru/learn/fifo-testbench/">21 Тестбенч FIFO</a></li></ol></li><li><h3>Модуль 8 · Ограничения</h3><p>Строки, сообщающие анализатору, что такое тактовый сигнал, какие пути не проверять и чему должны соответствовать выводы.</p><ol><li><a href="/ru/learn/create-clock/">22 Строка create_clock</a></li><li><a href="/ru/learn/false-path-and-max-delay/">23 False path и max delay</a></li><li><a href="/ru/learn/io-timing/">24 Тайминг выводов</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Отчёт CDC, одно пересечение, захваченное на триггерах, и дифф битстрима, замыкающий курс.</p><ol><li><a href="/ru/learn/cdc-report/">25 Отчёт CDC</a></li><li><a href="/ru/learn/capture-a-crossing/">26 Захватить пересечение</a></li><li><a href="/ru/learn/capstone-diff/">27 Капстоун: поменяй и сравни</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/hello-t27/index.html b/apps/website/public/ru/learn/hello-t27/index.html
index a1b00a96e1..7124a4abe5 100644
--- a/apps/website/public/ru/learn/hello-t27/index.html
+++ b/apps/website/public/ru/learn/hello-t27/index.html
@@ -153,5 +153,6 @@ <h1>Привет, t27</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Чип</h3><p>Что такое FPGA, какой чип у нас и как его выводы встречаются с платой.</p><ol><li><a href="/ru/learn/fpga-is-a-grid/">01 FPGA — это сетка, которую вы настраиваете</a></li><li><a href="/ru/learn/which-chip/">02 Какой это чип?</a></li><li><a href="/ru/learn/pins-and-package/">03 Выводы, шарики и корпус</a></li></ol></li><li><h3>Модуль 2 · Числа в железе</h3><p>Биты, триты и форматы чисел, и сколько стоит арифметика в логике.</p><ol><li><a href="/ru/learn/bits-and-trits/">04 Биты, триты и GF16</a></li><li><a href="/ru/learn/what-arithmetic-costs/">05 Сколько стоит арифметика</a></li><li><a href="/ru/learn/how-much-fits/">06 Сколько помещается на чип</a></li></ol></li><li><h3>Модуль 3 · Ваша программа на t27</h3><p>Напишите спеку, проверьте её и увидьте, почему «компилируется» не значит «верно».</p><ol><li class="on" aria-current="page"><a href="/ru/learn/hello-t27/">07 Привет, t27</a></li><li><a href="/ru/learn/tests-are-the-spec/">08 Тесты — часть спеки</a></li><li><a href="/ru/learn/compiles-is-not-correct/">09 Компилируется — не значит верно</a></li></ol></li><li><h3>Модуль 4 · Внутри t27c</h3><p>Как компилятор читает спеку и что он пишет, включая нативный t27b.</p><ol><li><a href="/ru/learn/compiler-stages/">10 Стадии t27c</a></li><li><a href="/ru/learn/t27c-on-a-real-machine/">11 t27c на настоящей машине</a></li><li><a href="/ru/learn/seven-backends-and-t27b/">12 Семь бэкендов и t27b</a></li></ol></li><li><h3>Модуль 5 · От спеки к железу</h3><p>Verilog, который пишет t27c, ячейки, в которые он превращается, и что делает одна LUT.</p><ol><li><a href="/ru/learn/spec-to-verilog/">13 От спеки к Verilog</a></li><li><a href="/ru/learn/count-the-cells/">14 Посчитайте ячейки</a></li><li><a href="/ru/learn/one-lut-up-close/">15 Одна LUT вблизи</a></li></ol></li><li><h3>Модуль 6 · Читаем синтез</h3><p>Что yosys сообщает о вашем дизайне и какие предупреждения важны.</p><ol><li><a href="/ru/learn/synthesis-warnings/">16 Важные предупреждения синтеза</a></li><li><a href="/ru/learn/state-machines/">17 Конечные автоматы</a></li><li><a href="/ru/learn/where-the-luts-go/">18 Куда уходят LUT</a></li></ol></li><li><h3>Модуль 7 · Размещение, трассировка, тайминг</h3><p>Где ячейки оказываются на кристалле и успевает ли тактовая частота.</p><ol><li><a href="/ru/learn/placement/">19 Размещение: где оказываются ячейки</a></li><li><a href="/ru/learn/logic-depth/">20 Глубина логики</a></li><li><a href="/ru/learn/timing-slack/">21 Тайминг и запас</a></li></ol></li><li><h3>Модуль 8 · Битстрим</h3><p>Как оттрассированный дизайн становится битами, которые загружает чип, без Vivado.</p><ol><li><a href="/ru/learn/fasm-to-frames/">22 От FASM к фреймам</a></li><li><a href="/ru/learn/a-build-receipt/">23 Сборка по пунктам</a></li><li><a href="/ru/learn/read-a-bitstream/">24 Читаем битстрим</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Спросите чип, кто он, загрузите биты и проверьте их.</p><ol><li><a href="/ru/learn/jtag-asks-the-chip/">25 JTAG спрашивает чип</a></li><li><a href="/ru/learn/same-bits-on-the-board/">26 Те же биты — на плате</a></li><li><a href="/ru/learn/capstone/">27 Итог: переугадайте yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/how-much-fits/index.html b/apps/website/public/ru/learn/how-much-fits/index.html
index c5a49476df..f858911a79 100644
--- a/apps/website/public/ru/learn/how-much-fits/index.html
+++ b/apps/website/public/ru/learn/how-much-fits/index.html
@@ -441,5 +441,6 @@ <h1>Сколько помещается на чип</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Чип</h3><p>Что такое FPGA, какой чип у нас и как его выводы встречаются с платой.</p><ol><li><a href="/ru/learn/fpga-is-a-grid/">01 FPGA — это сетка, которую вы настраиваете</a></li><li><a href="/ru/learn/which-chip/">02 Какой это чип?</a></li><li><a href="/ru/learn/pins-and-package/">03 Выводы, шарики и корпус</a></li></ol></li><li><h3>Модуль 2 · Числа в железе</h3><p>Биты, триты и форматы чисел, и сколько стоит арифметика в логике.</p><ol><li><a href="/ru/learn/bits-and-trits/">04 Биты, триты и GF16</a></li><li><a href="/ru/learn/what-arithmetic-costs/">05 Сколько стоит арифметика</a></li><li class="on" aria-current="page"><a href="/ru/learn/how-much-fits/">06 Сколько помещается на чип</a></li></ol></li><li><h3>Модуль 3 · Ваша программа на t27</h3><p>Напишите спеку, проверьте её и увидьте, почему «компилируется» не значит «верно».</p><ol><li><a href="/ru/learn/hello-t27/">07 Привет, t27</a></li><li><a href="/ru/learn/tests-are-the-spec/">08 Тесты — часть спеки</a></li><li><a href="/ru/learn/compiles-is-not-correct/">09 Компилируется — не значит верно</a></li></ol></li><li><h3>Модуль 4 · Внутри t27c</h3><p>Как компилятор читает спеку и что он пишет, включая нативный t27b.</p><ol><li><a href="/ru/learn/compiler-stages/">10 Стадии t27c</a></li><li><a href="/ru/learn/t27c-on-a-real-machine/">11 t27c на настоящей машине</a></li><li><a href="/ru/learn/seven-backends-and-t27b/">12 Семь бэкендов и t27b</a></li></ol></li><li><h3>Модуль 5 · От спеки к железу</h3><p>Verilog, который пишет t27c, ячейки, в которые он превращается, и что делает одна LUT.</p><ol><li><a href="/ru/learn/spec-to-verilog/">13 От спеки к Verilog</a></li><li><a href="/ru/learn/count-the-cells/">14 Посчитайте ячейки</a></li><li><a href="/ru/learn/one-lut-up-close/">15 Одна LUT вблизи</a></li></ol></li><li><h3>Модуль 6 · Читаем синтез</h3><p>Что yosys сообщает о вашем дизайне и какие предупреждения важны.</p><ol><li><a href="/ru/learn/synthesis-warnings/">16 Важные предупреждения синтеза</a></li><li><a href="/ru/learn/state-machines/">17 Конечные автоматы</a></li><li><a href="/ru/learn/where-the-luts-go/">18 Куда уходят LUT</a></li></ol></li><li><h3>Модуль 7 · Размещение, трассировка, тайминг</h3><p>Где ячейки оказываются на кристалле и успевает ли тактовая частота.</p><ol><li><a href="/ru/learn/placement/">19 Размещение: где оказываются ячейки</a></li><li><a href="/ru/learn/logic-depth/">20 Глубина логики</a></li><li><a href="/ru/learn/timing-slack/">21 Тайминг и запас</a></li></ol></li><li><h3>Модуль 8 · Битстрим</h3><p>Как оттрассированный дизайн становится битами, которые загружает чип, без Vivado.</p><ol><li><a href="/ru/learn/fasm-to-frames/">22 От FASM к фреймам</a></li><li><a href="/ru/learn/a-build-receipt/">23 Сборка по пунктам</a></li><li><a href="/ru/learn/read-a-bitstream/">24 Читаем битстрим</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Спросите чип, кто он, загрузите биты и проверьте их.</p><ol><li><a href="/ru/learn/jtag-asks-the-chip/">25 JTAG спрашивает чип</a></li><li><a href="/ru/learn/same-bits-on-the-board/">26 Те же биты — на плате</a></li><li><a href="/ru/learn/capstone/">27 Итог: переугадайте yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/if-and-else/index.html b/apps/website/public/ru/learn/if-and-else/index.html
index 4f73196657..ea579f6f4d 100644
--- a/apps/website/public/ru/learn/if-and-else/index.html
+++ b/apps/website/public/ru/learn/if-and-else/index.html
@@ -100,5 +100,6 @@ <h1>Выбор: if, else и switch</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Файл спеки</h3><p>Что лежит в файле .t27: строка module, строки прозы и комментарии.</p><ol><li><a href="/ru/learn/what-a-spec-is/">01 Что такое спека</a></li><li><a href="/ru/learn/the-module-line/">02 Строка module</a></li><li><a href="/ru/learn/comments-and-prose/">03 Комментарии и строки прозы</a></li></ol></li><li><h3>Модуль 2 · Значения и типы</h3><p>Константы, ширина целого числа, а также true, false и текст.</p><ol><li><a href="/ru/learn/constants/">04 Константы</a></li><li><a href="/ru/learn/whole-numbers/">05 Целые числа и их ширина</a></li><li><a href="/ru/learn/true-false-and-text/">06 Истина, ложь и текст</a></li></ol></li><li><h3>Модуль 3 · Массивы, триты и выражения</h3><p>Списки значений, три значения трита и что вычисляют операторы.</p><ol><li><a href="/ru/learn/arrays/">07 Массивы</a></li><li><a href="/ru/learn/trits/">08 Триты: минус один, ноль, один</a></li><li><a href="/ru/learn/expressions/">09 Выражения</a></li></ol></li><li><h3>Модуль 4 · Тесты и функции</h3><p>Блок test, несколько маленьких тестов в одной спеке и функция с типизированными входами.</p><ol><li><a href="/ru/learn/a-test-block/">10 Блок test</a></li><li><a href="/ru/learn/many-tests/">11 Много тестов в одной спеке</a></li><li><a href="/ru/learn/functions/">12 Функции</a></li></ol></li><li><h3>Модуль 5 · Внутри функции</h3><p>Имена через let и var, выбор через if и switch и циклы.</p><ol><li><a href="/ru/learn/local-names/">13 Локальные имена: let и var</a></li><li class="on" aria-current="page"><a href="/ru/learn/if-and-else/">14 Выбор: if, else и switch</a></li><li><a href="/ru/learn/loops/">15 Циклы: while и for</a></li></ol></li><li><h3>Модуль 6 · Видимость и свои типы</h3><p>Что pub открывает другим модулям, и структуры и перечисления, которые вы объявляете.</p><ol><li><a href="/ru/learn/pub-or-private/">16 pub или закрытое имя</a></li><li><a href="/ru/learn/structs/">17 Структуры</a></li><li><a href="/ru/learn/enums/">18 Перечисления</a></li></ol></li><li><h3>Модуль 7 · Модули, правила и gen-ts</h3><p>Чтение другого модуля, правило, которое выполняется всегда, и спека, превращённая в TypeScript.</p><ol><li><a href="/ru/learn/use-other-modules/">19 Другие модули</a></li><li><a href="/ru/learn/invariants/">20 Инварианты и замеры</a></li><li><a href="/ru/learn/gen-ts/">21 gen-ts: спека превращается в TypeScript</a></li></ol></li><li><h3>Модуль 8 · Компилятор и tri</h3><p>Бэкенды, которые пишет t27c, стадии, в которых он читает спеку, и команда tri.</p><ol><li><a href="/ru/learn/seven-backends/">22 Семь бэкендов</a></li><li><a href="/ru/learn/stages-of-t27c/">23 Стадии t27c</a></li><li><a href="/ru/learn/tri-commands/">24 tri: командная строка</a></li></ol></li><li><h3>Модуль 9 · Ошибки, программа и что дальше</h3><p>Как читать ошибку компилятора, одна маленькая законченная программа и какой курс пройти дальше.</p><ol><li><a href="/ru/learn/reading-errors/">25 Как читать ошибки компилятора</a></li><li><a href="/ru/learn/a-small-program/">26 Маленькая законченная программа</a></li><li><a href="/ru/learn/where-next/">27 Куда дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/index.html b/apps/website/public/ru/learn/index.html
index 0d498ffa8c..d918b0e455 100644
--- a/apps/website/public/ru/learn/index.html
+++ b/apps/website/public/ru/learn/index.html
@@ -67,5 +67,6 @@ <h1>Основы t27: язык за 27 уроков</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Файл спеки</h3><p>Что лежит в файле .t27: строка module, строки прозы и комментарии.</p><ol><li><a href="/ru/learn/what-a-spec-is/">01 Что такое спека</a></li><li><a href="/ru/learn/the-module-line/">02 Строка module</a></li><li><a href="/ru/learn/comments-and-prose/">03 Комментарии и строки прозы</a></li></ol></li><li><h3>Модуль 2 · Значения и типы</h3><p>Константы, ширина целого числа, а также true, false и текст.</p><ol><li><a href="/ru/learn/constants/">04 Константы</a></li><li><a href="/ru/learn/whole-numbers/">05 Целые числа и их ширина</a></li><li><a href="/ru/learn/true-false-and-text/">06 Истина, ложь и текст</a></li></ol></li><li><h3>Модуль 3 · Массивы, триты и выражения</h3><p>Списки значений, три значения трита и что вычисляют операторы.</p><ol><li><a href="/ru/learn/arrays/">07 Массивы</a></li><li><a href="/ru/learn/trits/">08 Триты: минус один, ноль, один</a></li><li><a href="/ru/learn/expressions/">09 Выражения</a></li></ol></li><li><h3>Модуль 4 · Тесты и функции</h3><p>Блок test, несколько маленьких тестов в одной спеке и функция с типизированными входами.</p><ol><li><a href="/ru/learn/a-test-block/">10 Блок test</a></li><li><a href="/ru/learn/many-tests/">11 Много тестов в одной спеке</a></li><li><a href="/ru/learn/functions/">12 Функции</a></li></ol></li><li><h3>Модуль 5 · Внутри функции</h3><p>Имена через let и var, выбор через if и switch и циклы.</p><ol><li><a href="/ru/learn/local-names/">13 Локальные имена: let и var</a></li><li><a href="/ru/learn/if-and-else/">14 Выбор: if, else и switch</a></li><li><a href="/ru/learn/loops/">15 Циклы: while и for</a></li></ol></li><li><h3>Модуль 6 · Видимость и свои типы</h3><p>Что pub открывает другим модулям, и структуры и перечисления, которые вы объявляете.</p><ol><li><a href="/ru/learn/pub-or-private/">16 pub или закрытое имя</a></li><li><a href="/ru/learn/structs/">17 Структуры</a></li><li><a href="/ru/learn/enums/">18 Перечисления</a></li></ol></li><li><h3>Модуль 7 · Модули, правила и gen-ts</h3><p>Чтение другого модуля, правило, которое выполняется всегда, и спека, превращённая в TypeScript.</p><ol><li><a href="/ru/learn/use-other-modules/">19 Другие модули</a></li><li><a href="/ru/learn/invariants/">20 Инварианты и замеры</a></li><li><a href="/ru/learn/gen-ts/">21 gen-ts: спека превращается в TypeScript</a></li></ol></li><li><h3>Модуль 8 · Компилятор и tri</h3><p>Бэкенды, которые пишет t27c, стадии, в которых он читает спеку, и команда tri.</p><ol><li><a href="/ru/learn/seven-backends/">22 Семь бэкендов</a></li><li><a href="/ru/learn/stages-of-t27c/">23 Стадии t27c</a></li><li><a href="/ru/learn/tri-commands/">24 tri: командная строка</a></li></ol></li><li><h3>Модуль 9 · Ошибки, программа и что дальше</h3><p>Как читать ошибку компилятора, одна маленькая законченная программа и какой курс пройти дальше.</p><ol><li><a href="/ru/learn/reading-errors/">25 Как читать ошибки компилятора</a></li><li><a href="/ru/learn/a-small-program/">26 Маленькая законченная программа</a></li><li><a href="/ru/learn/where-next/">27 Куда дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/induction/index.html b/apps/website/public/ru/learn/induction/index.html
index c98a2fe340..ed9c85aff3 100644
--- a/apps/website/public/ru/learn/induction/index.html
+++ b/apps/website/public/ru/learn/induction/index.html
@@ -207,5 +207,6 @@ <h1>Индукция</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Зачем проверять</h3><p>Дизайны, которые компилируются и ошибаются; модель, которая выносит вердикт; план, записанный до кода.</p><ol><li><a href="/ru/learn/bugs-that-compile/">01 Баги, которые компилируются</a></li><li><a href="/ru/learn/golden-model/">02 Золотая модель</a></li><li><a href="/ru/learn/plan-before-code/">03 Сначала план</a></li></ol></li><li><h3>Модуль 2 · Тестбенчи</h3><p>Стимулы, проверки и вердикт, записанные как один spec рядом с дизайном, который они судят.</p><ol><li><a href="/ru/learn/anatomy-of-a-tb/">04 Анатомия тестбенча</a></li><li><a href="/ru/learn/self-checking/">05 Самопроверка вместо всматривания</a></li><li><a href="/ru/learn/directed-vs-random/">06 Направленные против случайных</a></li></ol></li><li><h3>Модуль 3 · Временные диаграммы</h3><p>Трасса каждого сигнала, прочитанная так, как её читает инженер по железу, и два прогона, сравнённые между собой.</p><ol><li><a href="/ru/learn/vcd-format/">07 Формат VCD</a></li><li><a href="/ru/learn/debug-with-waves/">08 Отладка по волнам</a></li><li><a href="/ru/learn/compare-traces/">09 Сравнение двух трасс</a></li></ol></li><li><h3>Модуль 4 · Векторы соответствия</h3><p>Случаи с ответом, записанным рядом, там, откуда их достанет компилятор.</p><ol><li><a href="/ru/learn/what-a-vector-is/">10 Что такое вектор</a></li><li><a href="/ru/learn/golden-ruler-vectors/">11 Векторы как эталонная линейка</a></li><li><a href="/ru/learn/end-to-end-scenarios/">12 Сквозные сценарии</a></li></ol></li><li><h3>Модуль 5 · Косимуляция</h3><p>Spec, симулятор и плата сходятся в одном ответе на стенде Artix-7 XC7A200T, и что делать, когда не сходятся.</p><ol><li><a href="/ru/learn/three-way-agreement/">13 Тройное согласие</a></li><li><a href="/ru/learn/simulator-internals/">14 Внутри симулятора</a></li><li><a href="/ru/learn/when-they-disagree/">15 Когда они не сходятся</a></li></ol></li><li><h3>Модуль 6 · Покрытие</h3><p>Чего коснулись тесты: строки, переключения, состояния — и что прячет это число.</p><ol><li><a href="/ru/learn/line-and-toggle/">16 Строки и переключения</a></li><li><a href="/ru/learn/fsm-coverage/">17 Покрытие автоматов</a></li><li><a href="/ru/learn/coverage-lies/">18 Покрытие лжёт</a></li></ol></li><li><h3>Модуль 7 · Формальные методы</h3><p>Ассерты, верные каждый такт; ограниченный поиск контрпримера; и почему доказательству нужна индукция.</p><ol><li><a href="/ru/learn/assertions/">19 Ассерты</a></li><li><a href="/ru/learn/bounded-model-checking/">20 Ограниченная проверка моделей</a></li><li class="on" aria-current="page"><a href="/ru/learn/induction/">21 Индукция</a></li></ol></li><li><h3>Модуль 8 · Мутации</h3><p>Ломайте дизайн нарочно и считайте, что заметили тесты.</p><ol><li><a href="/ru/learn/kill-the-mutant/">22 Убить мутанта</a></li><li><a href="/ru/learn/surviving-mutants/">23 Выжившие мутанты</a></li><li><a href="/ru/learn/test-strength/">24 Сила тестов</a></li></ol></li><li><h3>Модуль 9 · Приёмка</h3><p>Одна команда, все квитанции, чистый вердикт, который можно показать.</p><ol><li><a href="/ru/learn/build-verify/">25 Сборка и проверка</a></li><li><a href="/ru/learn/board-selftest/">26 Стенд проверяет себя</a></li><li><a href="/ru/learn/verify-capstone/">27 Капстоун: проверить модуль целиком</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">Этот курс сам является spec t27: прочитайте его</a></p><p>Виджеты сохраняют свои английские слова: каждый живёт в собственном spec t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/invariants/index.html b/apps/website/public/ru/learn/invariants/index.html
index d5cfc9f936..7fb647f98f 100644
--- a/apps/website/public/ru/learn/invariants/index.html
+++ b/apps/website/public/ru/learn/invariants/index.html
@@ -84,5 +84,6 @@ <h1>Инварианты и замеры</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Файл спеки</h3><p>Что лежит в файле .t27: строка module, строки прозы и комментарии.</p><ol><li><a href="/ru/learn/what-a-spec-is/">01 Что такое спека</a></li><li><a href="/ru/learn/the-module-line/">02 Строка module</a></li><li><a href="/ru/learn/comments-and-prose/">03 Комментарии и строки прозы</a></li></ol></li><li><h3>Модуль 2 · Значения и типы</h3><p>Константы, ширина целого числа, а также true, false и текст.</p><ol><li><a href="/ru/learn/constants/">04 Константы</a></li><li><a href="/ru/learn/whole-numbers/">05 Целые числа и их ширина</a></li><li><a href="/ru/learn/true-false-and-text/">06 Истина, ложь и текст</a></li></ol></li><li><h3>Модуль 3 · Массивы, триты и выражения</h3><p>Списки значений, три значения трита и что вычисляют операторы.</p><ol><li><a href="/ru/learn/arrays/">07 Массивы</a></li><li><a href="/ru/learn/trits/">08 Триты: минус один, ноль, один</a></li><li><a href="/ru/learn/expressions/">09 Выражения</a></li></ol></li><li><h3>Модуль 4 · Тесты и функции</h3><p>Блок test, несколько маленьких тестов в одной спеке и функция с типизированными входами.</p><ol><li><a href="/ru/learn/a-test-block/">10 Блок test</a></li><li><a href="/ru/learn/many-tests/">11 Много тестов в одной спеке</a></li><li><a href="/ru/learn/functions/">12 Функции</a></li></ol></li><li><h3>Модуль 5 · Внутри функции</h3><p>Имена через let и var, выбор через if и switch и циклы.</p><ol><li><a href="/ru/learn/local-names/">13 Локальные имена: let и var</a></li><li><a href="/ru/learn/if-and-else/">14 Выбор: if, else и switch</a></li><li><a href="/ru/learn/loops/">15 Циклы: while и for</a></li></ol></li><li><h3>Модуль 6 · Видимость и свои типы</h3><p>Что pub открывает другим модулям, и структуры и перечисления, которые вы объявляете.</p><ol><li><a href="/ru/learn/pub-or-private/">16 pub или закрытое имя</a></li><li><a href="/ru/learn/structs/">17 Структуры</a></li><li><a href="/ru/learn/enums/">18 Перечисления</a></li></ol></li><li><h3>Модуль 7 · Модули, правила и gen-ts</h3><p>Чтение другого модуля, правило, которое выполняется всегда, и спека, превращённая в TypeScript.</p><ol><li><a href="/ru/learn/use-other-modules/">19 Другие модули</a></li><li class="on" aria-current="page"><a href="/ru/learn/invariants/">20 Инварианты и замеры</a></li><li><a href="/ru/learn/gen-ts/">21 gen-ts: спека превращается в TypeScript</a></li></ol></li><li><h3>Модуль 8 · Компилятор и tri</h3><p>Бэкенды, которые пишет t27c, стадии, в которых он читает спеку, и команда tri.</p><ol><li><a href="/ru/learn/seven-backends/">22 Семь бэкендов</a></li><li><a href="/ru/learn/stages-of-t27c/">23 Стадии t27c</a></li><li><a href="/ru/learn/tri-commands/">24 tri: командная строка</a></li></ol></li><li><h3>Модуль 9 · Ошибки, программа и что дальше</h3><p>Как читать ошибку компилятора, одна маленькая законченная программа и какой курс пройти дальше.</p><ol><li><a href="/ru/learn/reading-errors/">25 Как читать ошибки компилятора</a></li><li><a href="/ru/learn/a-small-program/">26 Маленькая законченная программа</a></li><li><a href="/ru/learn/where-next/">27 Куда дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/io-timing/index.html b/apps/website/public/ru/learn/io-timing/index.html
index 7a0ec57b1e..12751ec25d 100644
--- a/apps/website/public/ru/learn/io-timing/index.html
+++ b/apps/website/public/ru/learn/io-timing/index.html
@@ -364,5 +364,6 @@ <h1>Тайминг выводов</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое тактовый сигнал</h3><p>Один фронт, один мир: всё, что делит фронт тактового сигнала, делит и мир; период и джиттер настоящего фронта и место, где тактовый сигнал входит в плату.</p><ol><li><a href="/ru/learn/one-edge-one-world/">01 Один фронт, один мир</a></li><li><a href="/ru/learn/period-and-jitter/">02 Период, фронт и джиттер</a></li><li><a href="/ru/learn/where-the-clock-enters/">03 Где входит тактовый сигнал</a></li></ol></li><li><h3>Модуль 2 · Деревья тактового сигнала</h3><p>Перекос и задержка прохождения, глобальная сеть буферов и ловушка стробирования клока логикой.</p><ol><li><a href="/ru/learn/skew-and-insertion/">04 Перекос и задержка прохождения</a></li><li><a href="/ru/learn/global-buffers/">05 Глобальный буфер</a></li><li><a href="/ru/learn/gated-clock-trap/">06 Ловушка стробированного клока</a></li></ol></li><li><h3>Модуль 3 · PLL и MMCM</h3><p>Умножить и поделить один тактовый сигнал в другой, сдвинуть его фазу шагами VCO и какие тактовые сигналы анализатор считает родственными.</p><ol><li><a href="/ru/learn/multiply-and-divide/">07 Умножить и поделить</a></li><li><a href="/ru/learn/phase-and-lock/">08 Фаза и захват</a></li><li><a href="/ru/learn/related-or-not/">09 Родственные или чужие</a></li></ol></li><li><h3>Модуль 4 · Сбросы</h3><p>Ассертировать асинхронно, снимать синхронно: три вида сброса, конвейер снятия и дерево, которое растит сброс.</p><ol><li><a href="/ru/learn/sync-or-async/">10 Синхронный или асинхронный</a></li><li><a href="/ru/learn/reset-release/">11 Конвейер снятия</a></li><li><a href="/ru/learn/reset-fanout/">12 Разветвление сброса</a></li></ol></li><li><h3>Модуль 5 · Метастабильность</h3><p>Окно setup-hold, среднее время между отказами в целочисленной арифметике и два триггера, которые это чинят.</p><ol><li><a href="/ru/learn/setup-hold-window/">13 Окно setup-hold</a></li><li><a href="/ru/learn/mtbf/">14 Среднее время между отказами</a></li><li><a href="/ru/learn/two-flop-sync/">15 Двухтриггерный синхронизатор</a></li></ol></li><li><h3>Модуль 6 · Пересечение многих бит</h3><p>Почему двоичная шина рвётся, почему код Грея нет, и рукопожатие, переносящее импульс между мирами.</p><ol><li><a href="/ru/learn/why-binary-fails/">16 Почему двоичный код рвётся</a></li><li><a href="/ru/learn/gray-code/">17 Код Грея</a></li><li><a href="/ru/learn/handshakes-and-pulses/">18 Рукопожатия и импульсы</a></li></ol></li><li><h3>Модуль 7 · Асинхронный FIFO</h3><p>Указатели, флаги и глубина: буфер, переносящий поток между двумя тактовыми сигналами.</p><ol><li><a href="/ru/learn/pointers-and-flags/">19 Указатели и флаги</a></li><li><a href="/ru/learn/depth-sizing/">20 Выбор глубины</a></li><li><a href="/ru/learn/fifo-testbench/">21 Тестбенч FIFO</a></li></ol></li><li><h3>Модуль 8 · Ограничения</h3><p>Строки, сообщающие анализатору, что такое тактовый сигнал, какие пути не проверять и чему должны соответствовать выводы.</p><ol><li><a href="/ru/learn/create-clock/">22 Строка create_clock</a></li><li><a href="/ru/learn/false-path-and-max-delay/">23 False path и max delay</a></li><li class="on" aria-current="page"><a href="/ru/learn/io-timing/">24 Тайминг выводов</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Отчёт CDC, одно пересечение, захваченное на триггерах, и дифф битстрима, замыкающий курс.</p><ol><li><a href="/ru/learn/cdc-report/">25 Отчёт CDC</a></li><li><a href="/ru/learn/capture-a-crossing/">26 Захватить пересечение</a></li><li><a href="/ru/learn/capstone-diff/">27 Капстоун: поменяй и сравни</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/jtag-asks-the-chip/index.html b/apps/website/public/ru/learn/jtag-asks-the-chip/index.html
index 8db9c573ca..04c05d46e9 100644
--- a/apps/website/public/ru/learn/jtag-asks-the-chip/index.html
+++ b/apps/website/public/ru/learn/jtag-asks-the-chip/index.html
@@ -110,5 +110,6 @@ <h1>JTAG спрашивает чип</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Чип</h3><p>Что такое FPGA, какой чип у нас и как его выводы встречаются с платой.</p><ol><li><a href="/ru/learn/fpga-is-a-grid/">01 FPGA — это сетка, которую вы настраиваете</a></li><li><a href="/ru/learn/which-chip/">02 Какой это чип?</a></li><li><a href="/ru/learn/pins-and-package/">03 Выводы, шарики и корпус</a></li></ol></li><li><h3>Модуль 2 · Числа в железе</h3><p>Биты, триты и форматы чисел, и сколько стоит арифметика в логике.</p><ol><li><a href="/ru/learn/bits-and-trits/">04 Биты, триты и GF16</a></li><li><a href="/ru/learn/what-arithmetic-costs/">05 Сколько стоит арифметика</a></li><li><a href="/ru/learn/how-much-fits/">06 Сколько помещается на чип</a></li></ol></li><li><h3>Модуль 3 · Ваша программа на t27</h3><p>Напишите спеку, проверьте её и увидьте, почему «компилируется» не значит «верно».</p><ol><li><a href="/ru/learn/hello-t27/">07 Привет, t27</a></li><li><a href="/ru/learn/tests-are-the-spec/">08 Тесты — часть спеки</a></li><li><a href="/ru/learn/compiles-is-not-correct/">09 Компилируется — не значит верно</a></li></ol></li><li><h3>Модуль 4 · Внутри t27c</h3><p>Как компилятор читает спеку и что он пишет, включая нативный t27b.</p><ol><li><a href="/ru/learn/compiler-stages/">10 Стадии t27c</a></li><li><a href="/ru/learn/t27c-on-a-real-machine/">11 t27c на настоящей машине</a></li><li><a href="/ru/learn/seven-backends-and-t27b/">12 Семь бэкендов и t27b</a></li></ol></li><li><h3>Модуль 5 · От спеки к железу</h3><p>Verilog, который пишет t27c, ячейки, в которые он превращается, и что делает одна LUT.</p><ol><li><a href="/ru/learn/spec-to-verilog/">13 От спеки к Verilog</a></li><li><a href="/ru/learn/count-the-cells/">14 Посчитайте ячейки</a></li><li><a href="/ru/learn/one-lut-up-close/">15 Одна LUT вблизи</a></li></ol></li><li><h3>Модуль 6 · Читаем синтез</h3><p>Что yosys сообщает о вашем дизайне и какие предупреждения важны.</p><ol><li><a href="/ru/learn/synthesis-warnings/">16 Важные предупреждения синтеза</a></li><li><a href="/ru/learn/state-machines/">17 Конечные автоматы</a></li><li><a href="/ru/learn/where-the-luts-go/">18 Куда уходят LUT</a></li></ol></li><li><h3>Модуль 7 · Размещение, трассировка, тайминг</h3><p>Где ячейки оказываются на кристалле и успевает ли тактовая частота.</p><ol><li><a href="/ru/learn/placement/">19 Размещение: где оказываются ячейки</a></li><li><a href="/ru/learn/logic-depth/">20 Глубина логики</a></li><li><a href="/ru/learn/timing-slack/">21 Тайминг и запас</a></li></ol></li><li><h3>Модуль 8 · Битстрим</h3><p>Как оттрассированный дизайн становится битами, которые загружает чип, без Vivado.</p><ol><li><a href="/ru/learn/fasm-to-frames/">22 От FASM к фреймам</a></li><li><a href="/ru/learn/a-build-receipt/">23 Сборка по пунктам</a></li><li><a href="/ru/learn/read-a-bitstream/">24 Читаем битстрим</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Спросите чип, кто он, загрузите биты и проверьте их.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/jtag-asks-the-chip/">25 JTAG спрашивает чип</a></li><li><a href="/ru/learn/same-bits-on-the-board/">26 Те же биты — на плате</a></li><li><a href="/ru/learn/capstone/">27 Итог: переугадайте yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/kill-the-mutant/index.html b/apps/website/public/ru/learn/kill-the-mutant/index.html
index 9f794f21ea..76ef397392 100644
--- a/apps/website/public/ru/learn/kill-the-mutant/index.html
+++ b/apps/website/public/ru/learn/kill-the-mutant/index.html
@@ -110,5 +110,6 @@ <h1>Убить мутанта</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Зачем проверять</h3><p>Дизайны, которые компилируются и ошибаются; модель, которая выносит вердикт; план, записанный до кода.</p><ol><li><a href="/ru/learn/bugs-that-compile/">01 Баги, которые компилируются</a></li><li><a href="/ru/learn/golden-model/">02 Золотая модель</a></li><li><a href="/ru/learn/plan-before-code/">03 Сначала план</a></li></ol></li><li><h3>Модуль 2 · Тестбенчи</h3><p>Стимулы, проверки и вердикт, записанные как один spec рядом с дизайном, который они судят.</p><ol><li><a href="/ru/learn/anatomy-of-a-tb/">04 Анатомия тестбенча</a></li><li><a href="/ru/learn/self-checking/">05 Самопроверка вместо всматривания</a></li><li><a href="/ru/learn/directed-vs-random/">06 Направленные против случайных</a></li></ol></li><li><h3>Модуль 3 · Временные диаграммы</h3><p>Трасса каждого сигнала, прочитанная так, как её читает инженер по железу, и два прогона, сравнённые между собой.</p><ol><li><a href="/ru/learn/vcd-format/">07 Формат VCD</a></li><li><a href="/ru/learn/debug-with-waves/">08 Отладка по волнам</a></li><li><a href="/ru/learn/compare-traces/">09 Сравнение двух трасс</a></li></ol></li><li><h3>Модуль 4 · Векторы соответствия</h3><p>Случаи с ответом, записанным рядом, там, откуда их достанет компилятор.</p><ol><li><a href="/ru/learn/what-a-vector-is/">10 Что такое вектор</a></li><li><a href="/ru/learn/golden-ruler-vectors/">11 Векторы как эталонная линейка</a></li><li><a href="/ru/learn/end-to-end-scenarios/">12 Сквозные сценарии</a></li></ol></li><li><h3>Модуль 5 · Косимуляция</h3><p>Spec, симулятор и плата сходятся в одном ответе на стенде Artix-7 XC7A200T, и что делать, когда не сходятся.</p><ol><li><a href="/ru/learn/three-way-agreement/">13 Тройное согласие</a></li><li><a href="/ru/learn/simulator-internals/">14 Внутри симулятора</a></li><li><a href="/ru/learn/when-they-disagree/">15 Когда они не сходятся</a></li></ol></li><li><h3>Модуль 6 · Покрытие</h3><p>Чего коснулись тесты: строки, переключения, состояния — и что прячет это число.</p><ol><li><a href="/ru/learn/line-and-toggle/">16 Строки и переключения</a></li><li><a href="/ru/learn/fsm-coverage/">17 Покрытие автоматов</a></li><li><a href="/ru/learn/coverage-lies/">18 Покрытие лжёт</a></li></ol></li><li><h3>Модуль 7 · Формальные методы</h3><p>Ассерты, верные каждый такт; ограниченный поиск контрпримера; и почему доказательству нужна индукция.</p><ol><li><a href="/ru/learn/assertions/">19 Ассерты</a></li><li><a href="/ru/learn/bounded-model-checking/">20 Ограниченная проверка моделей</a></li><li><a href="/ru/learn/induction/">21 Индукция</a></li></ol></li><li><h3>Модуль 8 · Мутации</h3><p>Ломайте дизайн нарочно и считайте, что заметили тесты.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/kill-the-mutant/">22 Убить мутанта</a></li><li><a href="/ru/learn/surviving-mutants/">23 Выжившие мутанты</a></li><li><a href="/ru/learn/test-strength/">24 Сила тестов</a></li></ol></li><li><h3>Модуль 9 · Приёмка</h3><p>Одна команда, все квитанции, чистый вердикт, который можно показать.</p><ol><li><a href="/ru/learn/build-verify/">25 Сборка и проверка</a></li><li><a href="/ru/learn/board-selftest/">26 Стенд проверяет себя</a></li><li><a href="/ru/learn/verify-capstone/">27 Капстоун: проверить модуль целиком</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">Этот курс сам является spec t27: прочитайте его</a></p><p>Виджеты сохраняют свои английские слова: каждый живёт в собственном spec t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/kleene-logic/index.html b/apps/website/public/ru/learn/kleene-logic/index.html
index e55c6dceeb..662aa22741 100644
--- a/apps/website/public/ru/learn/kleene-logic/index.html
+++ b/apps/website/public/ru/learn/kleene-logic/index.html
@@ -769,5 +769,6 @@ <h1>Трёхзначная логика</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Три значения</h3><p>Почему три, как сбалансированная троичная система пишет любое число без знака и что дают переворот и отсечение тритов.</p><ol><li><a href="/ru/learn/why-three/">01 Почему три</a></li><li><a href="/ru/learn/balanced-ternary/">02 Сбалансированная троичная</a></li><li><a href="/ru/learn/flip-and-cut/">03 Переворот вместо минуса, отсечение вместо округления</a></li></ol></li><li><h3>Модуль 2 · Логика с «не знаю»</h3><p>Трёхзначные вентили Клини, два вентиля без двоичного двойника и сложение как пара таблиц.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/kleene-logic/">04 Трёхзначная логика</a></li><li><a href="/ru/learn/consensus-and-any/">05 Консенсус и «любой»</a></li><li><a href="/ru/learn/sum-and-carry/">06 Сумма и перенос</a></li></ol></li><li><h3>Модуль 3 · Складываем триты</h3><p>Полусумматор, полный сумматор и перенос, бегущий влево разряд за разрядом.</p><ol><li><a href="/ru/learn/half-adder/">07 Полусумматор</a></li><li><a href="/ru/learn/full-adder/">08 Полный сумматор</a></li><li><a href="/ru/learn/ripple-carry/">09 Сквозной перенос</a></li></ol></li><li><h3>Модуль 4 · Умножение без умножителя</h3><p>Копия, ноль или переворот: произведение на один трит, умножение столбиком и MAC, который только складывает.</p><ol><li><a href="/ru/learn/times-a-trit/">10 Умножить на один трит</a></li><li><a href="/ru/learn/long-multiplication/">11 Умножение столбиком</a></li><li><a href="/ru/learn/ternary-mac/">12 Троичный MAC</a></li></ol></li><li><h3>Модуль 5 · Триты в двоичной памяти</h3><p>Два бита на трит, пять тритов на байт и сколько бит нужно слову из 27 тритов.</p><ol><li><a href="/ru/learn/two-bits-per-trit/">13 Два бита на трит</a></li><li><a href="/ru/learn/five-trits-a-byte/">14 Пять тритов в байте</a></li><li><a href="/ru/learn/a-word-of-27/">15 Слово из 27 тритов</a></li></ol></li><li><h3>Модуль 6 · Командное слово TRI-27</h3><p>32-битное слово, которое декодирует эмулятор Trinity: его поля, 47 опкодов и 15-битное непосредственное значение.</p><ol><li><a href="/ru/learn/the-instruction-word/">16 Командное слово</a></li><li><a href="/ru/learn/the-opcode-map/">17 Карта опкодов</a></li><li><a href="/ru/learn/the-immediate/">18 Непосредственное значение</a></li></ol></li><li><h3>Модуль 7 · Троичная машина</h3><p>АЛУ из сумматоров этого курса, переход на три стороны и программа, которую можно пройти по шагам.</p><ol><li><a href="/ru/learn/a-ternary-alu/">19 Троичное АЛУ</a></li><li><a href="/ru/learn/three-way-jump/">20 Переход на три стороны</a></li><li><a href="/ru/learn/run-a-program/">21 Запустить программу</a></li></ol></li><li><h3>Модуль 8 · Три ответа</h3><p>Сравнение по старшему различному триту, поиск числа по третям и сортировка трёхзначными сравнениями.</p><ol><li><a href="/ru/learn/compare-at-the-top/">22 Сравнение сверху</a></li><li><a href="/ru/learn/questions-in-thirds/">23 Вопросы по третям</a></li><li><a href="/ru/learn/sort-by-three/">24 Сортировка трёхзначными сравнениями</a></li></ol></li><li><h3>Модуль 9 · Троичные нейроны</h3><p>Веса -1, 0, +1: один нейрон, детектор и маленький слой, и нигде нет умножителя.</p><ol><li><a href="/ru/learn/one-ternary-neuron/">25 Один троичный нейрон</a></li><li><a href="/ru/learn/a-bar-detector/">26 Детектор полосы</a></li><li><a href="/ru/learn/a-tiny-layer/">27 Маленький слой</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/latency/index.html b/apps/website/public/ru/learn/latency/index.html
index 9925bdad31..1d4ea37fc5 100644
--- a/apps/website/public/ru/learn/latency/index.html
+++ b/apps/website/public/ru/learn/latency/index.html
@@ -238,5 +238,6 @@ <h1>Задержка</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое шина</h3><p>Зачем вообще существует шина: разговор по проводам, с кадрами и адресами, и кому позволено говорить.</p><ol><li><a href="/ru/learn/wires-that-talk/">01 Провода, которые говорят</a></li><li><a href="/ru/learn/framing-a-conversation/">02 Кадрирование разговора</a></li><li><a href="/ru/learn/master-and-servant/">03 Хозяин и слуга</a></li></ol></li><li><h3>Модуль 2 · Разговор по UART</h3><p>Двухпроводная шина без такта: кадр, делитель, задающий скорость, и статус, который опрашивает драйвер.</p><ol><li><a href="/ru/learn/the-frame/">04 Кадр: старт, данные, стоп</a></li><li><a href="/ru/learn/baud-and-the-divisor/">05 Бод и делитель</a></li><li><a href="/ru/learn/status-and-fifos/">06 Статус и FIFO</a></li></ol></li><li><h3>Модуль 3 · SPI под тактом</h3><p>Разговор под тактом: четыре режима, лестница предделителей и выбор кристалла на каждого подчинённого.</p><ol><li><a href="/ru/learn/modes-cpol-cpha/">07 Режимы: CPOL и CPHA</a></li><li><a href="/ru/learn/the-prescaler-ladder/">08 Лестница предделителей</a></li><li><a href="/ru/learn/chip-select-and-widths/">09 Выбор кристалла и ширины</a></li></ol></li><li><h3>Модуль 4 · Регистровая шина APB</h3><p>Шина регистров: PSEL и PENABLE, строобы и ожидания, и сколько адресных бит стоит количество периферии.</p><ol><li><a href="/ru/learn/psel-and-penable/">10 PSEL и PENABLE</a></li><li><a href="/ru/learn/strobes-and-waits/">11 Строобы и ожидания</a></li><li><a href="/ru/learn/address-decode/">12 Дешифрация адреса</a></li></ol></li><li><h3>Модуль 5 · Пять каналов AXI4</h3><p>Пять каналов: адрес, данные и отклик в обе стороны, lite или full, пакеты и идентификаторы.</p><ol><li><a href="/ru/learn/five-channels/">13 Пять каналов</a></li><li><a href="/ru/learn/lite-or-full/">14 Lite или full</a></li><li><a href="/ru/learn/bursts-and-ids/">15 Пакеты и идентификаторы</a></li></ol></li><li><h3>Модуль 6 · Память</h3><p>Что находится на дальнем конце каждой шины: карты памяти, типы портов и задержка, которую должно покрывать ожидание.</p><ol><li><a href="/ru/learn/memory-maps/">16 Карты памяти</a></li><li><a href="/ru/learn/ports-and-kinds/">17 Порты и типы</a></li><li class="on" aria-current="page"><a href="/ru/learn/latency/">18 Задержка</a></li></ol></li><li><h3>Модуль 7 · Мосты</h3><p>Почему в дизайне появляется больше одной шины и пакетный мост, переносящий работу между ними.</p><ol><li><a href="/ru/learn/why-bridges/">19 Зачем мосты</a></li><li><a href="/ru/learn/the-packet-bridge/">20 Пакетный мост</a></li><li><a href="/ru/learn/everything-in-one-bench/">21 Всё в одном стенде</a></li></ol></li><li><h3>Модуль 8 · Ethernet: кадры и тайминг</h3><p>Кадры, контрольная сумма кадра, тайминг RGMII и заранее зарегистрированные шаги настоящего запуска.</p><ol><li><a href="/ru/learn/frames-and-the-fcs/">22 Кадры и контрольная сумма</a></li><li><a href="/ru/learn/rgmii-timing/">23 Тайминг RGMII</a></li><li><a href="/ru/learn/bring-up-on-the-bench/">24 Запуск на стенде</a></li></ol></li><li><h3>Модуль 9 · Стенд</h3><p>Дисциплина, оберегающая настоящее железо: кто держит IO, как взять и вернуть захват, и что запускается дальше.</p><ol><li><a href="/ru/learn/who-holds-the-io/">25 Кто держит IO</a></li><li><a href="/ru/learn/claim-and-release/">26 Захват и возврат</a></li><li><a href="/ru/learn/capstone-what-runs-next/">27 Капстоун: что запускается дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/line-and-toggle/index.html b/apps/website/public/ru/learn/line-and-toggle/index.html
index 1fff7126d5..c2a58c3594 100644
--- a/apps/website/public/ru/learn/line-and-toggle/index.html
+++ b/apps/website/public/ru/learn/line-and-toggle/index.html
@@ -576,5 +576,6 @@ <h1>Строки и переключения</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Зачем проверять</h3><p>Дизайны, которые компилируются и ошибаются; модель, которая выносит вердикт; план, записанный до кода.</p><ol><li><a href="/ru/learn/bugs-that-compile/">01 Баги, которые компилируются</a></li><li><a href="/ru/learn/golden-model/">02 Золотая модель</a></li><li><a href="/ru/learn/plan-before-code/">03 Сначала план</a></li></ol></li><li><h3>Модуль 2 · Тестбенчи</h3><p>Стимулы, проверки и вердикт, записанные как один spec рядом с дизайном, который они судят.</p><ol><li><a href="/ru/learn/anatomy-of-a-tb/">04 Анатомия тестбенча</a></li><li><a href="/ru/learn/self-checking/">05 Самопроверка вместо всматривания</a></li><li><a href="/ru/learn/directed-vs-random/">06 Направленные против случайных</a></li></ol></li><li><h3>Модуль 3 · Временные диаграммы</h3><p>Трасса каждого сигнала, прочитанная так, как её читает инженер по железу, и два прогона, сравнённые между собой.</p><ol><li><a href="/ru/learn/vcd-format/">07 Формат VCD</a></li><li><a href="/ru/learn/debug-with-waves/">08 Отладка по волнам</a></li><li><a href="/ru/learn/compare-traces/">09 Сравнение двух трасс</a></li></ol></li><li><h3>Модуль 4 · Векторы соответствия</h3><p>Случаи с ответом, записанным рядом, там, откуда их достанет компилятор.</p><ol><li><a href="/ru/learn/what-a-vector-is/">10 Что такое вектор</a></li><li><a href="/ru/learn/golden-ruler-vectors/">11 Векторы как эталонная линейка</a></li><li><a href="/ru/learn/end-to-end-scenarios/">12 Сквозные сценарии</a></li></ol></li><li><h3>Модуль 5 · Косимуляция</h3><p>Spec, симулятор и плата сходятся в одном ответе на стенде Artix-7 XC7A200T, и что делать, когда не сходятся.</p><ol><li><a href="/ru/learn/three-way-agreement/">13 Тройное согласие</a></li><li><a href="/ru/learn/simulator-internals/">14 Внутри симулятора</a></li><li><a href="/ru/learn/when-they-disagree/">15 Когда они не сходятся</a></li></ol></li><li><h3>Модуль 6 · Покрытие</h3><p>Чего коснулись тесты: строки, переключения, состояния — и что прячет это число.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/line-and-toggle/">16 Строки и переключения</a></li><li><a href="/ru/learn/fsm-coverage/">17 Покрытие автоматов</a></li><li><a href="/ru/learn/coverage-lies/">18 Покрытие лжёт</a></li></ol></li><li><h3>Модуль 7 · Формальные методы</h3><p>Ассерты, верные каждый такт; ограниченный поиск контрпримера; и почему доказательству нужна индукция.</p><ol><li><a href="/ru/learn/assertions/">19 Ассерты</a></li><li><a href="/ru/learn/bounded-model-checking/">20 Ограниченная проверка моделей</a></li><li><a href="/ru/learn/induction/">21 Индукция</a></li></ol></li><li><h3>Модуль 8 · Мутации</h3><p>Ломайте дизайн нарочно и считайте, что заметили тесты.</p><ol><li><a href="/ru/learn/kill-the-mutant/">22 Убить мутанта</a></li><li><a href="/ru/learn/surviving-mutants/">23 Выжившие мутанты</a></li><li><a href="/ru/learn/test-strength/">24 Сила тестов</a></li></ol></li><li><h3>Модуль 9 · Приёмка</h3><p>Одна команда, все квитанции, чистый вердикт, который можно показать.</p><ol><li><a href="/ru/learn/build-verify/">25 Сборка и проверка</a></li><li><a href="/ru/learn/board-selftest/">26 Стенд проверяет себя</a></li><li><a href="/ru/learn/verify-capstone/">27 Капстоун: проверить модуль целиком</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">Этот курс сам является spec t27: прочитайте его</a></p><p>Виджеты сохраняют свои английские слова: каждый живёт в собственном spec t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/lite-or-full/index.html b/apps/website/public/ru/learn/lite-or-full/index.html
index e963f69328..2395d87141 100644
--- a/apps/website/public/ru/learn/lite-or-full/index.html
+++ b/apps/website/public/ru/learn/lite-or-full/index.html
@@ -463,5 +463,6 @@ <h1>Lite или full</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое шина</h3><p>Зачем вообще существует шина: разговор по проводам, с кадрами и адресами, и кому позволено говорить.</p><ol><li><a href="/ru/learn/wires-that-talk/">01 Провода, которые говорят</a></li><li><a href="/ru/learn/framing-a-conversation/">02 Кадрирование разговора</a></li><li><a href="/ru/learn/master-and-servant/">03 Хозяин и слуга</a></li></ol></li><li><h3>Модуль 2 · Разговор по UART</h3><p>Двухпроводная шина без такта: кадр, делитель, задающий скорость, и статус, который опрашивает драйвер.</p><ol><li><a href="/ru/learn/the-frame/">04 Кадр: старт, данные, стоп</a></li><li><a href="/ru/learn/baud-and-the-divisor/">05 Бод и делитель</a></li><li><a href="/ru/learn/status-and-fifos/">06 Статус и FIFO</a></li></ol></li><li><h3>Модуль 3 · SPI под тактом</h3><p>Разговор под тактом: четыре режима, лестница предделителей и выбор кристалла на каждого подчинённого.</p><ol><li><a href="/ru/learn/modes-cpol-cpha/">07 Режимы: CPOL и CPHA</a></li><li><a href="/ru/learn/the-prescaler-ladder/">08 Лестница предделителей</a></li><li><a href="/ru/learn/chip-select-and-widths/">09 Выбор кристалла и ширины</a></li></ol></li><li><h3>Модуль 4 · Регистровая шина APB</h3><p>Шина регистров: PSEL и PENABLE, строобы и ожидания, и сколько адресных бит стоит количество периферии.</p><ol><li><a href="/ru/learn/psel-and-penable/">10 PSEL и PENABLE</a></li><li><a href="/ru/learn/strobes-and-waits/">11 Строобы и ожидания</a></li><li><a href="/ru/learn/address-decode/">12 Дешифрация адреса</a></li></ol></li><li><h3>Модуль 5 · Пять каналов AXI4</h3><p>Пять каналов: адрес, данные и отклик в обе стороны, lite или full, пакеты и идентификаторы.</p><ol><li><a href="/ru/learn/five-channels/">13 Пять каналов</a></li><li class="on" aria-current="page"><a href="/ru/learn/lite-or-full/">14 Lite или full</a></li><li><a href="/ru/learn/bursts-and-ids/">15 Пакеты и идентификаторы</a></li></ol></li><li><h3>Модуль 6 · Память</h3><p>Что находится на дальнем конце каждой шины: карты памяти, типы портов и задержка, которую должно покрывать ожидание.</p><ol><li><a href="/ru/learn/memory-maps/">16 Карты памяти</a></li><li><a href="/ru/learn/ports-and-kinds/">17 Порты и типы</a></li><li><a href="/ru/learn/latency/">18 Задержка</a></li></ol></li><li><h3>Модуль 7 · Мосты</h3><p>Почему в дизайне появляется больше одной шины и пакетный мост, переносящий работу между ними.</p><ol><li><a href="/ru/learn/why-bridges/">19 Зачем мосты</a></li><li><a href="/ru/learn/the-packet-bridge/">20 Пакетный мост</a></li><li><a href="/ru/learn/everything-in-one-bench/">21 Всё в одном стенде</a></li></ol></li><li><h3>Модуль 8 · Ethernet: кадры и тайминг</h3><p>Кадры, контрольная сумма кадра, тайминг RGMII и заранее зарегистрированные шаги настоящего запуска.</p><ol><li><a href="/ru/learn/frames-and-the-fcs/">22 Кадры и контрольная сумма</a></li><li><a href="/ru/learn/rgmii-timing/">23 Тайминг RGMII</a></li><li><a href="/ru/learn/bring-up-on-the-bench/">24 Запуск на стенде</a></li></ol></li><li><h3>Модуль 9 · Стенд</h3><p>Дисциплина, оберегающая настоящее железо: кто держит IO, как взять и вернуть захват, и что запускается дальше.</p><ol><li><a href="/ru/learn/who-holds-the-io/">25 Кто держит IO</a></li><li><a href="/ru/learn/claim-and-release/">26 Захват и возврат</a></li><li><a href="/ru/learn/capstone-what-runs-next/">27 Капстоун: что запускается дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/local-names/index.html b/apps/website/public/ru/learn/local-names/index.html
index 0853753675..2f95789526 100644
--- a/apps/website/public/ru/learn/local-names/index.html
+++ b/apps/website/public/ru/learn/local-names/index.html
@@ -94,5 +94,6 @@ <h1>Локальные имена: let и var</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Файл спеки</h3><p>Что лежит в файле .t27: строка module, строки прозы и комментарии.</p><ol><li><a href="/ru/learn/what-a-spec-is/">01 Что такое спека</a></li><li><a href="/ru/learn/the-module-line/">02 Строка module</a></li><li><a href="/ru/learn/comments-and-prose/">03 Комментарии и строки прозы</a></li></ol></li><li><h3>Модуль 2 · Значения и типы</h3><p>Константы, ширина целого числа, а также true, false и текст.</p><ol><li><a href="/ru/learn/constants/">04 Константы</a></li><li><a href="/ru/learn/whole-numbers/">05 Целые числа и их ширина</a></li><li><a href="/ru/learn/true-false-and-text/">06 Истина, ложь и текст</a></li></ol></li><li><h3>Модуль 3 · Массивы, триты и выражения</h3><p>Списки значений, три значения трита и что вычисляют операторы.</p><ol><li><a href="/ru/learn/arrays/">07 Массивы</a></li><li><a href="/ru/learn/trits/">08 Триты: минус один, ноль, один</a></li><li><a href="/ru/learn/expressions/">09 Выражения</a></li></ol></li><li><h3>Модуль 4 · Тесты и функции</h3><p>Блок test, несколько маленьких тестов в одной спеке и функция с типизированными входами.</p><ol><li><a href="/ru/learn/a-test-block/">10 Блок test</a></li><li><a href="/ru/learn/many-tests/">11 Много тестов в одной спеке</a></li><li><a href="/ru/learn/functions/">12 Функции</a></li></ol></li><li><h3>Модуль 5 · Внутри функции</h3><p>Имена через let и var, выбор через if и switch и циклы.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/local-names/">13 Локальные имена: let и var</a></li><li><a href="/ru/learn/if-and-else/">14 Выбор: if, else и switch</a></li><li><a href="/ru/learn/loops/">15 Циклы: while и for</a></li></ol></li><li><h3>Модуль 6 · Видимость и свои типы</h3><p>Что pub открывает другим модулям, и структуры и перечисления, которые вы объявляете.</p><ol><li><a href="/ru/learn/pub-or-private/">16 pub или закрытое имя</a></li><li><a href="/ru/learn/structs/">17 Структуры</a></li><li><a href="/ru/learn/enums/">18 Перечисления</a></li></ol></li><li><h3>Модуль 7 · Модули, правила и gen-ts</h3><p>Чтение другого модуля, правило, которое выполняется всегда, и спека, превращённая в TypeScript.</p><ol><li><a href="/ru/learn/use-other-modules/">19 Другие модули</a></li><li><a href="/ru/learn/invariants/">20 Инварианты и замеры</a></li><li><a href="/ru/learn/gen-ts/">21 gen-ts: спека превращается в TypeScript</a></li></ol></li><li><h3>Модуль 8 · Компилятор и tri</h3><p>Бэкенды, которые пишет t27c, стадии, в которых он читает спеку, и команда tri.</p><ol><li><a href="/ru/learn/seven-backends/">22 Семь бэкендов</a></li><li><a href="/ru/learn/stages-of-t27c/">23 Стадии t27c</a></li><li><a href="/ru/learn/tri-commands/">24 tri: командная строка</a></li></ol></li><li><h3>Модуль 9 · Ошибки, программа и что дальше</h3><p>Как читать ошибку компилятора, одна маленькая законченная программа и какой курс пройти дальше.</p><ol><li><a href="/ru/learn/reading-errors/">25 Как читать ошибки компилятора</a></li><li><a href="/ru/learn/a-small-program/">26 Маленькая законченная программа</a></li><li><a href="/ru/learn/where-next/">27 Куда дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/logic-depth/index.html b/apps/website/public/ru/learn/logic-depth/index.html
index 39a0ce9c73..9411b9100e 100644
--- a/apps/website/public/ru/learn/logic-depth/index.html
+++ b/apps/website/public/ru/learn/logic-depth/index.html
@@ -448,5 +448,6 @@ <h1>Глубина логики</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Чип</h3><p>Что такое FPGA, какой чип у нас и как его выводы встречаются с платой.</p><ol><li><a href="/ru/learn/fpga-is-a-grid/">01 FPGA — это сетка, которую вы настраиваете</a></li><li><a href="/ru/learn/which-chip/">02 Какой это чип?</a></li><li><a href="/ru/learn/pins-and-package/">03 Выводы, шарики и корпус</a></li></ol></li><li><h3>Модуль 2 · Числа в железе</h3><p>Биты, триты и форматы чисел, и сколько стоит арифметика в логике.</p><ol><li><a href="/ru/learn/bits-and-trits/">04 Биты, триты и GF16</a></li><li><a href="/ru/learn/what-arithmetic-costs/">05 Сколько стоит арифметика</a></li><li><a href="/ru/learn/how-much-fits/">06 Сколько помещается на чип</a></li></ol></li><li><h3>Модуль 3 · Ваша программа на t27</h3><p>Напишите спеку, проверьте её и увидьте, почему «компилируется» не значит «верно».</p><ol><li><a href="/ru/learn/hello-t27/">07 Привет, t27</a></li><li><a href="/ru/learn/tests-are-the-spec/">08 Тесты — часть спеки</a></li><li><a href="/ru/learn/compiles-is-not-correct/">09 Компилируется — не значит верно</a></li></ol></li><li><h3>Модуль 4 · Внутри t27c</h3><p>Как компилятор читает спеку и что он пишет, включая нативный t27b.</p><ol><li><a href="/ru/learn/compiler-stages/">10 Стадии t27c</a></li><li><a href="/ru/learn/t27c-on-a-real-machine/">11 t27c на настоящей машине</a></li><li><a href="/ru/learn/seven-backends-and-t27b/">12 Семь бэкендов и t27b</a></li></ol></li><li><h3>Модуль 5 · От спеки к железу</h3><p>Verilog, который пишет t27c, ячейки, в которые он превращается, и что делает одна LUT.</p><ol><li><a href="/ru/learn/spec-to-verilog/">13 От спеки к Verilog</a></li><li><a href="/ru/learn/count-the-cells/">14 Посчитайте ячейки</a></li><li><a href="/ru/learn/one-lut-up-close/">15 Одна LUT вблизи</a></li></ol></li><li><h3>Модуль 6 · Читаем синтез</h3><p>Что yosys сообщает о вашем дизайне и какие предупреждения важны.</p><ol><li><a href="/ru/learn/synthesis-warnings/">16 Важные предупреждения синтеза</a></li><li><a href="/ru/learn/state-machines/">17 Конечные автоматы</a></li><li><a href="/ru/learn/where-the-luts-go/">18 Куда уходят LUT</a></li></ol></li><li><h3>Модуль 7 · Размещение, трассировка, тайминг</h3><p>Где ячейки оказываются на кристалле и успевает ли тактовая частота.</p><ol><li><a href="/ru/learn/placement/">19 Размещение: где оказываются ячейки</a></li><li class="on" aria-current="page"><a href="/ru/learn/logic-depth/">20 Глубина логики</a></li><li><a href="/ru/learn/timing-slack/">21 Тайминг и запас</a></li></ol></li><li><h3>Модуль 8 · Битстрим</h3><p>Как оттрассированный дизайн становится битами, которые загружает чип, без Vivado.</p><ol><li><a href="/ru/learn/fasm-to-frames/">22 От FASM к фреймам</a></li><li><a href="/ru/learn/a-build-receipt/">23 Сборка по пунктам</a></li><li><a href="/ru/learn/read-a-bitstream/">24 Читаем битстрим</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Спросите чип, кто он, загрузите биты и проверьте их.</p><ol><li><a href="/ru/learn/jtag-asks-the-chip/">25 JTAG спрашивает чип</a></li><li><a href="/ru/learn/same-bits-on-the-board/">26 Те же биты — на плате</a></li><li><a href="/ru/learn/capstone/">27 Итог: переугадайте yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/long-multiplication/index.html b/apps/website/public/ru/learn/long-multiplication/index.html
index 06d62653a7..94f66a59b6 100644
--- a/apps/website/public/ru/learn/long-multiplication/index.html
+++ b/apps/website/public/ru/learn/long-multiplication/index.html
@@ -769,5 +769,6 @@ <h1>Умножение столбиком</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Три значения</h3><p>Почему три, как сбалансированная троичная система пишет любое число без знака и что дают переворот и отсечение тритов.</p><ol><li><a href="/ru/learn/why-three/">01 Почему три</a></li><li><a href="/ru/learn/balanced-ternary/">02 Сбалансированная троичная</a></li><li><a href="/ru/learn/flip-and-cut/">03 Переворот вместо минуса, отсечение вместо округления</a></li></ol></li><li><h3>Модуль 2 · Логика с «не знаю»</h3><p>Трёхзначные вентили Клини, два вентиля без двоичного двойника и сложение как пара таблиц.</p><ol><li><a href="/ru/learn/kleene-logic/">04 Трёхзначная логика</a></li><li><a href="/ru/learn/consensus-and-any/">05 Консенсус и «любой»</a></li><li><a href="/ru/learn/sum-and-carry/">06 Сумма и перенос</a></li></ol></li><li><h3>Модуль 3 · Складываем триты</h3><p>Полусумматор, полный сумматор и перенос, бегущий влево разряд за разрядом.</p><ol><li><a href="/ru/learn/half-adder/">07 Полусумматор</a></li><li><a href="/ru/learn/full-adder/">08 Полный сумматор</a></li><li><a href="/ru/learn/ripple-carry/">09 Сквозной перенос</a></li></ol></li><li><h3>Модуль 4 · Умножение без умножителя</h3><p>Копия, ноль или переворот: произведение на один трит, умножение столбиком и MAC, который только складывает.</p><ol><li><a href="/ru/learn/times-a-trit/">10 Умножить на один трит</a></li><li class="on" aria-current="page"><a href="/ru/learn/long-multiplication/">11 Умножение столбиком</a></li><li><a href="/ru/learn/ternary-mac/">12 Троичный MAC</a></li></ol></li><li><h3>Модуль 5 · Триты в двоичной памяти</h3><p>Два бита на трит, пять тритов на байт и сколько бит нужно слову из 27 тритов.</p><ol><li><a href="/ru/learn/two-bits-per-trit/">13 Два бита на трит</a></li><li><a href="/ru/learn/five-trits-a-byte/">14 Пять тритов в байте</a></li><li><a href="/ru/learn/a-word-of-27/">15 Слово из 27 тритов</a></li></ol></li><li><h3>Модуль 6 · Командное слово TRI-27</h3><p>32-битное слово, которое декодирует эмулятор Trinity: его поля, 47 опкодов и 15-битное непосредственное значение.</p><ol><li><a href="/ru/learn/the-instruction-word/">16 Командное слово</a></li><li><a href="/ru/learn/the-opcode-map/">17 Карта опкодов</a></li><li><a href="/ru/learn/the-immediate/">18 Непосредственное значение</a></li></ol></li><li><h3>Модуль 7 · Троичная машина</h3><p>АЛУ из сумматоров этого курса, переход на три стороны и программа, которую можно пройти по шагам.</p><ol><li><a href="/ru/learn/a-ternary-alu/">19 Троичное АЛУ</a></li><li><a href="/ru/learn/three-way-jump/">20 Переход на три стороны</a></li><li><a href="/ru/learn/run-a-program/">21 Запустить программу</a></li></ol></li><li><h3>Модуль 8 · Три ответа</h3><p>Сравнение по старшему различному триту, поиск числа по третям и сортировка трёхзначными сравнениями.</p><ol><li><a href="/ru/learn/compare-at-the-top/">22 Сравнение сверху</a></li><li><a href="/ru/learn/questions-in-thirds/">23 Вопросы по третям</a></li><li><a href="/ru/learn/sort-by-three/">24 Сортировка трёхзначными сравнениями</a></li></ol></li><li><h3>Модуль 9 · Троичные нейроны</h3><p>Веса -1, 0, +1: один нейрон, детектор и маленький слой, и нигде нет умножителя.</p><ol><li><a href="/ru/learn/one-ternary-neuron/">25 Один троичный нейрон</a></li><li><a href="/ru/learn/a-bar-detector/">26 Детектор полосы</a></li><li><a href="/ru/learn/a-tiny-layer/">27 Маленький слой</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/loops/index.html b/apps/website/public/ru/learn/loops/index.html
index 816220e3fe..f893bda896 100644
--- a/apps/website/public/ru/learn/loops/index.html
+++ b/apps/website/public/ru/learn/loops/index.html
@@ -97,5 +97,6 @@ <h1>Циклы: while и for</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Файл спеки</h3><p>Что лежит в файле .t27: строка module, строки прозы и комментарии.</p><ol><li><a href="/ru/learn/what-a-spec-is/">01 Что такое спека</a></li><li><a href="/ru/learn/the-module-line/">02 Строка module</a></li><li><a href="/ru/learn/comments-and-prose/">03 Комментарии и строки прозы</a></li></ol></li><li><h3>Модуль 2 · Значения и типы</h3><p>Константы, ширина целого числа, а также true, false и текст.</p><ol><li><a href="/ru/learn/constants/">04 Константы</a></li><li><a href="/ru/learn/whole-numbers/">05 Целые числа и их ширина</a></li><li><a href="/ru/learn/true-false-and-text/">06 Истина, ложь и текст</a></li></ol></li><li><h3>Модуль 3 · Массивы, триты и выражения</h3><p>Списки значений, три значения трита и что вычисляют операторы.</p><ol><li><a href="/ru/learn/arrays/">07 Массивы</a></li><li><a href="/ru/learn/trits/">08 Триты: минус один, ноль, один</a></li><li><a href="/ru/learn/expressions/">09 Выражения</a></li></ol></li><li><h3>Модуль 4 · Тесты и функции</h3><p>Блок test, несколько маленьких тестов в одной спеке и функция с типизированными входами.</p><ol><li><a href="/ru/learn/a-test-block/">10 Блок test</a></li><li><a href="/ru/learn/many-tests/">11 Много тестов в одной спеке</a></li><li><a href="/ru/learn/functions/">12 Функции</a></li></ol></li><li><h3>Модуль 5 · Внутри функции</h3><p>Имена через let и var, выбор через if и switch и циклы.</p><ol><li><a href="/ru/learn/local-names/">13 Локальные имена: let и var</a></li><li><a href="/ru/learn/if-and-else/">14 Выбор: if, else и switch</a></li><li class="on" aria-current="page"><a href="/ru/learn/loops/">15 Циклы: while и for</a></li></ol></li><li><h3>Модуль 6 · Видимость и свои типы</h3><p>Что pub открывает другим модулям, и структуры и перечисления, которые вы объявляете.</p><ol><li><a href="/ru/learn/pub-or-private/">16 pub или закрытое имя</a></li><li><a href="/ru/learn/structs/">17 Структуры</a></li><li><a href="/ru/learn/enums/">18 Перечисления</a></li></ol></li><li><h3>Модуль 7 · Модули, правила и gen-ts</h3><p>Чтение другого модуля, правило, которое выполняется всегда, и спека, превращённая в TypeScript.</p><ol><li><a href="/ru/learn/use-other-modules/">19 Другие модули</a></li><li><a href="/ru/learn/invariants/">20 Инварианты и замеры</a></li><li><a href="/ru/learn/gen-ts/">21 gen-ts: спека превращается в TypeScript</a></li></ol></li><li><h3>Модуль 8 · Компилятор и tri</h3><p>Бэкенды, которые пишет t27c, стадии, в которых он читает спеку, и команда tri.</p><ol><li><a href="/ru/learn/seven-backends/">22 Семь бэкендов</a></li><li><a href="/ru/learn/stages-of-t27c/">23 Стадии t27c</a></li><li><a href="/ru/learn/tri-commands/">24 tri: командная строка</a></li></ol></li><li><h3>Модуль 9 · Ошибки, программа и что дальше</h3><p>Как читать ошибку компилятора, одна маленькая законченная программа и какой курс пройти дальше.</p><ol><li><a href="/ru/learn/reading-errors/">25 Как читать ошибки компилятора</a></li><li><a href="/ru/learn/a-small-program/">26 Маленькая законченная программа</a></li><li><a href="/ru/learn/where-next/">27 Куда дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/lowering-the-rate/index.html b/apps/website/public/ru/learn/lowering-the-rate/index.html
index 3246bf58b6..8adcc761a2 100644
--- a/apps/website/public/ru/learn/lowering-the-rate/index.html
+++ b/apps/website/public/ru/learn/lowering-the-rate/index.html
@@ -1122,5 +1122,6 @@ <h1>Понижение частоты</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Сложение</h3><p>Почему сложение медленное: перенос, который идёт вверх по слову, выделенная цепь в каждом слайсе, префиксные сети и деревья сохранения переноса.</p><ol><li><a href="/ru/learn/the-carry-chain/">01 Цепь переноса</a></li><li><a href="/ru/learn/prefix-adders/">02 Взгляд вперёд: префиксные сумматоры</a></li><li><a href="/ru/learn/carry-save/">03 Три числа сразу</a></li></ol></li><li><h3>Модуль 2 · Умножение</h3><p>Произведение — сумма сдвинутых копий; перекодирование Бута вдвое сокращает их, а слайс DSP48E1 делает остальное одним блоком.</p><ol><li><a href="/ru/learn/shift-and-add/">04 Сдвиг и сложение</a></li><li><a href="/ru/learn/booths-trick/">05 Приём Бута</a></li><li><a href="/ru/learn/the-dsp48e1-slice/">06 Слайс DSP48E1</a></li></ol></li><li><h3>Модуль 3 · Фиксированная точка</h3><p>Где стоит двоичная точка, что округление делает со значением и с его средним и что даёт каждый бит квантователя.</p><ol><li><a href="/ru/learn/q-formats/">07 Форматы Q</a></li><li><a href="/ru/learn/rounding-and-saturation/">08 Округление и насыщение</a></li><li><a href="/ru/learn/quantization-noise/">09 Шум квантования</a></li></ol></li><li><h3>Модуль 4 · Функции в железе</h3><p>Синус, косинус, угол и длина из сдвигов и сложений — и когда таблица оказывается выгоднее.</p><ol><li><a href="/ru/learn/rotating-by-shifts/">10 Поворот сдвигами</a></li><li><a href="/ru/learn/angle-and-magnitude/">11 Угол и модуль</a></li><li><a href="/ru/learn/tables-and-interpolation/">12 Таблицы и интерполяция</a></li></ol></li><li><h3>Модуль 5 · Сигналы и дискретизация</h3><p>Что дискретизация делает с частотой, как генератор строится из сумматора и что измеряет бин ДПФ.</p><ol><li><a href="/ru/learn/sampling-and-aliasing/">13 Дискретизация и наложение спектров</a></li><li><a href="/ru/learn/the-nco/">14 Генератор с числовым управлением</a></li><li><a href="/ru/learn/bins-of-a-spectrum/">15 Бины спектра</a></li></ol></li><li><h3>Модуль 6 · Фильтры FIR</h3><p>Скользящее среднее, расчёт окном sinc с целыми коэффициентами и раскладка отводов по слайсам DSP.</p><ol><li><a href="/ru/learn/the-moving-average/">16 Скользящее среднее</a></li><li><a href="/ru/learn/designing-taps/">17 Расчёт коэффициентов</a></li><li><a href="/ru/learn/folding-taps-onto-slices/">18 Раскладка отводов по слайсам</a></li></ol></li><li><h3>Модуль 7 · Многоскоростная обработка</h3><p>Понижение частоты дискретизации без заворота шума: децимация, фильтр CIC и полифазная форма.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/lowering-the-rate/">19 Понижение частоты</a></li><li><a href="/ru/learn/the-cic-filter/">20 Фильтр CIC</a></li><li><a href="/ru/learn/polyphase/">21 Полифазная форма</a></li></ol></li><li><h3>Модуль 8 · БПФ</h3><p>N log N вместо N^2: бабочки, бит-реверсный порядок входа и биты, которые добавляет каждый этап.</p><ol><li><a href="/ru/learn/the-butterfly/">22 Бабочка</a></li><li><a href="/ru/learn/twiddles-and-bit-reversal/">23 Поворачивающие множители и бит-реверс</a></li><li><a href="/ru/learn/growth-through-the-stages/">24 Рост через этапы</a></li></ol></li><li><h3>Модуль 9 · На стенде</h3><p>От арифметики к плате: коррелятор из спеки модема, бюджет слайсов и тайминга и итоговый фильтр.</p><ol><li><a href="/ru/learn/a-correlator-finds-the-frame/">25 Коррелятор находит кадр</a></li><li><a href="/ru/learn/does-it-fit-and-close-timing/">26 Помещается ли и сходится ли тайминг?</a></li><li><a href="/ru/learn/capstone-a-filter-on-the-board/">27 Итог: фильтр на плате</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/lucas-numbers-stay-whole/index.html b/apps/website/public/ru/learn/lucas-numbers-stay-whole/index.html
index b958f7bb19..daafa1b681 100644
--- a/apps/website/public/ru/learn/lucas-numbers-stay-whole/index.html
+++ b/apps/website/public/ru/learn/lucas-numbers-stay-whole/index.html
@@ -207,5 +207,6 @@ <h1>Числа Люка остаются целыми</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Правило и его числа</h3><p>Одно правило делит каждую ширину, отношение, к которому оно стремится, и числа Люка за тройкой 3.</p><ol><li><a href="/ru/learn/a-float-cut-by-phi/">01 Число с плавающей точкой, разрезанное по phi</a></li><li><a href="/ru/learn/phi-as-a-ratio/">02 Phi как отношение</a></li><li class="on" aria-current="page"><a href="/ru/learn/lucas-numbers-stay-whole/">03 Числа Люка остаются целыми</a></li></ol></li><li><h3>Модуль 2 · Почему phi, почему три</h3><p>Почему деление идёт по phi, почему основание три и как спека проверяет, что GF16 хранит phi.</p><ol><li><a href="/ru/learn/why-the-split-is-phi/">04 Почему деление по phi</a></li><li><a href="/ru/learn/why-base-three/">05 Почему основание три</a></li><li><a href="/ru/learn/phi-in-sixteen-bits/">06 Phi в шестнадцати битах</a></li></ol></li><li><h3>Модуль 3 · Малые ступени: от GF4 до GF8</h3><p>GF4, GF6 и GF8 — меньше всего битов, и округление до целых битов стоит здесь дороже всего.</p><ol><li><a href="/ru/learn/gf4-four-bits/">07 GF4: четыре бита</a></li><li><a href="/ru/learn/gf6-six-bits/">08 GF6: шесть битов</a></li><li><a href="/ru/learn/gf8-one-byte/">09 GF8: один байт</a></li></ol></li><li><h3>Модуль 4 · От десяти до четырнадцати битов</h3><p>GF10, GF12 и GF14 и то, как расстояние до 1 / phi меняется с ростом слова.</p><ol><li><a href="/ru/learn/gf10-ten-bits/">10 GF10: десять битов</a></li><li><a href="/ru/learn/gf12-twelve-bits/">11 GF12: двенадцать битов</a></li><li><a href="/ru/learn/gf14-fourteen-bits/">12 GF14: четырнадцать битов</a></li></ol></li><li><h3>Модуль 5 · GF16 в работе</h3><p>Основной 16-битный формат, скалярное произведение из двух слагаемых в GF-T16, затем GF20 и GF24.</p><ol><li><a href="/ru/learn/gf16-the-primary/">13 GF16: основной формат</a></li><li><a href="/ru/learn/two-products-one-sum/">14 Два произведения, одна сумма</a></li><li><a href="/ru/learn/gf24-twenty-four-bits/">15 GF20 и GF24</a></li></ol></li><li><h3>Модуль 6 · От GF32 до GF64</h3><p>GF32 рядом с IEEE single, GF48 без пары в IEEE, GF64 рядом с IEEE double.</p><ol><li><a href="/ru/learn/gf32-a-single/">16 GF32: одинарная точность</a></li><li><a href="/ru/learn/gf48-forty-eight-bits/">17 GF48: сорок восемь битов</a></li><li><a href="/ru/learn/gf64-a-double/">18 GF64: двойная точность</a></li></ol></li><li><h3>Модуль 7 · От GF96 до GF256</h3><p>GF96, GF128 и GF256, где спеки держат раскладку инвариантами.</p><ol><li><a href="/ru/learn/gf96-ninety-six-bits/">19 GF96: девяносто шесть битов</a></li><li><a href="/ru/learn/gf128-a-quad/">20 GF128: четверная точность</a></li><li><a href="/ru/learn/gf256-two-five-six/">21 GF256: двести пятьдесят шесть битов</a></li></ol></li><li><h3>Модуль 8 · Самые широкие ступени, затем триты</h3><p>GF512 и GF1024, две самые широкие ступени, затем GF-T8, где порядок уходит в триты.</p><ol><li><a href="/ru/learn/gf512-five-twelve/">22 GF512: пятьсот двенадцать битов</a></li><li><a href="/ru/learn/gf1024-one-kilobit/">23 GF1024: один килобит</a></li><li><a href="/ru/learn/gft8-exponent-in-trits/">24 GF-T8: порядок в тритах</a></li></ol></li><li><h3>Модуль 9 · Ещё триты, затем декодирование</h3><p>GF-T16 и GF-T32, затем почему фиксированные поля декодируются параллельно, а posit — нет.</p><ol><li><a href="/ru/learn/gft16-six-trits/">25 GF-T16: шесть тритов</a></li><li><a href="/ru/learn/gft32-twelve-trits/">26 GF-T32: двенадцать тритов</a></li><li><a href="/ru/learn/decode-in-one-step/">27 Раскодировать за один шаг</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/many-tests/index.html b/apps/website/public/ru/learn/many-tests/index.html
index f5c3715079..2b6099f9e8 100644
--- a/apps/website/public/ru/learn/many-tests/index.html
+++ b/apps/website/public/ru/learn/many-tests/index.html
@@ -93,5 +93,6 @@ <h1>Много тестов в одной спеке</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Файл спеки</h3><p>Что лежит в файле .t27: строка module, строки прозы и комментарии.</p><ol><li><a href="/ru/learn/what-a-spec-is/">01 Что такое спека</a></li><li><a href="/ru/learn/the-module-line/">02 Строка module</a></li><li><a href="/ru/learn/comments-and-prose/">03 Комментарии и строки прозы</a></li></ol></li><li><h3>Модуль 2 · Значения и типы</h3><p>Константы, ширина целого числа, а также true, false и текст.</p><ol><li><a href="/ru/learn/constants/">04 Константы</a></li><li><a href="/ru/learn/whole-numbers/">05 Целые числа и их ширина</a></li><li><a href="/ru/learn/true-false-and-text/">06 Истина, ложь и текст</a></li></ol></li><li><h3>Модуль 3 · Массивы, триты и выражения</h3><p>Списки значений, три значения трита и что вычисляют операторы.</p><ol><li><a href="/ru/learn/arrays/">07 Массивы</a></li><li><a href="/ru/learn/trits/">08 Триты: минус один, ноль, один</a></li><li><a href="/ru/learn/expressions/">09 Выражения</a></li></ol></li><li><h3>Модуль 4 · Тесты и функции</h3><p>Блок test, несколько маленьких тестов в одной спеке и функция с типизированными входами.</p><ol><li><a href="/ru/learn/a-test-block/">10 Блок test</a></li><li class="on" aria-current="page"><a href="/ru/learn/many-tests/">11 Много тестов в одной спеке</a></li><li><a href="/ru/learn/functions/">12 Функции</a></li></ol></li><li><h3>Модуль 5 · Внутри функции</h3><p>Имена через let и var, выбор через if и switch и циклы.</p><ol><li><a href="/ru/learn/local-names/">13 Локальные имена: let и var</a></li><li><a href="/ru/learn/if-and-else/">14 Выбор: if, else и switch</a></li><li><a href="/ru/learn/loops/">15 Циклы: while и for</a></li></ol></li><li><h3>Модуль 6 · Видимость и свои типы</h3><p>Что pub открывает другим модулям, и структуры и перечисления, которые вы объявляете.</p><ol><li><a href="/ru/learn/pub-or-private/">16 pub или закрытое имя</a></li><li><a href="/ru/learn/structs/">17 Структуры</a></li><li><a href="/ru/learn/enums/">18 Перечисления</a></li></ol></li><li><h3>Модуль 7 · Модули, правила и gen-ts</h3><p>Чтение другого модуля, правило, которое выполняется всегда, и спека, превращённая в TypeScript.</p><ol><li><a href="/ru/learn/use-other-modules/">19 Другие модули</a></li><li><a href="/ru/learn/invariants/">20 Инварианты и замеры</a></li><li><a href="/ru/learn/gen-ts/">21 gen-ts: спека превращается в TypeScript</a></li></ol></li><li><h3>Модуль 8 · Компилятор и tri</h3><p>Бэкенды, которые пишет t27c, стадии, в которых он читает спеку, и команда tri.</p><ol><li><a href="/ru/learn/seven-backends/">22 Семь бэкендов</a></li><li><a href="/ru/learn/stages-of-t27c/">23 Стадии t27c</a></li><li><a href="/ru/learn/tri-commands/">24 tri: командная строка</a></li></ol></li><li><h3>Модуль 9 · Ошибки, программа и что дальше</h3><p>Как читать ошибку компилятора, одна маленькая законченная программа и какой курс пройти дальше.</p><ol><li><a href="/ru/learn/reading-errors/">25 Как читать ошибки компилятора</a></li><li><a href="/ru/learn/a-small-program/">26 Маленькая законченная программа</a></li><li><a href="/ru/learn/where-next/">27 Куда дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/master-and-servant/index.html b/apps/website/public/ru/learn/master-and-servant/index.html
index 2c01c1cbcf..4ae0bc6202 100644
--- a/apps/website/public/ru/learn/master-and-servant/index.html
+++ b/apps/website/public/ru/learn/master-and-servant/index.html
@@ -417,5 +417,6 @@ <h1>Хозяин и слуга</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое шина</h3><p>Зачем вообще существует шина: разговор по проводам, с кадрами и адресами, и кому позволено говорить.</p><ol><li><a href="/ru/learn/wires-that-talk/">01 Провода, которые говорят</a></li><li><a href="/ru/learn/framing-a-conversation/">02 Кадрирование разговора</a></li><li class="on" aria-current="page"><a href="/ru/learn/master-and-servant/">03 Хозяин и слуга</a></li></ol></li><li><h3>Модуль 2 · Разговор по UART</h3><p>Двухпроводная шина без такта: кадр, делитель, задающий скорость, и статус, который опрашивает драйвер.</p><ol><li><a href="/ru/learn/the-frame/">04 Кадр: старт, данные, стоп</a></li><li><a href="/ru/learn/baud-and-the-divisor/">05 Бод и делитель</a></li><li><a href="/ru/learn/status-and-fifos/">06 Статус и FIFO</a></li></ol></li><li><h3>Модуль 3 · SPI под тактом</h3><p>Разговор под тактом: четыре режима, лестница предделителей и выбор кристалла на каждого подчинённого.</p><ol><li><a href="/ru/learn/modes-cpol-cpha/">07 Режимы: CPOL и CPHA</a></li><li><a href="/ru/learn/the-prescaler-ladder/">08 Лестница предделителей</a></li><li><a href="/ru/learn/chip-select-and-widths/">09 Выбор кристалла и ширины</a></li></ol></li><li><h3>Модуль 4 · Регистровая шина APB</h3><p>Шина регистров: PSEL и PENABLE, строобы и ожидания, и сколько адресных бит стоит количество периферии.</p><ol><li><a href="/ru/learn/psel-and-penable/">10 PSEL и PENABLE</a></li><li><a href="/ru/learn/strobes-and-waits/">11 Строобы и ожидания</a></li><li><a href="/ru/learn/address-decode/">12 Дешифрация адреса</a></li></ol></li><li><h3>Модуль 5 · Пять каналов AXI4</h3><p>Пять каналов: адрес, данные и отклик в обе стороны, lite или full, пакеты и идентификаторы.</p><ol><li><a href="/ru/learn/five-channels/">13 Пять каналов</a></li><li><a href="/ru/learn/lite-or-full/">14 Lite или full</a></li><li><a href="/ru/learn/bursts-and-ids/">15 Пакеты и идентификаторы</a></li></ol></li><li><h3>Модуль 6 · Память</h3><p>Что находится на дальнем конце каждой шины: карты памяти, типы портов и задержка, которую должно покрывать ожидание.</p><ol><li><a href="/ru/learn/memory-maps/">16 Карты памяти</a></li><li><a href="/ru/learn/ports-and-kinds/">17 Порты и типы</a></li><li><a href="/ru/learn/latency/">18 Задержка</a></li></ol></li><li><h3>Модуль 7 · Мосты</h3><p>Почему в дизайне появляется больше одной шины и пакетный мост, переносящий работу между ними.</p><ol><li><a href="/ru/learn/why-bridges/">19 Зачем мосты</a></li><li><a href="/ru/learn/the-packet-bridge/">20 Пакетный мост</a></li><li><a href="/ru/learn/everything-in-one-bench/">21 Всё в одном стенде</a></li></ol></li><li><h3>Модуль 8 · Ethernet: кадры и тайминг</h3><p>Кадры, контрольная сумма кадра, тайминг RGMII и заранее зарегистрированные шаги настоящего запуска.</p><ol><li><a href="/ru/learn/frames-and-the-fcs/">22 Кадры и контрольная сумма</a></li><li><a href="/ru/learn/rgmii-timing/">23 Тайминг RGMII</a></li><li><a href="/ru/learn/bring-up-on-the-bench/">24 Запуск на стенде</a></li></ol></li><li><h3>Модуль 9 · Стенд</h3><p>Дисциплина, оберегающая настоящее железо: кто держит IO, как взять и вернуть захват, и что запускается дальше.</p><ol><li><a href="/ru/learn/who-holds-the-io/">25 Кто держит IO</a></li><li><a href="/ru/learn/claim-and-release/">26 Захват и возврат</a></li><li><a href="/ru/learn/capstone-what-runs-next/">27 Капстоун: что запускается дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/memory-maps/index.html b/apps/website/public/ru/learn/memory-maps/index.html
index e16b2b8384..5f8ce16e7c 100644
--- a/apps/website/public/ru/learn/memory-maps/index.html
+++ b/apps/website/public/ru/learn/memory-maps/index.html
@@ -424,5 +424,6 @@ <h1>Карты памяти</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое шина</h3><p>Зачем вообще существует шина: разговор по проводам, с кадрами и адресами, и кому позволено говорить.</p><ol><li><a href="/ru/learn/wires-that-talk/">01 Провода, которые говорят</a></li><li><a href="/ru/learn/framing-a-conversation/">02 Кадрирование разговора</a></li><li><a href="/ru/learn/master-and-servant/">03 Хозяин и слуга</a></li></ol></li><li><h3>Модуль 2 · Разговор по UART</h3><p>Двухпроводная шина без такта: кадр, делитель, задающий скорость, и статус, который опрашивает драйвер.</p><ol><li><a href="/ru/learn/the-frame/">04 Кадр: старт, данные, стоп</a></li><li><a href="/ru/learn/baud-and-the-divisor/">05 Бод и делитель</a></li><li><a href="/ru/learn/status-and-fifos/">06 Статус и FIFO</a></li></ol></li><li><h3>Модуль 3 · SPI под тактом</h3><p>Разговор под тактом: четыре режима, лестница предделителей и выбор кристалла на каждого подчинённого.</p><ol><li><a href="/ru/learn/modes-cpol-cpha/">07 Режимы: CPOL и CPHA</a></li><li><a href="/ru/learn/the-prescaler-ladder/">08 Лестница предделителей</a></li><li><a href="/ru/learn/chip-select-and-widths/">09 Выбор кристалла и ширины</a></li></ol></li><li><h3>Модуль 4 · Регистровая шина APB</h3><p>Шина регистров: PSEL и PENABLE, строобы и ожидания, и сколько адресных бит стоит количество периферии.</p><ol><li><a href="/ru/learn/psel-and-penable/">10 PSEL и PENABLE</a></li><li><a href="/ru/learn/strobes-and-waits/">11 Строобы и ожидания</a></li><li><a href="/ru/learn/address-decode/">12 Дешифрация адреса</a></li></ol></li><li><h3>Модуль 5 · Пять каналов AXI4</h3><p>Пять каналов: адрес, данные и отклик в обе стороны, lite или full, пакеты и идентификаторы.</p><ol><li><a href="/ru/learn/five-channels/">13 Пять каналов</a></li><li><a href="/ru/learn/lite-or-full/">14 Lite или full</a></li><li><a href="/ru/learn/bursts-and-ids/">15 Пакеты и идентификаторы</a></li></ol></li><li><h3>Модуль 6 · Память</h3><p>Что находится на дальнем конце каждой шины: карты памяти, типы портов и задержка, которую должно покрывать ожидание.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/memory-maps/">16 Карты памяти</a></li><li><a href="/ru/learn/ports-and-kinds/">17 Порты и типы</a></li><li><a href="/ru/learn/latency/">18 Задержка</a></li></ol></li><li><h3>Модуль 7 · Мосты</h3><p>Почему в дизайне появляется больше одной шины и пакетный мост, переносящий работу между ними.</p><ol><li><a href="/ru/learn/why-bridges/">19 Зачем мосты</a></li><li><a href="/ru/learn/the-packet-bridge/">20 Пакетный мост</a></li><li><a href="/ru/learn/everything-in-one-bench/">21 Всё в одном стенде</a></li></ol></li><li><h3>Модуль 8 · Ethernet: кадры и тайминг</h3><p>Кадры, контрольная сумма кадра, тайминг RGMII и заранее зарегистрированные шаги настоящего запуска.</p><ol><li><a href="/ru/learn/frames-and-the-fcs/">22 Кадры и контрольная сумма</a></li><li><a href="/ru/learn/rgmii-timing/">23 Тайминг RGMII</a></li><li><a href="/ru/learn/bring-up-on-the-bench/">24 Запуск на стенде</a></li></ol></li><li><h3>Модуль 9 · Стенд</h3><p>Дисциплина, оберегающая настоящее железо: кто держит IO, как взять и вернуть захват, и что запускается дальше.</p><ol><li><a href="/ru/learn/who-holds-the-io/">25 Кто держит IO</a></li><li><a href="/ru/learn/claim-and-release/">26 Захват и возврат</a></li><li><a href="/ru/learn/capstone-what-runs-next/">27 Капстоун: что запускается дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/modes-cpol-cpha/index.html b/apps/website/public/ru/learn/modes-cpol-cpha/index.html
index 6fb1357578..d948e5760b 100644
--- a/apps/website/public/ru/learn/modes-cpol-cpha/index.html
+++ b/apps/website/public/ru/learn/modes-cpol-cpha/index.html
@@ -476,5 +476,6 @@ <h1>Режимы: CPOL и CPHA</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое шина</h3><p>Зачем вообще существует шина: разговор по проводам, с кадрами и адресами, и кому позволено говорить.</p><ol><li><a href="/ru/learn/wires-that-talk/">01 Провода, которые говорят</a></li><li><a href="/ru/learn/framing-a-conversation/">02 Кадрирование разговора</a></li><li><a href="/ru/learn/master-and-servant/">03 Хозяин и слуга</a></li></ol></li><li><h3>Модуль 2 · Разговор по UART</h3><p>Двухпроводная шина без такта: кадр, делитель, задающий скорость, и статус, который опрашивает драйвер.</p><ol><li><a href="/ru/learn/the-frame/">04 Кадр: старт, данные, стоп</a></li><li><a href="/ru/learn/baud-and-the-divisor/">05 Бод и делитель</a></li><li><a href="/ru/learn/status-and-fifos/">06 Статус и FIFO</a></li></ol></li><li><h3>Модуль 3 · SPI под тактом</h3><p>Разговор под тактом: четыре режима, лестница предделителей и выбор кристалла на каждого подчинённого.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/modes-cpol-cpha/">07 Режимы: CPOL и CPHA</a></li><li><a href="/ru/learn/the-prescaler-ladder/">08 Лестница предделителей</a></li><li><a href="/ru/learn/chip-select-and-widths/">09 Выбор кристалла и ширины</a></li></ol></li><li><h3>Модуль 4 · Регистровая шина APB</h3><p>Шина регистров: PSEL и PENABLE, строобы и ожидания, и сколько адресных бит стоит количество периферии.</p><ol><li><a href="/ru/learn/psel-and-penable/">10 PSEL и PENABLE</a></li><li><a href="/ru/learn/strobes-and-waits/">11 Строобы и ожидания</a></li><li><a href="/ru/learn/address-decode/">12 Дешифрация адреса</a></li></ol></li><li><h3>Модуль 5 · Пять каналов AXI4</h3><p>Пять каналов: адрес, данные и отклик в обе стороны, lite или full, пакеты и идентификаторы.</p><ol><li><a href="/ru/learn/five-channels/">13 Пять каналов</a></li><li><a href="/ru/learn/lite-or-full/">14 Lite или full</a></li><li><a href="/ru/learn/bursts-and-ids/">15 Пакеты и идентификаторы</a></li></ol></li><li><h3>Модуль 6 · Память</h3><p>Что находится на дальнем конце каждой шины: карты памяти, типы портов и задержка, которую должно покрывать ожидание.</p><ol><li><a href="/ru/learn/memory-maps/">16 Карты памяти</a></li><li><a href="/ru/learn/ports-and-kinds/">17 Порты и типы</a></li><li><a href="/ru/learn/latency/">18 Задержка</a></li></ol></li><li><h3>Модуль 7 · Мосты</h3><p>Почему в дизайне появляется больше одной шины и пакетный мост, переносящий работу между ними.</p><ol><li><a href="/ru/learn/why-bridges/">19 Зачем мосты</a></li><li><a href="/ru/learn/the-packet-bridge/">20 Пакетный мост</a></li><li><a href="/ru/learn/everything-in-one-bench/">21 Всё в одном стенде</a></li></ol></li><li><h3>Модуль 8 · Ethernet: кадры и тайминг</h3><p>Кадры, контрольная сумма кадра, тайминг RGMII и заранее зарегистрированные шаги настоящего запуска.</p><ol><li><a href="/ru/learn/frames-and-the-fcs/">22 Кадры и контрольная сумма</a></li><li><a href="/ru/learn/rgmii-timing/">23 Тайминг RGMII</a></li><li><a href="/ru/learn/bring-up-on-the-bench/">24 Запуск на стенде</a></li></ol></li><li><h3>Модуль 9 · Стенд</h3><p>Дисциплина, оберегающая настоящее железо: кто держит IO, как взять и вернуть захват, и что запускается дальше.</p><ol><li><a href="/ru/learn/who-holds-the-io/">25 Кто держит IO</a></li><li><a href="/ru/learn/claim-and-release/">26 Захват и возврат</a></li><li><a href="/ru/learn/capstone-what-runs-next/">27 Капстоун: что запускается дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/mtbf/index.html b/apps/website/public/ru/learn/mtbf/index.html
index 14c02c3d61..ab1bb245e5 100644
--- a/apps/website/public/ru/learn/mtbf/index.html
+++ b/apps/website/public/ru/learn/mtbf/index.html
@@ -330,5 +330,6 @@ <h1>Среднее время между отказами</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое тактовый сигнал</h3><p>Один фронт, один мир: всё, что делит фронт тактового сигнала, делит и мир; период и джиттер настоящего фронта и место, где тактовый сигнал входит в плату.</p><ol><li><a href="/ru/learn/one-edge-one-world/">01 Один фронт, один мир</a></li><li><a href="/ru/learn/period-and-jitter/">02 Период, фронт и джиттер</a></li><li><a href="/ru/learn/where-the-clock-enters/">03 Где входит тактовый сигнал</a></li></ol></li><li><h3>Модуль 2 · Деревья тактового сигнала</h3><p>Перекос и задержка прохождения, глобальная сеть буферов и ловушка стробирования клока логикой.</p><ol><li><a href="/ru/learn/skew-and-insertion/">04 Перекос и задержка прохождения</a></li><li><a href="/ru/learn/global-buffers/">05 Глобальный буфер</a></li><li><a href="/ru/learn/gated-clock-trap/">06 Ловушка стробированного клока</a></li></ol></li><li><h3>Модуль 3 · PLL и MMCM</h3><p>Умножить и поделить один тактовый сигнал в другой, сдвинуть его фазу шагами VCO и какие тактовые сигналы анализатор считает родственными.</p><ol><li><a href="/ru/learn/multiply-and-divide/">07 Умножить и поделить</a></li><li><a href="/ru/learn/phase-and-lock/">08 Фаза и захват</a></li><li><a href="/ru/learn/related-or-not/">09 Родственные или чужие</a></li></ol></li><li><h3>Модуль 4 · Сбросы</h3><p>Ассертировать асинхронно, снимать синхронно: три вида сброса, конвейер снятия и дерево, которое растит сброс.</p><ol><li><a href="/ru/learn/sync-or-async/">10 Синхронный или асинхронный</a></li><li><a href="/ru/learn/reset-release/">11 Конвейер снятия</a></li><li><a href="/ru/learn/reset-fanout/">12 Разветвление сброса</a></li></ol></li><li><h3>Модуль 5 · Метастабильность</h3><p>Окно setup-hold, среднее время между отказами в целочисленной арифметике и два триггера, которые это чинят.</p><ol><li><a href="/ru/learn/setup-hold-window/">13 Окно setup-hold</a></li><li class="on" aria-current="page"><a href="/ru/learn/mtbf/">14 Среднее время между отказами</a></li><li><a href="/ru/learn/two-flop-sync/">15 Двухтриггерный синхронизатор</a></li></ol></li><li><h3>Модуль 6 · Пересечение многих бит</h3><p>Почему двоичная шина рвётся, почему код Грея нет, и рукопожатие, переносящее импульс между мирами.</p><ol><li><a href="/ru/learn/why-binary-fails/">16 Почему двоичный код рвётся</a></li><li><a href="/ru/learn/gray-code/">17 Код Грея</a></li><li><a href="/ru/learn/handshakes-and-pulses/">18 Рукопожатия и импульсы</a></li></ol></li><li><h3>Модуль 7 · Асинхронный FIFO</h3><p>Указатели, флаги и глубина: буфер, переносящий поток между двумя тактовыми сигналами.</p><ol><li><a href="/ru/learn/pointers-and-flags/">19 Указатели и флаги</a></li><li><a href="/ru/learn/depth-sizing/">20 Выбор глубины</a></li><li><a href="/ru/learn/fifo-testbench/">21 Тестбенч FIFO</a></li></ol></li><li><h3>Модуль 8 · Ограничения</h3><p>Строки, сообщающие анализатору, что такое тактовый сигнал, какие пути не проверять и чему должны соответствовать выводы.</p><ol><li><a href="/ru/learn/create-clock/">22 Строка create_clock</a></li><li><a href="/ru/learn/false-path-and-max-delay/">23 False path и max delay</a></li><li><a href="/ru/learn/io-timing/">24 Тайминг выводов</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Отчёт CDC, одно пересечение, захваченное на триггерах, и дифф битстрима, замыкающий курс.</p><ol><li><a href="/ru/learn/cdc-report/">25 Отчёт CDC</a></li><li><a href="/ru/learn/capture-a-crossing/">26 Захватить пересечение</a></li><li><a href="/ru/learn/capstone-diff/">27 Капстоун: поменяй и сравни</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/multiply-and-divide/index.html b/apps/website/public/ru/learn/multiply-and-divide/index.html
index d8fec512fb..55eea0ce9c 100644
--- a/apps/website/public/ru/learn/multiply-and-divide/index.html
+++ b/apps/website/public/ru/learn/multiply-and-divide/index.html
@@ -341,5 +341,6 @@ <h1>Умножить и поделить</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое тактовый сигнал</h3><p>Один фронт, один мир: всё, что делит фронт тактового сигнала, делит и мир; период и джиттер настоящего фронта и место, где тактовый сигнал входит в плату.</p><ol><li><a href="/ru/learn/one-edge-one-world/">01 Один фронт, один мир</a></li><li><a href="/ru/learn/period-and-jitter/">02 Период, фронт и джиттер</a></li><li><a href="/ru/learn/where-the-clock-enters/">03 Где входит тактовый сигнал</a></li></ol></li><li><h3>Модуль 2 · Деревья тактового сигнала</h3><p>Перекос и задержка прохождения, глобальная сеть буферов и ловушка стробирования клока логикой.</p><ol><li><a href="/ru/learn/skew-and-insertion/">04 Перекос и задержка прохождения</a></li><li><a href="/ru/learn/global-buffers/">05 Глобальный буфер</a></li><li><a href="/ru/learn/gated-clock-trap/">06 Ловушка стробированного клока</a></li></ol></li><li><h3>Модуль 3 · PLL и MMCM</h3><p>Умножить и поделить один тактовый сигнал в другой, сдвинуть его фазу шагами VCO и какие тактовые сигналы анализатор считает родственными.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/multiply-and-divide/">07 Умножить и поделить</a></li><li><a href="/ru/learn/phase-and-lock/">08 Фаза и захват</a></li><li><a href="/ru/learn/related-or-not/">09 Родственные или чужие</a></li></ol></li><li><h3>Модуль 4 · Сбросы</h3><p>Ассертировать асинхронно, снимать синхронно: три вида сброса, конвейер снятия и дерево, которое растит сброс.</p><ol><li><a href="/ru/learn/sync-or-async/">10 Синхронный или асинхронный</a></li><li><a href="/ru/learn/reset-release/">11 Конвейер снятия</a></li><li><a href="/ru/learn/reset-fanout/">12 Разветвление сброса</a></li></ol></li><li><h3>Модуль 5 · Метастабильность</h3><p>Окно setup-hold, среднее время между отказами в целочисленной арифметике и два триггера, которые это чинят.</p><ol><li><a href="/ru/learn/setup-hold-window/">13 Окно setup-hold</a></li><li><a href="/ru/learn/mtbf/">14 Среднее время между отказами</a></li><li><a href="/ru/learn/two-flop-sync/">15 Двухтриггерный синхронизатор</a></li></ol></li><li><h3>Модуль 6 · Пересечение многих бит</h3><p>Почему двоичная шина рвётся, почему код Грея нет, и рукопожатие, переносящее импульс между мирами.</p><ol><li><a href="/ru/learn/why-binary-fails/">16 Почему двоичный код рвётся</a></li><li><a href="/ru/learn/gray-code/">17 Код Грея</a></li><li><a href="/ru/learn/handshakes-and-pulses/">18 Рукопожатия и импульсы</a></li></ol></li><li><h3>Модуль 7 · Асинхронный FIFO</h3><p>Указатели, флаги и глубина: буфер, переносящий поток между двумя тактовыми сигналами.</p><ol><li><a href="/ru/learn/pointers-and-flags/">19 Указатели и флаги</a></li><li><a href="/ru/learn/depth-sizing/">20 Выбор глубины</a></li><li><a href="/ru/learn/fifo-testbench/">21 Тестбенч FIFO</a></li></ol></li><li><h3>Модуль 8 · Ограничения</h3><p>Строки, сообщающие анализатору, что такое тактовый сигнал, какие пути не проверять и чему должны соответствовать выводы.</p><ol><li><a href="/ru/learn/create-clock/">22 Строка create_clock</a></li><li><a href="/ru/learn/false-path-and-max-delay/">23 False path и max delay</a></li><li><a href="/ru/learn/io-timing/">24 Тайминг выводов</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Отчёт CDC, одно пересечение, захваченное на триггерах, и дифф битстрима, замыкающий курс.</p><ol><li><a href="/ru/learn/cdc-report/">25 Отчёт CDC</a></li><li><a href="/ru/learn/capture-a-crossing/">26 Захватить пересечение</a></li><li><a href="/ru/learn/capstone-diff/">27 Капстоун: поменяй и сравни</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/multiply-then-add/index.html b/apps/website/public/ru/learn/multiply-then-add/index.html
index bf87e002f0..5f535ab6da 100644
--- a/apps/website/public/ru/learn/multiply-then-add/index.html
+++ b/apps/website/public/ru/learn/multiply-then-add/index.html
@@ -239,5 +239,6 @@ <h1>Умножить, потом сложить</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Лаборатория: наши исследования</h3><p>Свой формат чисел, честная таблица результатов и таблицы модели, перемноженные на плате.</p><ol><li><a href="/ru/learn/t27-number-format/">01 T27: наш собственный формат чисел</a></li><li><a href="/ru/learn/an-honest-scoreboard/">02 Честная таблица результатов</a></li><li><a href="/ru/learn/a-model-on-the-board/">03 Таблицы модели на плате</a></li></ol></li><li><h3>Модуль 2 · ИИ-числа: блок MX</h3><p>Как ИИ-чипы хранят веса в нескольких битах: один общий масштаб на блок, сам байт масштаба и что один выброс делает с соседями.</p><ol><li><a href="/ru/learn/one-scale-per-block/">04 Один масштаб на блок весов</a></li><li><a href="/ru/learn/scale-byte-native/">05 Байт масштаба на настоящей машине</a></li><li><a href="/ru/learn/one-outlier/">06 Один выброс, много нулей</a></li></ol></li><li><h3>Модуль 3 · Тернарные веса</h3><p>Веса, которые бывают только минус масштаб, ноль или плюс масштаб, пять правил, которые должен пройти тернарный алфавит, и прогон тестов, который ничего не проверил.</p><ol><li><a href="/ru/learn/phi-scaled-ternary/">07 Веса минус фи, ноль и плюс фи</a></li><li><a href="/ru/learn/a-pass-that-checked-nothing/">08 Прогон, который ничего не проверил</a></li><li><a href="/ru/learn/five-rules-for-weights/">09 Пять правил для алфавита весов</a></li></ol></li><li><h3>Модуль 4 · Тернарный формат Ternary Network Float</h3><p>Правило, которое компилятор проверяет до запуска любого теста, и 17-битное число с плавающей точкой, порядок которого — четыре сбалансированных трита.</p><ol><li><a href="/ru/learn/blocked-is-not-failed/">10 Заблокирован — не значит провален</a></li><li><a href="/ru/learn/four-trit-exponent/">11 Число, у которого порядок — четыре трита</a></li><li><a href="/ru/learn/negate-twice/">12 Дважды сменить знак — получить число обратно</a></li></ol></li><li><h3>Модуль 5 · Арифметика чисел со знаком</h3><p>Умножить два числа со знаком, сложить их, когда знаки разные, и сделать то и другое сразу в умножении с накоплением.</p><ol><li><a href="/ru/learn/a-sign-is-one-bit/">13 Знак — это один бит</a></li><li><a href="/ru/learn/opposite-signs-subtract/">14 Разные знаки — вычитание</a></li><li class="on" aria-current="page"><a href="/ru/learn/multiply-then-add/">15 Умножить, потом сложить</a></li></ol></li><li><h3>Модуль 6 · Части нейрона</h3><p>ReLU с изломом в нуле, степень двойки для softmax и argmax, который называет ответ.</p><ol><li><a href="/ru/learn/a-bend-at-zero/">16 Излом в нуле</a></li><li><a href="/ru/learn/two-to-the-x/">17 Два в степени x</a></li><li><a href="/ru/learn/pick-the-largest/">18 Выбрать наибольшее</a></li></ol></li><li><h3>Модуль 7 · Обучение на ошибке</h3><p>Потеря, которая оценивает неверную догадку в битах, один шаг, который сдвигает вес против градиента, и скрытый слой, который нужен XOR.</p><ol><li><a href="/ru/learn/a-loss-in-bits/">19 Потеря в битах</a></li><li><a href="/ru/learn/one-step-downhill/">20 Один шаг вниз по склону</a></li><li><a href="/ru/learn/xor-needs-a-bend/">21 XOR нужен излом</a></li></ol></li><li><h3>Модуль 8 · BitNet: тернарные сети</h3><p>Порог, который сжимает сумму обратно до трёх значений, один нейрон, который становится другой функцией при смене весов, и нейрон, который читает входы по 27 тритов за раз.</p><ol><li><a href="/ru/learn/back-to-three-values/">22 Обратно в три значения</a></li><li><a href="/ru/learn/same-neuron-new-weights/">23 Тот же нейрон, другие веса</a></li><li><a href="/ru/learn/a-neuron-in-chunks/">24 Нейрон по кускам</a></li></ol></li><li><h3>Модуль 9 · Тернарный MAC как чип</h3><p>Скалярное произведение 27 тритов как провода без регистра, та же сумма, которую регистр копит на каждом такте, и небольшая целая сеть в конце курса.</p><ol><li><a href="/ru/learn/a-dot-product-in-wires/">25 Скалярное произведение из проводов</a></li><li><a href="/ru/learn/add-it-up-every-clock/">26 Складывать на каждом такте</a></li><li><a href="/ru/learn/a-whole-network/">27 Целая сеть</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/negate-twice/index.html b/apps/website/public/ru/learn/negate-twice/index.html
index 876a4946fd..cd18ddf6e2 100644
--- a/apps/website/public/ru/learn/negate-twice/index.html
+++ b/apps/website/public/ru/learn/negate-twice/index.html
@@ -436,5 +436,6 @@ <h1>Дважды сменить знак — получить число обр
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Лаборатория: наши исследования</h3><p>Свой формат чисел, честная таблица результатов и таблицы модели, перемноженные на плате.</p><ol><li><a href="/ru/learn/t27-number-format/">01 T27: наш собственный формат чисел</a></li><li><a href="/ru/learn/an-honest-scoreboard/">02 Честная таблица результатов</a></li><li><a href="/ru/learn/a-model-on-the-board/">03 Таблицы модели на плате</a></li></ol></li><li><h3>Модуль 2 · ИИ-числа: блок MX</h3><p>Как ИИ-чипы хранят веса в нескольких битах: один общий масштаб на блок, сам байт масштаба и что один выброс делает с соседями.</p><ol><li><a href="/ru/learn/one-scale-per-block/">04 Один масштаб на блок весов</a></li><li><a href="/ru/learn/scale-byte-native/">05 Байт масштаба на настоящей машине</a></li><li><a href="/ru/learn/one-outlier/">06 Один выброс, много нулей</a></li></ol></li><li><h3>Модуль 3 · Тернарные веса</h3><p>Веса, которые бывают только минус масштаб, ноль или плюс масштаб, пять правил, которые должен пройти тернарный алфавит, и прогон тестов, который ничего не проверил.</p><ol><li><a href="/ru/learn/phi-scaled-ternary/">07 Веса минус фи, ноль и плюс фи</a></li><li><a href="/ru/learn/a-pass-that-checked-nothing/">08 Прогон, который ничего не проверил</a></li><li><a href="/ru/learn/five-rules-for-weights/">09 Пять правил для алфавита весов</a></li></ol></li><li><h3>Модуль 4 · Тернарный формат Ternary Network Float</h3><p>Правило, которое компилятор проверяет до запуска любого теста, и 17-битное число с плавающей точкой, порядок которого — четыре сбалансированных трита.</p><ol><li><a href="/ru/learn/blocked-is-not-failed/">10 Заблокирован — не значит провален</a></li><li><a href="/ru/learn/four-trit-exponent/">11 Число, у которого порядок — четыре трита</a></li><li class="on" aria-current="page"><a href="/ru/learn/negate-twice/">12 Дважды сменить знак — получить число обратно</a></li></ol></li><li><h3>Модуль 5 · Арифметика чисел со знаком</h3><p>Умножить два числа со знаком, сложить их, когда знаки разные, и сделать то и другое сразу в умножении с накоплением.</p><ol><li><a href="/ru/learn/a-sign-is-one-bit/">13 Знак — это один бит</a></li><li><a href="/ru/learn/opposite-signs-subtract/">14 Разные знаки — вычитание</a></li><li><a href="/ru/learn/multiply-then-add/">15 Умножить, потом сложить</a></li></ol></li><li><h3>Модуль 6 · Части нейрона</h3><p>ReLU с изломом в нуле, степень двойки для softmax и argmax, который называет ответ.</p><ol><li><a href="/ru/learn/a-bend-at-zero/">16 Излом в нуле</a></li><li><a href="/ru/learn/two-to-the-x/">17 Два в степени x</a></li><li><a href="/ru/learn/pick-the-largest/">18 Выбрать наибольшее</a></li></ol></li><li><h3>Модуль 7 · Обучение на ошибке</h3><p>Потеря, которая оценивает неверную догадку в битах, один шаг, который сдвигает вес против градиента, и скрытый слой, который нужен XOR.</p><ol><li><a href="/ru/learn/a-loss-in-bits/">19 Потеря в битах</a></li><li><a href="/ru/learn/one-step-downhill/">20 Один шаг вниз по склону</a></li><li><a href="/ru/learn/xor-needs-a-bend/">21 XOR нужен излом</a></li></ol></li><li><h3>Модуль 8 · BitNet: тернарные сети</h3><p>Порог, который сжимает сумму обратно до трёх значений, один нейрон, который становится другой функцией при смене весов, и нейрон, который читает входы по 27 тритов за раз.</p><ol><li><a href="/ru/learn/back-to-three-values/">22 Обратно в три значения</a></li><li><a href="/ru/learn/same-neuron-new-weights/">23 Тот же нейрон, другие веса</a></li><li><a href="/ru/learn/a-neuron-in-chunks/">24 Нейрон по кускам</a></li></ol></li><li><h3>Модуль 9 · Тернарный MAC как чип</h3><p>Скалярное произведение 27 тритов как провода без регистра, та же сумма, которую регистр копит на каждом такте, и небольшая целая сеть в конце курса.</p><ol><li><a href="/ru/learn/a-dot-product-in-wires/">25 Скалярное произведение из проводов</a></li><li><a href="/ru/learn/add-it-up-every-clock/">26 Складывать на каждом такте</a></li><li><a href="/ru/learn/a-whole-network/">27 Целая сеть</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/one-edge-one-world/index.html b/apps/website/public/ru/learn/one-edge-one-world/index.html
index a9beec284d..377ac72f87 100644
--- a/apps/website/public/ru/learn/one-edge-one-world/index.html
+++ b/apps/website/public/ru/learn/one-edge-one-world/index.html
@@ -300,5 +300,6 @@ <h1>Один фронт, один мир</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое тактовый сигнал</h3><p>Один фронт, один мир: всё, что делит фронт тактового сигнала, делит и мир; период и джиттер настоящего фронта и место, где тактовый сигнал входит в плату.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/one-edge-one-world/">01 Один фронт, один мир</a></li><li><a href="/ru/learn/period-and-jitter/">02 Период, фронт и джиттер</a></li><li><a href="/ru/learn/where-the-clock-enters/">03 Где входит тактовый сигнал</a></li></ol></li><li><h3>Модуль 2 · Деревья тактового сигнала</h3><p>Перекос и задержка прохождения, глобальная сеть буферов и ловушка стробирования клока логикой.</p><ol><li><a href="/ru/learn/skew-and-insertion/">04 Перекос и задержка прохождения</a></li><li><a href="/ru/learn/global-buffers/">05 Глобальный буфер</a></li><li><a href="/ru/learn/gated-clock-trap/">06 Ловушка стробированного клока</a></li></ol></li><li><h3>Модуль 3 · PLL и MMCM</h3><p>Умножить и поделить один тактовый сигнал в другой, сдвинуть его фазу шагами VCO и какие тактовые сигналы анализатор считает родственными.</p><ol><li><a href="/ru/learn/multiply-and-divide/">07 Умножить и поделить</a></li><li><a href="/ru/learn/phase-and-lock/">08 Фаза и захват</a></li><li><a href="/ru/learn/related-or-not/">09 Родственные или чужие</a></li></ol></li><li><h3>Модуль 4 · Сбросы</h3><p>Ассертировать асинхронно, снимать синхронно: три вида сброса, конвейер снятия и дерево, которое растит сброс.</p><ol><li><a href="/ru/learn/sync-or-async/">10 Синхронный или асинхронный</a></li><li><a href="/ru/learn/reset-release/">11 Конвейер снятия</a></li><li><a href="/ru/learn/reset-fanout/">12 Разветвление сброса</a></li></ol></li><li><h3>Модуль 5 · Метастабильность</h3><p>Окно setup-hold, среднее время между отказами в целочисленной арифметике и два триггера, которые это чинят.</p><ol><li><a href="/ru/learn/setup-hold-window/">13 Окно setup-hold</a></li><li><a href="/ru/learn/mtbf/">14 Среднее время между отказами</a></li><li><a href="/ru/learn/two-flop-sync/">15 Двухтриггерный синхронизатор</a></li></ol></li><li><h3>Модуль 6 · Пересечение многих бит</h3><p>Почему двоичная шина рвётся, почему код Грея нет, и рукопожатие, переносящее импульс между мирами.</p><ol><li><a href="/ru/learn/why-binary-fails/">16 Почему двоичный код рвётся</a></li><li><a href="/ru/learn/gray-code/">17 Код Грея</a></li><li><a href="/ru/learn/handshakes-and-pulses/">18 Рукопожатия и импульсы</a></li></ol></li><li><h3>Модуль 7 · Асинхронный FIFO</h3><p>Указатели, флаги и глубина: буфер, переносящий поток между двумя тактовыми сигналами.</p><ol><li><a href="/ru/learn/pointers-and-flags/">19 Указатели и флаги</a></li><li><a href="/ru/learn/depth-sizing/">20 Выбор глубины</a></li><li><a href="/ru/learn/fifo-testbench/">21 Тестбенч FIFO</a></li></ol></li><li><h3>Модуль 8 · Ограничения</h3><p>Строки, сообщающие анализатору, что такое тактовый сигнал, какие пути не проверять и чему должны соответствовать выводы.</p><ol><li><a href="/ru/learn/create-clock/">22 Строка create_clock</a></li><li><a href="/ru/learn/false-path-and-max-delay/">23 False path и max delay</a></li><li><a href="/ru/learn/io-timing/">24 Тайминг выводов</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Отчёт CDC, одно пересечение, захваченное на триггерах, и дифф битстрима, замыкающий курс.</p><ol><li><a href="/ru/learn/cdc-report/">25 Отчёт CDC</a></li><li><a href="/ru/learn/capture-a-crossing/">26 Захватить пересечение</a></li><li><a href="/ru/learn/capstone-diff/">27 Капстоун: поменяй и сравни</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/one-lut-up-close/index.html b/apps/website/public/ru/learn/one-lut-up-close/index.html
index 8c7ca5e64b..ae5680c5b9 100644
--- a/apps/website/public/ru/learn/one-lut-up-close/index.html
+++ b/apps/website/public/ru/learn/one-lut-up-close/index.html
@@ -148,5 +148,6 @@ <h1>Одна LUT вблизи</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Чип</h3><p>Что такое FPGA, какой чип у нас и как его выводы встречаются с платой.</p><ol><li><a href="/ru/learn/fpga-is-a-grid/">01 FPGA — это сетка, которую вы настраиваете</a></li><li><a href="/ru/learn/which-chip/">02 Какой это чип?</a></li><li><a href="/ru/learn/pins-and-package/">03 Выводы, шарики и корпус</a></li></ol></li><li><h3>Модуль 2 · Числа в железе</h3><p>Биты, триты и форматы чисел, и сколько стоит арифметика в логике.</p><ol><li><a href="/ru/learn/bits-and-trits/">04 Биты, триты и GF16</a></li><li><a href="/ru/learn/what-arithmetic-costs/">05 Сколько стоит арифметика</a></li><li><a href="/ru/learn/how-much-fits/">06 Сколько помещается на чип</a></li></ol></li><li><h3>Модуль 3 · Ваша программа на t27</h3><p>Напишите спеку, проверьте её и увидьте, почему «компилируется» не значит «верно».</p><ol><li><a href="/ru/learn/hello-t27/">07 Привет, t27</a></li><li><a href="/ru/learn/tests-are-the-spec/">08 Тесты — часть спеки</a></li><li><a href="/ru/learn/compiles-is-not-correct/">09 Компилируется — не значит верно</a></li></ol></li><li><h3>Модуль 4 · Внутри t27c</h3><p>Как компилятор читает спеку и что он пишет, включая нативный t27b.</p><ol><li><a href="/ru/learn/compiler-stages/">10 Стадии t27c</a></li><li><a href="/ru/learn/t27c-on-a-real-machine/">11 t27c на настоящей машине</a></li><li><a href="/ru/learn/seven-backends-and-t27b/">12 Семь бэкендов и t27b</a></li></ol></li><li><h3>Модуль 5 · От спеки к железу</h3><p>Verilog, который пишет t27c, ячейки, в которые он превращается, и что делает одна LUT.</p><ol><li><a href="/ru/learn/spec-to-verilog/">13 От спеки к Verilog</a></li><li><a href="/ru/learn/count-the-cells/">14 Посчитайте ячейки</a></li><li class="on" aria-current="page"><a href="/ru/learn/one-lut-up-close/">15 Одна LUT вблизи</a></li></ol></li><li><h3>Модуль 6 · Читаем синтез</h3><p>Что yosys сообщает о вашем дизайне и какие предупреждения важны.</p><ol><li><a href="/ru/learn/synthesis-warnings/">16 Важные предупреждения синтеза</a></li><li><a href="/ru/learn/state-machines/">17 Конечные автоматы</a></li><li><a href="/ru/learn/where-the-luts-go/">18 Куда уходят LUT</a></li></ol></li><li><h3>Модуль 7 · Размещение, трассировка, тайминг</h3><p>Где ячейки оказываются на кристалле и успевает ли тактовая частота.</p><ol><li><a href="/ru/learn/placement/">19 Размещение: где оказываются ячейки</a></li><li><a href="/ru/learn/logic-depth/">20 Глубина логики</a></li><li><a href="/ru/learn/timing-slack/">21 Тайминг и запас</a></li></ol></li><li><h3>Модуль 8 · Битстрим</h3><p>Как оттрассированный дизайн становится битами, которые загружает чип, без Vivado.</p><ol><li><a href="/ru/learn/fasm-to-frames/">22 От FASM к фреймам</a></li><li><a href="/ru/learn/a-build-receipt/">23 Сборка по пунктам</a></li><li><a href="/ru/learn/read-a-bitstream/">24 Читаем битстрим</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Спросите чип, кто он, загрузите биты и проверьте их.</p><ol><li><a href="/ru/learn/jtag-asks-the-chip/">25 JTAG спрашивает чип</a></li><li><a href="/ru/learn/same-bits-on-the-board/">26 Те же биты — на плате</a></li><li><a href="/ru/learn/capstone/">27 Итог: переугадайте yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/one-outlier/index.html b/apps/website/public/ru/learn/one-outlier/index.html
index c2918440c4..068a837f67 100644
--- a/apps/website/public/ru/learn/one-outlier/index.html
+++ b/apps/website/public/ru/learn/one-outlier/index.html
@@ -375,5 +375,6 @@ <h1>Один выброс, много нулей</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Лаборатория: наши исследования</h3><p>Свой формат чисел, честная таблица результатов и таблицы модели, перемноженные на плате.</p><ol><li><a href="/ru/learn/t27-number-format/">01 T27: наш собственный формат чисел</a></li><li><a href="/ru/learn/an-honest-scoreboard/">02 Честная таблица результатов</a></li><li><a href="/ru/learn/a-model-on-the-board/">03 Таблицы модели на плате</a></li></ol></li><li><h3>Модуль 2 · ИИ-числа: блок MX</h3><p>Как ИИ-чипы хранят веса в нескольких битах: один общий масштаб на блок, сам байт масштаба и что один выброс делает с соседями.</p><ol><li><a href="/ru/learn/one-scale-per-block/">04 Один масштаб на блок весов</a></li><li><a href="/ru/learn/scale-byte-native/">05 Байт масштаба на настоящей машине</a></li><li class="on" aria-current="page"><a href="/ru/learn/one-outlier/">06 Один выброс, много нулей</a></li></ol></li><li><h3>Модуль 3 · Тернарные веса</h3><p>Веса, которые бывают только минус масштаб, ноль или плюс масштаб, пять правил, которые должен пройти тернарный алфавит, и прогон тестов, который ничего не проверил.</p><ol><li><a href="/ru/learn/phi-scaled-ternary/">07 Веса минус фи, ноль и плюс фи</a></li><li><a href="/ru/learn/a-pass-that-checked-nothing/">08 Прогон, который ничего не проверил</a></li><li><a href="/ru/learn/five-rules-for-weights/">09 Пять правил для алфавита весов</a></li></ol></li><li><h3>Модуль 4 · Тернарный формат Ternary Network Float</h3><p>Правило, которое компилятор проверяет до запуска любого теста, и 17-битное число с плавающей точкой, порядок которого — четыре сбалансированных трита.</p><ol><li><a href="/ru/learn/blocked-is-not-failed/">10 Заблокирован — не значит провален</a></li><li><a href="/ru/learn/four-trit-exponent/">11 Число, у которого порядок — четыре трита</a></li><li><a href="/ru/learn/negate-twice/">12 Дважды сменить знак — получить число обратно</a></li></ol></li><li><h3>Модуль 5 · Арифметика чисел со знаком</h3><p>Умножить два числа со знаком, сложить их, когда знаки разные, и сделать то и другое сразу в умножении с накоплением.</p><ol><li><a href="/ru/learn/a-sign-is-one-bit/">13 Знак — это один бит</a></li><li><a href="/ru/learn/opposite-signs-subtract/">14 Разные знаки — вычитание</a></li><li><a href="/ru/learn/multiply-then-add/">15 Умножить, потом сложить</a></li></ol></li><li><h3>Модуль 6 · Части нейрона</h3><p>ReLU с изломом в нуле, степень двойки для softmax и argmax, который называет ответ.</p><ol><li><a href="/ru/learn/a-bend-at-zero/">16 Излом в нуле</a></li><li><a href="/ru/learn/two-to-the-x/">17 Два в степени x</a></li><li><a href="/ru/learn/pick-the-largest/">18 Выбрать наибольшее</a></li></ol></li><li><h3>Модуль 7 · Обучение на ошибке</h3><p>Потеря, которая оценивает неверную догадку в битах, один шаг, который сдвигает вес против градиента, и скрытый слой, который нужен XOR.</p><ol><li><a href="/ru/learn/a-loss-in-bits/">19 Потеря в битах</a></li><li><a href="/ru/learn/one-step-downhill/">20 Один шаг вниз по склону</a></li><li><a href="/ru/learn/xor-needs-a-bend/">21 XOR нужен излом</a></li></ol></li><li><h3>Модуль 8 · BitNet: тернарные сети</h3><p>Порог, который сжимает сумму обратно до трёх значений, один нейрон, который становится другой функцией при смене весов, и нейрон, который читает входы по 27 тритов за раз.</p><ol><li><a href="/ru/learn/back-to-three-values/">22 Обратно в три значения</a></li><li><a href="/ru/learn/same-neuron-new-weights/">23 Тот же нейрон, другие веса</a></li><li><a href="/ru/learn/a-neuron-in-chunks/">24 Нейрон по кускам</a></li></ol></li><li><h3>Модуль 9 · Тернарный MAC как чип</h3><p>Скалярное произведение 27 тритов как провода без регистра, та же сумма, которую регистр копит на каждом такте, и небольшая целая сеть в конце курса.</p><ol><li><a href="/ru/learn/a-dot-product-in-wires/">25 Скалярное произведение из проводов</a></li><li><a href="/ru/learn/add-it-up-every-clock/">26 Складывать на каждом такте</a></li><li><a href="/ru/learn/a-whole-network/">27 Целая сеть</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/one-scale-per-block/index.html b/apps/website/public/ru/learn/one-scale-per-block/index.html
index 68503af2fc..b5ca92ed24 100644
--- a/apps/website/public/ru/learn/one-scale-per-block/index.html
+++ b/apps/website/public/ru/learn/one-scale-per-block/index.html
@@ -565,5 +565,6 @@ <h1>Один масштаб на блок весов</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Лаборатория: наши исследования</h3><p>Свой формат чисел, честная таблица результатов и таблицы модели, перемноженные на плате.</p><ol><li><a href="/ru/learn/t27-number-format/">01 T27: наш собственный формат чисел</a></li><li><a href="/ru/learn/an-honest-scoreboard/">02 Честная таблица результатов</a></li><li><a href="/ru/learn/a-model-on-the-board/">03 Таблицы модели на плате</a></li></ol></li><li><h3>Модуль 2 · ИИ-числа: блок MX</h3><p>Как ИИ-чипы хранят веса в нескольких битах: один общий масштаб на блок, сам байт масштаба и что один выброс делает с соседями.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/one-scale-per-block/">04 Один масштаб на блок весов</a></li><li><a href="/ru/learn/scale-byte-native/">05 Байт масштаба на настоящей машине</a></li><li><a href="/ru/learn/one-outlier/">06 Один выброс, много нулей</a></li></ol></li><li><h3>Модуль 3 · Тернарные веса</h3><p>Веса, которые бывают только минус масштаб, ноль или плюс масштаб, пять правил, которые должен пройти тернарный алфавит, и прогон тестов, который ничего не проверил.</p><ol><li><a href="/ru/learn/phi-scaled-ternary/">07 Веса минус фи, ноль и плюс фи</a></li><li><a href="/ru/learn/a-pass-that-checked-nothing/">08 Прогон, который ничего не проверил</a></li><li><a href="/ru/learn/five-rules-for-weights/">09 Пять правил для алфавита весов</a></li></ol></li><li><h3>Модуль 4 · Тернарный формат Ternary Network Float</h3><p>Правило, которое компилятор проверяет до запуска любого теста, и 17-битное число с плавающей точкой, порядок которого — четыре сбалансированных трита.</p><ol><li><a href="/ru/learn/blocked-is-not-failed/">10 Заблокирован — не значит провален</a></li><li><a href="/ru/learn/four-trit-exponent/">11 Число, у которого порядок — четыре трита</a></li><li><a href="/ru/learn/negate-twice/">12 Дважды сменить знак — получить число обратно</a></li></ol></li><li><h3>Модуль 5 · Арифметика чисел со знаком</h3><p>Умножить два числа со знаком, сложить их, когда знаки разные, и сделать то и другое сразу в умножении с накоплением.</p><ol><li><a href="/ru/learn/a-sign-is-one-bit/">13 Знак — это один бит</a></li><li><a href="/ru/learn/opposite-signs-subtract/">14 Разные знаки — вычитание</a></li><li><a href="/ru/learn/multiply-then-add/">15 Умножить, потом сложить</a></li></ol></li><li><h3>Модуль 6 · Части нейрона</h3><p>ReLU с изломом в нуле, степень двойки для softmax и argmax, который называет ответ.</p><ol><li><a href="/ru/learn/a-bend-at-zero/">16 Излом в нуле</a></li><li><a href="/ru/learn/two-to-the-x/">17 Два в степени x</a></li><li><a href="/ru/learn/pick-the-largest/">18 Выбрать наибольшее</a></li></ol></li><li><h3>Модуль 7 · Обучение на ошибке</h3><p>Потеря, которая оценивает неверную догадку в битах, один шаг, который сдвигает вес против градиента, и скрытый слой, который нужен XOR.</p><ol><li><a href="/ru/learn/a-loss-in-bits/">19 Потеря в битах</a></li><li><a href="/ru/learn/one-step-downhill/">20 Один шаг вниз по склону</a></li><li><a href="/ru/learn/xor-needs-a-bend/">21 XOR нужен излом</a></li></ol></li><li><h3>Модуль 8 · BitNet: тернарные сети</h3><p>Порог, который сжимает сумму обратно до трёх значений, один нейрон, который становится другой функцией при смене весов, и нейрон, который читает входы по 27 тритов за раз.</p><ol><li><a href="/ru/learn/back-to-three-values/">22 Обратно в три значения</a></li><li><a href="/ru/learn/same-neuron-new-weights/">23 Тот же нейрон, другие веса</a></li><li><a href="/ru/learn/a-neuron-in-chunks/">24 Нейрон по кускам</a></li></ol></li><li><h3>Модуль 9 · Тернарный MAC как чип</h3><p>Скалярное произведение 27 тритов как провода без регистра, та же сумма, которую регистр копит на каждом такте, и небольшая целая сеть в конце курса.</p><ol><li><a href="/ru/learn/a-dot-product-in-wires/">25 Скалярное произведение из проводов</a></li><li><a href="/ru/learn/add-it-up-every-clock/">26 Складывать на каждом такте</a></li><li><a href="/ru/learn/a-whole-network/">27 Целая сеть</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/one-step-downhill/index.html b/apps/website/public/ru/learn/one-step-downhill/index.html
index 638e85a747..b8a1ea768c 100644
--- a/apps/website/public/ru/learn/one-step-downhill/index.html
+++ b/apps/website/public/ru/learn/one-step-downhill/index.html
@@ -198,5 +198,6 @@ <h1>Один шаг вниз по склону</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Лаборатория: наши исследования</h3><p>Свой формат чисел, честная таблица результатов и таблицы модели, перемноженные на плате.</p><ol><li><a href="/ru/learn/t27-number-format/">01 T27: наш собственный формат чисел</a></li><li><a href="/ru/learn/an-honest-scoreboard/">02 Честная таблица результатов</a></li><li><a href="/ru/learn/a-model-on-the-board/">03 Таблицы модели на плате</a></li></ol></li><li><h3>Модуль 2 · ИИ-числа: блок MX</h3><p>Как ИИ-чипы хранят веса в нескольких битах: один общий масштаб на блок, сам байт масштаба и что один выброс делает с соседями.</p><ol><li><a href="/ru/learn/one-scale-per-block/">04 Один масштаб на блок весов</a></li><li><a href="/ru/learn/scale-byte-native/">05 Байт масштаба на настоящей машине</a></li><li><a href="/ru/learn/one-outlier/">06 Один выброс, много нулей</a></li></ol></li><li><h3>Модуль 3 · Тернарные веса</h3><p>Веса, которые бывают только минус масштаб, ноль или плюс масштаб, пять правил, которые должен пройти тернарный алфавит, и прогон тестов, который ничего не проверил.</p><ol><li><a href="/ru/learn/phi-scaled-ternary/">07 Веса минус фи, ноль и плюс фи</a></li><li><a href="/ru/learn/a-pass-that-checked-nothing/">08 Прогон, который ничего не проверил</a></li><li><a href="/ru/learn/five-rules-for-weights/">09 Пять правил для алфавита весов</a></li></ol></li><li><h3>Модуль 4 · Тернарный формат Ternary Network Float</h3><p>Правило, которое компилятор проверяет до запуска любого теста, и 17-битное число с плавающей точкой, порядок которого — четыре сбалансированных трита.</p><ol><li><a href="/ru/learn/blocked-is-not-failed/">10 Заблокирован — не значит провален</a></li><li><a href="/ru/learn/four-trit-exponent/">11 Число, у которого порядок — четыре трита</a></li><li><a href="/ru/learn/negate-twice/">12 Дважды сменить знак — получить число обратно</a></li></ol></li><li><h3>Модуль 5 · Арифметика чисел со знаком</h3><p>Умножить два числа со знаком, сложить их, когда знаки разные, и сделать то и другое сразу в умножении с накоплением.</p><ol><li><a href="/ru/learn/a-sign-is-one-bit/">13 Знак — это один бит</a></li><li><a href="/ru/learn/opposite-signs-subtract/">14 Разные знаки — вычитание</a></li><li><a href="/ru/learn/multiply-then-add/">15 Умножить, потом сложить</a></li></ol></li><li><h3>Модуль 6 · Части нейрона</h3><p>ReLU с изломом в нуле, степень двойки для softmax и argmax, который называет ответ.</p><ol><li><a href="/ru/learn/a-bend-at-zero/">16 Излом в нуле</a></li><li><a href="/ru/learn/two-to-the-x/">17 Два в степени x</a></li><li><a href="/ru/learn/pick-the-largest/">18 Выбрать наибольшее</a></li></ol></li><li><h3>Модуль 7 · Обучение на ошибке</h3><p>Потеря, которая оценивает неверную догадку в битах, один шаг, который сдвигает вес против градиента, и скрытый слой, который нужен XOR.</p><ol><li><a href="/ru/learn/a-loss-in-bits/">19 Потеря в битах</a></li><li class="on" aria-current="page"><a href="/ru/learn/one-step-downhill/">20 Один шаг вниз по склону</a></li><li><a href="/ru/learn/xor-needs-a-bend/">21 XOR нужен излом</a></li></ol></li><li><h3>Модуль 8 · BitNet: тернарные сети</h3><p>Порог, который сжимает сумму обратно до трёх значений, один нейрон, который становится другой функцией при смене весов, и нейрон, который читает входы по 27 тритов за раз.</p><ol><li><a href="/ru/learn/back-to-three-values/">22 Обратно в три значения</a></li><li><a href="/ru/learn/same-neuron-new-weights/">23 Тот же нейрон, другие веса</a></li><li><a href="/ru/learn/a-neuron-in-chunks/">24 Нейрон по кускам</a></li></ol></li><li><h3>Модуль 9 · Тернарный MAC как чип</h3><p>Скалярное произведение 27 тритов как провода без регистра, та же сумма, которую регистр копит на каждом такте, и небольшая целая сеть в конце курса.</p><ol><li><a href="/ru/learn/a-dot-product-in-wires/">25 Скалярное произведение из проводов</a></li><li><a href="/ru/learn/add-it-up-every-clock/">26 Складывать на каждом такте</a></li><li><a href="/ru/learn/a-whole-network/">27 Целая сеть</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/one-ternary-neuron/index.html b/apps/website/public/ru/learn/one-ternary-neuron/index.html
index f3aa7c1afe..b699e72c9c 100644
--- a/apps/website/public/ru/learn/one-ternary-neuron/index.html
+++ b/apps/website/public/ru/learn/one-ternary-neuron/index.html
@@ -769,5 +769,6 @@ <h1>Один троичный нейрон</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Три значения</h3><p>Почему три, как сбалансированная троичная система пишет любое число без знака и что дают переворот и отсечение тритов.</p><ol><li><a href="/ru/learn/why-three/">01 Почему три</a></li><li><a href="/ru/learn/balanced-ternary/">02 Сбалансированная троичная</a></li><li><a href="/ru/learn/flip-and-cut/">03 Переворот вместо минуса, отсечение вместо округления</a></li></ol></li><li><h3>Модуль 2 · Логика с «не знаю»</h3><p>Трёхзначные вентили Клини, два вентиля без двоичного двойника и сложение как пара таблиц.</p><ol><li><a href="/ru/learn/kleene-logic/">04 Трёхзначная логика</a></li><li><a href="/ru/learn/consensus-and-any/">05 Консенсус и «любой»</a></li><li><a href="/ru/learn/sum-and-carry/">06 Сумма и перенос</a></li></ol></li><li><h3>Модуль 3 · Складываем триты</h3><p>Полусумматор, полный сумматор и перенос, бегущий влево разряд за разрядом.</p><ol><li><a href="/ru/learn/half-adder/">07 Полусумматор</a></li><li><a href="/ru/learn/full-adder/">08 Полный сумматор</a></li><li><a href="/ru/learn/ripple-carry/">09 Сквозной перенос</a></li></ol></li><li><h3>Модуль 4 · Умножение без умножителя</h3><p>Копия, ноль или переворот: произведение на один трит, умножение столбиком и MAC, который только складывает.</p><ol><li><a href="/ru/learn/times-a-trit/">10 Умножить на один трит</a></li><li><a href="/ru/learn/long-multiplication/">11 Умножение столбиком</a></li><li><a href="/ru/learn/ternary-mac/">12 Троичный MAC</a></li></ol></li><li><h3>Модуль 5 · Триты в двоичной памяти</h3><p>Два бита на трит, пять тритов на байт и сколько бит нужно слову из 27 тритов.</p><ol><li><a href="/ru/learn/two-bits-per-trit/">13 Два бита на трит</a></li><li><a href="/ru/learn/five-trits-a-byte/">14 Пять тритов в байте</a></li><li><a href="/ru/learn/a-word-of-27/">15 Слово из 27 тритов</a></li></ol></li><li><h3>Модуль 6 · Командное слово TRI-27</h3><p>32-битное слово, которое декодирует эмулятор Trinity: его поля, 47 опкодов и 15-битное непосредственное значение.</p><ol><li><a href="/ru/learn/the-instruction-word/">16 Командное слово</a></li><li><a href="/ru/learn/the-opcode-map/">17 Карта опкодов</a></li><li><a href="/ru/learn/the-immediate/">18 Непосредственное значение</a></li></ol></li><li><h3>Модуль 7 · Троичная машина</h3><p>АЛУ из сумматоров этого курса, переход на три стороны и программа, которую можно пройти по шагам.</p><ol><li><a href="/ru/learn/a-ternary-alu/">19 Троичное АЛУ</a></li><li><a href="/ru/learn/three-way-jump/">20 Переход на три стороны</a></li><li><a href="/ru/learn/run-a-program/">21 Запустить программу</a></li></ol></li><li><h3>Модуль 8 · Три ответа</h3><p>Сравнение по старшему различному триту, поиск числа по третям и сортировка трёхзначными сравнениями.</p><ol><li><a href="/ru/learn/compare-at-the-top/">22 Сравнение сверху</a></li><li><a href="/ru/learn/questions-in-thirds/">23 Вопросы по третям</a></li><li><a href="/ru/learn/sort-by-three/">24 Сортировка трёхзначными сравнениями</a></li></ol></li><li><h3>Модуль 9 · Троичные нейроны</h3><p>Веса -1, 0, +1: один нейрон, детектор и маленький слой, и нигде нет умножителя.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/one-ternary-neuron/">25 Один троичный нейрон</a></li><li><a href="/ru/learn/a-bar-detector/">26 Детектор полосы</a></li><li><a href="/ru/learn/a-tiny-layer/">27 Маленький слой</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/opposite-signs-subtract/index.html b/apps/website/public/ru/learn/opposite-signs-subtract/index.html
index aed422d280..c0098f9c2b 100644
--- a/apps/website/public/ru/learn/opposite-signs-subtract/index.html
+++ b/apps/website/public/ru/learn/opposite-signs-subtract/index.html
@@ -191,5 +191,6 @@ <h1>Разные знаки — вычитание</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Лаборатория: наши исследования</h3><p>Свой формат чисел, честная таблица результатов и таблицы модели, перемноженные на плате.</p><ol><li><a href="/ru/learn/t27-number-format/">01 T27: наш собственный формат чисел</a></li><li><a href="/ru/learn/an-honest-scoreboard/">02 Честная таблица результатов</a></li><li><a href="/ru/learn/a-model-on-the-board/">03 Таблицы модели на плате</a></li></ol></li><li><h3>Модуль 2 · ИИ-числа: блок MX</h3><p>Как ИИ-чипы хранят веса в нескольких битах: один общий масштаб на блок, сам байт масштаба и что один выброс делает с соседями.</p><ol><li><a href="/ru/learn/one-scale-per-block/">04 Один масштаб на блок весов</a></li><li><a href="/ru/learn/scale-byte-native/">05 Байт масштаба на настоящей машине</a></li><li><a href="/ru/learn/one-outlier/">06 Один выброс, много нулей</a></li></ol></li><li><h3>Модуль 3 · Тернарные веса</h3><p>Веса, которые бывают только минус масштаб, ноль или плюс масштаб, пять правил, которые должен пройти тернарный алфавит, и прогон тестов, который ничего не проверил.</p><ol><li><a href="/ru/learn/phi-scaled-ternary/">07 Веса минус фи, ноль и плюс фи</a></li><li><a href="/ru/learn/a-pass-that-checked-nothing/">08 Прогон, который ничего не проверил</a></li><li><a href="/ru/learn/five-rules-for-weights/">09 Пять правил для алфавита весов</a></li></ol></li><li><h3>Модуль 4 · Тернарный формат Ternary Network Float</h3><p>Правило, которое компилятор проверяет до запуска любого теста, и 17-битное число с плавающей точкой, порядок которого — четыре сбалансированных трита.</p><ol><li><a href="/ru/learn/blocked-is-not-failed/">10 Заблокирован — не значит провален</a></li><li><a href="/ru/learn/four-trit-exponent/">11 Число, у которого порядок — четыре трита</a></li><li><a href="/ru/learn/negate-twice/">12 Дважды сменить знак — получить число обратно</a></li></ol></li><li><h3>Модуль 5 · Арифметика чисел со знаком</h3><p>Умножить два числа со знаком, сложить их, когда знаки разные, и сделать то и другое сразу в умножении с накоплением.</p><ol><li><a href="/ru/learn/a-sign-is-one-bit/">13 Знак — это один бит</a></li><li class="on" aria-current="page"><a href="/ru/learn/opposite-signs-subtract/">14 Разные знаки — вычитание</a></li><li><a href="/ru/learn/multiply-then-add/">15 Умножить, потом сложить</a></li></ol></li><li><h3>Модуль 6 · Части нейрона</h3><p>ReLU с изломом в нуле, степень двойки для softmax и argmax, который называет ответ.</p><ol><li><a href="/ru/learn/a-bend-at-zero/">16 Излом в нуле</a></li><li><a href="/ru/learn/two-to-the-x/">17 Два в степени x</a></li><li><a href="/ru/learn/pick-the-largest/">18 Выбрать наибольшее</a></li></ol></li><li><h3>Модуль 7 · Обучение на ошибке</h3><p>Потеря, которая оценивает неверную догадку в битах, один шаг, который сдвигает вес против градиента, и скрытый слой, который нужен XOR.</p><ol><li><a href="/ru/learn/a-loss-in-bits/">19 Потеря в битах</a></li><li><a href="/ru/learn/one-step-downhill/">20 Один шаг вниз по склону</a></li><li><a href="/ru/learn/xor-needs-a-bend/">21 XOR нужен излом</a></li></ol></li><li><h3>Модуль 8 · BitNet: тернарные сети</h3><p>Порог, который сжимает сумму обратно до трёх значений, один нейрон, который становится другой функцией при смене весов, и нейрон, который читает входы по 27 тритов за раз.</p><ol><li><a href="/ru/learn/back-to-three-values/">22 Обратно в три значения</a></li><li><a href="/ru/learn/same-neuron-new-weights/">23 Тот же нейрон, другие веса</a></li><li><a href="/ru/learn/a-neuron-in-chunks/">24 Нейрон по кускам</a></li></ol></li><li><h3>Модуль 9 · Тернарный MAC как чип</h3><p>Скалярное произведение 27 тритов как провода без регистра, та же сумма, которую регистр копит на каждом такте, и небольшая целая сеть в конце курса.</p><ol><li><a href="/ru/learn/a-dot-product-in-wires/">25 Скалярное произведение из проводов</a></li><li><a href="/ru/learn/add-it-up-every-clock/">26 Складывать на каждом такте</a></li><li><a href="/ru/learn/a-whole-network/">27 Целая сеть</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/period-and-jitter/index.html b/apps/website/public/ru/learn/period-and-jitter/index.html
index 0442943d9e..aa46afcbbe 100644
--- a/apps/website/public/ru/learn/period-and-jitter/index.html
+++ b/apps/website/public/ru/learn/period-and-jitter/index.html
@@ -107,5 +107,6 @@ <h1>Период, фронт и джиттер</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое тактовый сигнал</h3><p>Один фронт, один мир: всё, что делит фронт тактового сигнала, делит и мир; период и джиттер настоящего фронта и место, где тактовый сигнал входит в плату.</p><ol><li><a href="/ru/learn/one-edge-one-world/">01 Один фронт, один мир</a></li><li class="on" aria-current="page"><a href="/ru/learn/period-and-jitter/">02 Период, фронт и джиттер</a></li><li><a href="/ru/learn/where-the-clock-enters/">03 Где входит тактовый сигнал</a></li></ol></li><li><h3>Модуль 2 · Деревья тактового сигнала</h3><p>Перекос и задержка прохождения, глобальная сеть буферов и ловушка стробирования клока логикой.</p><ol><li><a href="/ru/learn/skew-and-insertion/">04 Перекос и задержка прохождения</a></li><li><a href="/ru/learn/global-buffers/">05 Глобальный буфер</a></li><li><a href="/ru/learn/gated-clock-trap/">06 Ловушка стробированного клока</a></li></ol></li><li><h3>Модуль 3 · PLL и MMCM</h3><p>Умножить и поделить один тактовый сигнал в другой, сдвинуть его фазу шагами VCO и какие тактовые сигналы анализатор считает родственными.</p><ol><li><a href="/ru/learn/multiply-and-divide/">07 Умножить и поделить</a></li><li><a href="/ru/learn/phase-and-lock/">08 Фаза и захват</a></li><li><a href="/ru/learn/related-or-not/">09 Родственные или чужие</a></li></ol></li><li><h3>Модуль 4 · Сбросы</h3><p>Ассертировать асинхронно, снимать синхронно: три вида сброса, конвейер снятия и дерево, которое растит сброс.</p><ol><li><a href="/ru/learn/sync-or-async/">10 Синхронный или асинхронный</a></li><li><a href="/ru/learn/reset-release/">11 Конвейер снятия</a></li><li><a href="/ru/learn/reset-fanout/">12 Разветвление сброса</a></li></ol></li><li><h3>Модуль 5 · Метастабильность</h3><p>Окно setup-hold, среднее время между отказами в целочисленной арифметике и два триггера, которые это чинят.</p><ol><li><a href="/ru/learn/setup-hold-window/">13 Окно setup-hold</a></li><li><a href="/ru/learn/mtbf/">14 Среднее время между отказами</a></li><li><a href="/ru/learn/two-flop-sync/">15 Двухтриггерный синхронизатор</a></li></ol></li><li><h3>Модуль 6 · Пересечение многих бит</h3><p>Почему двоичная шина рвётся, почему код Грея нет, и рукопожатие, переносящее импульс между мирами.</p><ol><li><a href="/ru/learn/why-binary-fails/">16 Почему двоичный код рвётся</a></li><li><a href="/ru/learn/gray-code/">17 Код Грея</a></li><li><a href="/ru/learn/handshakes-and-pulses/">18 Рукопожатия и импульсы</a></li></ol></li><li><h3>Модуль 7 · Асинхронный FIFO</h3><p>Указатели, флаги и глубина: буфер, переносящий поток между двумя тактовыми сигналами.</p><ol><li><a href="/ru/learn/pointers-and-flags/">19 Указатели и флаги</a></li><li><a href="/ru/learn/depth-sizing/">20 Выбор глубины</a></li><li><a href="/ru/learn/fifo-testbench/">21 Тестбенч FIFO</a></li></ol></li><li><h3>Модуль 8 · Ограничения</h3><p>Строки, сообщающие анализатору, что такое тактовый сигнал, какие пути не проверять и чему должны соответствовать выводы.</p><ol><li><a href="/ru/learn/create-clock/">22 Строка create_clock</a></li><li><a href="/ru/learn/false-path-and-max-delay/">23 False path и max delay</a></li><li><a href="/ru/learn/io-timing/">24 Тайминг выводов</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Отчёт CDC, одно пересечение, захваченное на триггерах, и дифф битстрима, замыкающий курс.</p><ol><li><a href="/ru/learn/cdc-report/">25 Отчёт CDC</a></li><li><a href="/ru/learn/capture-a-crossing/">26 Захватить пересечение</a></li><li><a href="/ru/learn/capstone-diff/">27 Капстоун: поменяй и сравни</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/phase-and-lock/index.html b/apps/website/public/ru/learn/phase-and-lock/index.html
index a0c0e19b60..8515a97584 100644
--- a/apps/website/public/ru/learn/phase-and-lock/index.html
+++ b/apps/website/public/ru/learn/phase-and-lock/index.html
@@ -341,5 +341,6 @@ <h1>Фаза и захват</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое тактовый сигнал</h3><p>Один фронт, один мир: всё, что делит фронт тактового сигнала, делит и мир; период и джиттер настоящего фронта и место, где тактовый сигнал входит в плату.</p><ol><li><a href="/ru/learn/one-edge-one-world/">01 Один фронт, один мир</a></li><li><a href="/ru/learn/period-and-jitter/">02 Период, фронт и джиттер</a></li><li><a href="/ru/learn/where-the-clock-enters/">03 Где входит тактовый сигнал</a></li></ol></li><li><h3>Модуль 2 · Деревья тактового сигнала</h3><p>Перекос и задержка прохождения, глобальная сеть буферов и ловушка стробирования клока логикой.</p><ol><li><a href="/ru/learn/skew-and-insertion/">04 Перекос и задержка прохождения</a></li><li><a href="/ru/learn/global-buffers/">05 Глобальный буфер</a></li><li><a href="/ru/learn/gated-clock-trap/">06 Ловушка стробированного клока</a></li></ol></li><li><h3>Модуль 3 · PLL и MMCM</h3><p>Умножить и поделить один тактовый сигнал в другой, сдвинуть его фазу шагами VCO и какие тактовые сигналы анализатор считает родственными.</p><ol><li><a href="/ru/learn/multiply-and-divide/">07 Умножить и поделить</a></li><li class="on" aria-current="page"><a href="/ru/learn/phase-and-lock/">08 Фаза и захват</a></li><li><a href="/ru/learn/related-or-not/">09 Родственные или чужие</a></li></ol></li><li><h3>Модуль 4 · Сбросы</h3><p>Ассертировать асинхронно, снимать синхронно: три вида сброса, конвейер снятия и дерево, которое растит сброс.</p><ol><li><a href="/ru/learn/sync-or-async/">10 Синхронный или асинхронный</a></li><li><a href="/ru/learn/reset-release/">11 Конвейер снятия</a></li><li><a href="/ru/learn/reset-fanout/">12 Разветвление сброса</a></li></ol></li><li><h3>Модуль 5 · Метастабильность</h3><p>Окно setup-hold, среднее время между отказами в целочисленной арифметике и два триггера, которые это чинят.</p><ol><li><a href="/ru/learn/setup-hold-window/">13 Окно setup-hold</a></li><li><a href="/ru/learn/mtbf/">14 Среднее время между отказами</a></li><li><a href="/ru/learn/two-flop-sync/">15 Двухтриггерный синхронизатор</a></li></ol></li><li><h3>Модуль 6 · Пересечение многих бит</h3><p>Почему двоичная шина рвётся, почему код Грея нет, и рукопожатие, переносящее импульс между мирами.</p><ol><li><a href="/ru/learn/why-binary-fails/">16 Почему двоичный код рвётся</a></li><li><a href="/ru/learn/gray-code/">17 Код Грея</a></li><li><a href="/ru/learn/handshakes-and-pulses/">18 Рукопожатия и импульсы</a></li></ol></li><li><h3>Модуль 7 · Асинхронный FIFO</h3><p>Указатели, флаги и глубина: буфер, переносящий поток между двумя тактовыми сигналами.</p><ol><li><a href="/ru/learn/pointers-and-flags/">19 Указатели и флаги</a></li><li><a href="/ru/learn/depth-sizing/">20 Выбор глубины</a></li><li><a href="/ru/learn/fifo-testbench/">21 Тестбенч FIFO</a></li></ol></li><li><h3>Модуль 8 · Ограничения</h3><p>Строки, сообщающие анализатору, что такое тактовый сигнал, какие пути не проверять и чему должны соответствовать выводы.</p><ol><li><a href="/ru/learn/create-clock/">22 Строка create_clock</a></li><li><a href="/ru/learn/false-path-and-max-delay/">23 False path и max delay</a></li><li><a href="/ru/learn/io-timing/">24 Тайминг выводов</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Отчёт CDC, одно пересечение, захваченное на триггерах, и дифф битстрима, замыкающий курс.</p><ol><li><a href="/ru/learn/cdc-report/">25 Отчёт CDC</a></li><li><a href="/ru/learn/capture-a-crossing/">26 Захватить пересечение</a></li><li><a href="/ru/learn/capstone-diff/">27 Капстоун: поменяй и сравни</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/phi-as-a-ratio/index.html b/apps/website/public/ru/learn/phi-as-a-ratio/index.html
index a9c6867495..6511c57b35 100644
--- a/apps/website/public/ru/learn/phi-as-a-ratio/index.html
+++ b/apps/website/public/ru/learn/phi-as-a-ratio/index.html
@@ -828,5 +828,6 @@ <h1>Phi как отношение</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Правило и его числа</h3><p>Одно правило делит каждую ширину, отношение, к которому оно стремится, и числа Люка за тройкой 3.</p><ol><li><a href="/ru/learn/a-float-cut-by-phi/">01 Число с плавающей точкой, разрезанное по phi</a></li><li class="on" aria-current="page"><a href="/ru/learn/phi-as-a-ratio/">02 Phi как отношение</a></li><li><a href="/ru/learn/lucas-numbers-stay-whole/">03 Числа Люка остаются целыми</a></li></ol></li><li><h3>Модуль 2 · Почему phi, почему три</h3><p>Почему деление идёт по phi, почему основание три и как спека проверяет, что GF16 хранит phi.</p><ol><li><a href="/ru/learn/why-the-split-is-phi/">04 Почему деление по phi</a></li><li><a href="/ru/learn/why-base-three/">05 Почему основание три</a></li><li><a href="/ru/learn/phi-in-sixteen-bits/">06 Phi в шестнадцати битах</a></li></ol></li><li><h3>Модуль 3 · Малые ступени: от GF4 до GF8</h3><p>GF4, GF6 и GF8 — меньше всего битов, и округление до целых битов стоит здесь дороже всего.</p><ol><li><a href="/ru/learn/gf4-four-bits/">07 GF4: четыре бита</a></li><li><a href="/ru/learn/gf6-six-bits/">08 GF6: шесть битов</a></li><li><a href="/ru/learn/gf8-one-byte/">09 GF8: один байт</a></li></ol></li><li><h3>Модуль 4 · От десяти до четырнадцати битов</h3><p>GF10, GF12 и GF14 и то, как расстояние до 1 / phi меняется с ростом слова.</p><ol><li><a href="/ru/learn/gf10-ten-bits/">10 GF10: десять битов</a></li><li><a href="/ru/learn/gf12-twelve-bits/">11 GF12: двенадцать битов</a></li><li><a href="/ru/learn/gf14-fourteen-bits/">12 GF14: четырнадцать битов</a></li></ol></li><li><h3>Модуль 5 · GF16 в работе</h3><p>Основной 16-битный формат, скалярное произведение из двух слагаемых в GF-T16, затем GF20 и GF24.</p><ol><li><a href="/ru/learn/gf16-the-primary/">13 GF16: основной формат</a></li><li><a href="/ru/learn/two-products-one-sum/">14 Два произведения, одна сумма</a></li><li><a href="/ru/learn/gf24-twenty-four-bits/">15 GF20 и GF24</a></li></ol></li><li><h3>Модуль 6 · От GF32 до GF64</h3><p>GF32 рядом с IEEE single, GF48 без пары в IEEE, GF64 рядом с IEEE double.</p><ol><li><a href="/ru/learn/gf32-a-single/">16 GF32: одинарная точность</a></li><li><a href="/ru/learn/gf48-forty-eight-bits/">17 GF48: сорок восемь битов</a></li><li><a href="/ru/learn/gf64-a-double/">18 GF64: двойная точность</a></li></ol></li><li><h3>Модуль 7 · От GF96 до GF256</h3><p>GF96, GF128 и GF256, где спеки держат раскладку инвариантами.</p><ol><li><a href="/ru/learn/gf96-ninety-six-bits/">19 GF96: девяносто шесть битов</a></li><li><a href="/ru/learn/gf128-a-quad/">20 GF128: четверная точность</a></li><li><a href="/ru/learn/gf256-two-five-six/">21 GF256: двести пятьдесят шесть битов</a></li></ol></li><li><h3>Модуль 8 · Самые широкие ступени, затем триты</h3><p>GF512 и GF1024, две самые широкие ступени, затем GF-T8, где порядок уходит в триты.</p><ol><li><a href="/ru/learn/gf512-five-twelve/">22 GF512: пятьсот двенадцать битов</a></li><li><a href="/ru/learn/gf1024-one-kilobit/">23 GF1024: один килобит</a></li><li><a href="/ru/learn/gft8-exponent-in-trits/">24 GF-T8: порядок в тритах</a></li></ol></li><li><h3>Модуль 9 · Ещё триты, затем декодирование</h3><p>GF-T16 и GF-T32, затем почему фиксированные поля декодируются параллельно, а posit — нет.</p><ol><li><a href="/ru/learn/gft16-six-trits/">25 GF-T16: шесть тритов</a></li><li><a href="/ru/learn/gft32-twelve-trits/">26 GF-T32: двенадцать тритов</a></li><li><a href="/ru/learn/decode-in-one-step/">27 Раскодировать за один шаг</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/phi-in-sixteen-bits/index.html b/apps/website/public/ru/learn/phi-in-sixteen-bits/index.html
index a36199ca89..28dd12555d 100644
--- a/apps/website/public/ru/learn/phi-in-sixteen-bits/index.html
+++ b/apps/website/public/ru/learn/phi-in-sixteen-bits/index.html
@@ -177,5 +177,6 @@ <h1>Phi в шестнадцати битах</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Правило и его числа</h3><p>Одно правило делит каждую ширину, отношение, к которому оно стремится, и числа Люка за тройкой 3.</p><ol><li><a href="/ru/learn/a-float-cut-by-phi/">01 Число с плавающей точкой, разрезанное по phi</a></li><li><a href="/ru/learn/phi-as-a-ratio/">02 Phi как отношение</a></li><li><a href="/ru/learn/lucas-numbers-stay-whole/">03 Числа Люка остаются целыми</a></li></ol></li><li><h3>Модуль 2 · Почему phi, почему три</h3><p>Почему деление идёт по phi, почему основание три и как спека проверяет, что GF16 хранит phi.</p><ol><li><a href="/ru/learn/why-the-split-is-phi/">04 Почему деление по phi</a></li><li><a href="/ru/learn/why-base-three/">05 Почему основание три</a></li><li class="on" aria-current="page"><a href="/ru/learn/phi-in-sixteen-bits/">06 Phi в шестнадцати битах</a></li></ol></li><li><h3>Модуль 3 · Малые ступени: от GF4 до GF8</h3><p>GF4, GF6 и GF8 — меньше всего битов, и округление до целых битов стоит здесь дороже всего.</p><ol><li><a href="/ru/learn/gf4-four-bits/">07 GF4: четыре бита</a></li><li><a href="/ru/learn/gf6-six-bits/">08 GF6: шесть битов</a></li><li><a href="/ru/learn/gf8-one-byte/">09 GF8: один байт</a></li></ol></li><li><h3>Модуль 4 · От десяти до четырнадцати битов</h3><p>GF10, GF12 и GF14 и то, как расстояние до 1 / phi меняется с ростом слова.</p><ol><li><a href="/ru/learn/gf10-ten-bits/">10 GF10: десять битов</a></li><li><a href="/ru/learn/gf12-twelve-bits/">11 GF12: двенадцать битов</a></li><li><a href="/ru/learn/gf14-fourteen-bits/">12 GF14: четырнадцать битов</a></li></ol></li><li><h3>Модуль 5 · GF16 в работе</h3><p>Основной 16-битный формат, скалярное произведение из двух слагаемых в GF-T16, затем GF20 и GF24.</p><ol><li><a href="/ru/learn/gf16-the-primary/">13 GF16: основной формат</a></li><li><a href="/ru/learn/two-products-one-sum/">14 Два произведения, одна сумма</a></li><li><a href="/ru/learn/gf24-twenty-four-bits/">15 GF20 и GF24</a></li></ol></li><li><h3>Модуль 6 · От GF32 до GF64</h3><p>GF32 рядом с IEEE single, GF48 без пары в IEEE, GF64 рядом с IEEE double.</p><ol><li><a href="/ru/learn/gf32-a-single/">16 GF32: одинарная точность</a></li><li><a href="/ru/learn/gf48-forty-eight-bits/">17 GF48: сорок восемь битов</a></li><li><a href="/ru/learn/gf64-a-double/">18 GF64: двойная точность</a></li></ol></li><li><h3>Модуль 7 · От GF96 до GF256</h3><p>GF96, GF128 и GF256, где спеки держат раскладку инвариантами.</p><ol><li><a href="/ru/learn/gf96-ninety-six-bits/">19 GF96: девяносто шесть битов</a></li><li><a href="/ru/learn/gf128-a-quad/">20 GF128: четверная точность</a></li><li><a href="/ru/learn/gf256-two-five-six/">21 GF256: двести пятьдесят шесть битов</a></li></ol></li><li><h3>Модуль 8 · Самые широкие ступени, затем триты</h3><p>GF512 и GF1024, две самые широкие ступени, затем GF-T8, где порядок уходит в триты.</p><ol><li><a href="/ru/learn/gf512-five-twelve/">22 GF512: пятьсот двенадцать битов</a></li><li><a href="/ru/learn/gf1024-one-kilobit/">23 GF1024: один килобит</a></li><li><a href="/ru/learn/gft8-exponent-in-trits/">24 GF-T8: порядок в тритах</a></li></ol></li><li><h3>Модуль 9 · Ещё триты, затем декодирование</h3><p>GF-T16 и GF-T32, затем почему фиксированные поля декодируются параллельно, а posit — нет.</p><ol><li><a href="/ru/learn/gft16-six-trits/">25 GF-T16: шесть тритов</a></li><li><a href="/ru/learn/gft32-twelve-trits/">26 GF-T32: двенадцать тритов</a></li><li><a href="/ru/learn/decode-in-one-step/">27 Раскодировать за один шаг</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/phi-scaled-ternary/index.html b/apps/website/public/ru/learn/phi-scaled-ternary/index.html
index 101b9580a1..4c5675fcef 100644
--- a/apps/website/public/ru/learn/phi-scaled-ternary/index.html
+++ b/apps/website/public/ru/learn/phi-scaled-ternary/index.html
@@ -125,5 +125,6 @@ <h1>Веса минус фи, ноль и плюс фи</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Лаборатория: наши исследования</h3><p>Свой формат чисел, честная таблица результатов и таблицы модели, перемноженные на плате.</p><ol><li><a href="/ru/learn/t27-number-format/">01 T27: наш собственный формат чисел</a></li><li><a href="/ru/learn/an-honest-scoreboard/">02 Честная таблица результатов</a></li><li><a href="/ru/learn/a-model-on-the-board/">03 Таблицы модели на плате</a></li></ol></li><li><h3>Модуль 2 · ИИ-числа: блок MX</h3><p>Как ИИ-чипы хранят веса в нескольких битах: один общий масштаб на блок, сам байт масштаба и что один выброс делает с соседями.</p><ol><li><a href="/ru/learn/one-scale-per-block/">04 Один масштаб на блок весов</a></li><li><a href="/ru/learn/scale-byte-native/">05 Байт масштаба на настоящей машине</a></li><li><a href="/ru/learn/one-outlier/">06 Один выброс, много нулей</a></li></ol></li><li><h3>Модуль 3 · Тернарные веса</h3><p>Веса, которые бывают только минус масштаб, ноль или плюс масштаб, пять правил, которые должен пройти тернарный алфавит, и прогон тестов, который ничего не проверил.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/phi-scaled-ternary/">07 Веса минус фи, ноль и плюс фи</a></li><li><a href="/ru/learn/a-pass-that-checked-nothing/">08 Прогон, который ничего не проверил</a></li><li><a href="/ru/learn/five-rules-for-weights/">09 Пять правил для алфавита весов</a></li></ol></li><li><h3>Модуль 4 · Тернарный формат Ternary Network Float</h3><p>Правило, которое компилятор проверяет до запуска любого теста, и 17-битное число с плавающей точкой, порядок которого — четыре сбалансированных трита.</p><ol><li><a href="/ru/learn/blocked-is-not-failed/">10 Заблокирован — не значит провален</a></li><li><a href="/ru/learn/four-trit-exponent/">11 Число, у которого порядок — четыре трита</a></li><li><a href="/ru/learn/negate-twice/">12 Дважды сменить знак — получить число обратно</a></li></ol></li><li><h3>Модуль 5 · Арифметика чисел со знаком</h3><p>Умножить два числа со знаком, сложить их, когда знаки разные, и сделать то и другое сразу в умножении с накоплением.</p><ol><li><a href="/ru/learn/a-sign-is-one-bit/">13 Знак — это один бит</a></li><li><a href="/ru/learn/opposite-signs-subtract/">14 Разные знаки — вычитание</a></li><li><a href="/ru/learn/multiply-then-add/">15 Умножить, потом сложить</a></li></ol></li><li><h3>Модуль 6 · Части нейрона</h3><p>ReLU с изломом в нуле, степень двойки для softmax и argmax, который называет ответ.</p><ol><li><a href="/ru/learn/a-bend-at-zero/">16 Излом в нуле</a></li><li><a href="/ru/learn/two-to-the-x/">17 Два в степени x</a></li><li><a href="/ru/learn/pick-the-largest/">18 Выбрать наибольшее</a></li></ol></li><li><h3>Модуль 7 · Обучение на ошибке</h3><p>Потеря, которая оценивает неверную догадку в битах, один шаг, который сдвигает вес против градиента, и скрытый слой, который нужен XOR.</p><ol><li><a href="/ru/learn/a-loss-in-bits/">19 Потеря в битах</a></li><li><a href="/ru/learn/one-step-downhill/">20 Один шаг вниз по склону</a></li><li><a href="/ru/learn/xor-needs-a-bend/">21 XOR нужен излом</a></li></ol></li><li><h3>Модуль 8 · BitNet: тернарные сети</h3><p>Порог, который сжимает сумму обратно до трёх значений, один нейрон, который становится другой функцией при смене весов, и нейрон, который читает входы по 27 тритов за раз.</p><ol><li><a href="/ru/learn/back-to-three-values/">22 Обратно в три значения</a></li><li><a href="/ru/learn/same-neuron-new-weights/">23 Тот же нейрон, другие веса</a></li><li><a href="/ru/learn/a-neuron-in-chunks/">24 Нейрон по кускам</a></li></ol></li><li><h3>Модуль 9 · Тернарный MAC как чип</h3><p>Скалярное произведение 27 тритов как провода без регистра, та же сумма, которую регистр копит на каждом такте, и небольшая целая сеть в конце курса.</p><ol><li><a href="/ru/learn/a-dot-product-in-wires/">25 Скалярное произведение из проводов</a></li><li><a href="/ru/learn/add-it-up-every-clock/">26 Складывать на каждом такте</a></li><li><a href="/ru/learn/a-whole-network/">27 Целая сеть</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/pick-the-largest/index.html b/apps/website/public/ru/learn/pick-the-largest/index.html
index 1b410dfc34..ce78574876 100644
--- a/apps/website/public/ru/learn/pick-the-largest/index.html
+++ b/apps/website/public/ru/learn/pick-the-largest/index.html
@@ -121,5 +121,6 @@ <h1>Выбрать наибольшее</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Лаборатория: наши исследования</h3><p>Свой формат чисел, честная таблица результатов и таблицы модели, перемноженные на плате.</p><ol><li><a href="/ru/learn/t27-number-format/">01 T27: наш собственный формат чисел</a></li><li><a href="/ru/learn/an-honest-scoreboard/">02 Честная таблица результатов</a></li><li><a href="/ru/learn/a-model-on-the-board/">03 Таблицы модели на плате</a></li></ol></li><li><h3>Модуль 2 · ИИ-числа: блок MX</h3><p>Как ИИ-чипы хранят веса в нескольких битах: один общий масштаб на блок, сам байт масштаба и что один выброс делает с соседями.</p><ol><li><a href="/ru/learn/one-scale-per-block/">04 Один масштаб на блок весов</a></li><li><a href="/ru/learn/scale-byte-native/">05 Байт масштаба на настоящей машине</a></li><li><a href="/ru/learn/one-outlier/">06 Один выброс, много нулей</a></li></ol></li><li><h3>Модуль 3 · Тернарные веса</h3><p>Веса, которые бывают только минус масштаб, ноль или плюс масштаб, пять правил, которые должен пройти тернарный алфавит, и прогон тестов, который ничего не проверил.</p><ol><li><a href="/ru/learn/phi-scaled-ternary/">07 Веса минус фи, ноль и плюс фи</a></li><li><a href="/ru/learn/a-pass-that-checked-nothing/">08 Прогон, который ничего не проверил</a></li><li><a href="/ru/learn/five-rules-for-weights/">09 Пять правил для алфавита весов</a></li></ol></li><li><h3>Модуль 4 · Тернарный формат Ternary Network Float</h3><p>Правило, которое компилятор проверяет до запуска любого теста, и 17-битное число с плавающей точкой, порядок которого — четыре сбалансированных трита.</p><ol><li><a href="/ru/learn/blocked-is-not-failed/">10 Заблокирован — не значит провален</a></li><li><a href="/ru/learn/four-trit-exponent/">11 Число, у которого порядок — четыре трита</a></li><li><a href="/ru/learn/negate-twice/">12 Дважды сменить знак — получить число обратно</a></li></ol></li><li><h3>Модуль 5 · Арифметика чисел со знаком</h3><p>Умножить два числа со знаком, сложить их, когда знаки разные, и сделать то и другое сразу в умножении с накоплением.</p><ol><li><a href="/ru/learn/a-sign-is-one-bit/">13 Знак — это один бит</a></li><li><a href="/ru/learn/opposite-signs-subtract/">14 Разные знаки — вычитание</a></li><li><a href="/ru/learn/multiply-then-add/">15 Умножить, потом сложить</a></li></ol></li><li><h3>Модуль 6 · Части нейрона</h3><p>ReLU с изломом в нуле, степень двойки для softmax и argmax, который называет ответ.</p><ol><li><a href="/ru/learn/a-bend-at-zero/">16 Излом в нуле</a></li><li><a href="/ru/learn/two-to-the-x/">17 Два в степени x</a></li><li class="on" aria-current="page"><a href="/ru/learn/pick-the-largest/">18 Выбрать наибольшее</a></li></ol></li><li><h3>Модуль 7 · Обучение на ошибке</h3><p>Потеря, которая оценивает неверную догадку в битах, один шаг, который сдвигает вес против градиента, и скрытый слой, который нужен XOR.</p><ol><li><a href="/ru/learn/a-loss-in-bits/">19 Потеря в битах</a></li><li><a href="/ru/learn/one-step-downhill/">20 Один шаг вниз по склону</a></li><li><a href="/ru/learn/xor-needs-a-bend/">21 XOR нужен излом</a></li></ol></li><li><h3>Модуль 8 · BitNet: тернарные сети</h3><p>Порог, который сжимает сумму обратно до трёх значений, один нейрон, который становится другой функцией при смене весов, и нейрон, который читает входы по 27 тритов за раз.</p><ol><li><a href="/ru/learn/back-to-three-values/">22 Обратно в три значения</a></li><li><a href="/ru/learn/same-neuron-new-weights/">23 Тот же нейрон, другие веса</a></li><li><a href="/ru/learn/a-neuron-in-chunks/">24 Нейрон по кускам</a></li></ol></li><li><h3>Модуль 9 · Тернарный MAC как чип</h3><p>Скалярное произведение 27 тритов как провода без регистра, та же сумма, которую регистр копит на каждом такте, и небольшая целая сеть в конце курса.</p><ol><li><a href="/ru/learn/a-dot-product-in-wires/">25 Скалярное произведение из проводов</a></li><li><a href="/ru/learn/add-it-up-every-clock/">26 Складывать на каждом такте</a></li><li><a href="/ru/learn/a-whole-network/">27 Целая сеть</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/pins-and-package/index.html b/apps/website/public/ru/learn/pins-and-package/index.html
index 9a65cd3fd6..2b2f4a8bc8 100644
--- a/apps/website/public/ru/learn/pins-and-package/index.html
+++ b/apps/website/public/ru/learn/pins-and-package/index.html
@@ -462,5 +462,6 @@ <h1>Выводы, шарики и корпус</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Чип</h3><p>Что такое FPGA, какой чип у нас и как его выводы встречаются с платой.</p><ol><li><a href="/ru/learn/fpga-is-a-grid/">01 FPGA — это сетка, которую вы настраиваете</a></li><li><a href="/ru/learn/which-chip/">02 Какой это чип?</a></li><li class="on" aria-current="page"><a href="/ru/learn/pins-and-package/">03 Выводы, шарики и корпус</a></li></ol></li><li><h3>Модуль 2 · Числа в железе</h3><p>Биты, триты и форматы чисел, и сколько стоит арифметика в логике.</p><ol><li><a href="/ru/learn/bits-and-trits/">04 Биты, триты и GF16</a></li><li><a href="/ru/learn/what-arithmetic-costs/">05 Сколько стоит арифметика</a></li><li><a href="/ru/learn/how-much-fits/">06 Сколько помещается на чип</a></li></ol></li><li><h3>Модуль 3 · Ваша программа на t27</h3><p>Напишите спеку, проверьте её и увидьте, почему «компилируется» не значит «верно».</p><ol><li><a href="/ru/learn/hello-t27/">07 Привет, t27</a></li><li><a href="/ru/learn/tests-are-the-spec/">08 Тесты — часть спеки</a></li><li><a href="/ru/learn/compiles-is-not-correct/">09 Компилируется — не значит верно</a></li></ol></li><li><h3>Модуль 4 · Внутри t27c</h3><p>Как компилятор читает спеку и что он пишет, включая нативный t27b.</p><ol><li><a href="/ru/learn/compiler-stages/">10 Стадии t27c</a></li><li><a href="/ru/learn/t27c-on-a-real-machine/">11 t27c на настоящей машине</a></li><li><a href="/ru/learn/seven-backends-and-t27b/">12 Семь бэкендов и t27b</a></li></ol></li><li><h3>Модуль 5 · От спеки к железу</h3><p>Verilog, который пишет t27c, ячейки, в которые он превращается, и что делает одна LUT.</p><ol><li><a href="/ru/learn/spec-to-verilog/">13 От спеки к Verilog</a></li><li><a href="/ru/learn/count-the-cells/">14 Посчитайте ячейки</a></li><li><a href="/ru/learn/one-lut-up-close/">15 Одна LUT вблизи</a></li></ol></li><li><h3>Модуль 6 · Читаем синтез</h3><p>Что yosys сообщает о вашем дизайне и какие предупреждения важны.</p><ol><li><a href="/ru/learn/synthesis-warnings/">16 Важные предупреждения синтеза</a></li><li><a href="/ru/learn/state-machines/">17 Конечные автоматы</a></li><li><a href="/ru/learn/where-the-luts-go/">18 Куда уходят LUT</a></li></ol></li><li><h3>Модуль 7 · Размещение, трассировка, тайминг</h3><p>Где ячейки оказываются на кристалле и успевает ли тактовая частота.</p><ol><li><a href="/ru/learn/placement/">19 Размещение: где оказываются ячейки</a></li><li><a href="/ru/learn/logic-depth/">20 Глубина логики</a></li><li><a href="/ru/learn/timing-slack/">21 Тайминг и запас</a></li></ol></li><li><h3>Модуль 8 · Битстрим</h3><p>Как оттрассированный дизайн становится битами, которые загружает чип, без Vivado.</p><ol><li><a href="/ru/learn/fasm-to-frames/">22 От FASM к фреймам</a></li><li><a href="/ru/learn/a-build-receipt/">23 Сборка по пунктам</a></li><li><a href="/ru/learn/read-a-bitstream/">24 Читаем битстрим</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Спросите чип, кто он, загрузите биты и проверьте их.</p><ol><li><a href="/ru/learn/jtag-asks-the-chip/">25 JTAG спрашивает чип</a></li><li><a href="/ru/learn/same-bits-on-the-board/">26 Те же биты — на плате</a></li><li><a href="/ru/learn/capstone/">27 Итог: переугадайте yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/placement/index.html b/apps/website/public/ru/learn/placement/index.html
index db7be7e45d..0f30cfc95a 100644
--- a/apps/website/public/ru/learn/placement/index.html
+++ b/apps/website/public/ru/learn/placement/index.html
@@ -312,5 +312,6 @@ <h1>Размещение: где оказываются ячейки</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Чип</h3><p>Что такое FPGA, какой чип у нас и как его выводы встречаются с платой.</p><ol><li><a href="/ru/learn/fpga-is-a-grid/">01 FPGA — это сетка, которую вы настраиваете</a></li><li><a href="/ru/learn/which-chip/">02 Какой это чип?</a></li><li><a href="/ru/learn/pins-and-package/">03 Выводы, шарики и корпус</a></li></ol></li><li><h3>Модуль 2 · Числа в железе</h3><p>Биты, триты и форматы чисел, и сколько стоит арифметика в логике.</p><ol><li><a href="/ru/learn/bits-and-trits/">04 Биты, триты и GF16</a></li><li><a href="/ru/learn/what-arithmetic-costs/">05 Сколько стоит арифметика</a></li><li><a href="/ru/learn/how-much-fits/">06 Сколько помещается на чип</a></li></ol></li><li><h3>Модуль 3 · Ваша программа на t27</h3><p>Напишите спеку, проверьте её и увидьте, почему «компилируется» не значит «верно».</p><ol><li><a href="/ru/learn/hello-t27/">07 Привет, t27</a></li><li><a href="/ru/learn/tests-are-the-spec/">08 Тесты — часть спеки</a></li><li><a href="/ru/learn/compiles-is-not-correct/">09 Компилируется — не значит верно</a></li></ol></li><li><h3>Модуль 4 · Внутри t27c</h3><p>Как компилятор читает спеку и что он пишет, включая нативный t27b.</p><ol><li><a href="/ru/learn/compiler-stages/">10 Стадии t27c</a></li><li><a href="/ru/learn/t27c-on-a-real-machine/">11 t27c на настоящей машине</a></li><li><a href="/ru/learn/seven-backends-and-t27b/">12 Семь бэкендов и t27b</a></li></ol></li><li><h3>Модуль 5 · От спеки к железу</h3><p>Verilog, который пишет t27c, ячейки, в которые он превращается, и что делает одна LUT.</p><ol><li><a href="/ru/learn/spec-to-verilog/">13 От спеки к Verilog</a></li><li><a href="/ru/learn/count-the-cells/">14 Посчитайте ячейки</a></li><li><a href="/ru/learn/one-lut-up-close/">15 Одна LUT вблизи</a></li></ol></li><li><h3>Модуль 6 · Читаем синтез</h3><p>Что yosys сообщает о вашем дизайне и какие предупреждения важны.</p><ol><li><a href="/ru/learn/synthesis-warnings/">16 Важные предупреждения синтеза</a></li><li><a href="/ru/learn/state-machines/">17 Конечные автоматы</a></li><li><a href="/ru/learn/where-the-luts-go/">18 Куда уходят LUT</a></li></ol></li><li><h3>Модуль 7 · Размещение, трассировка, тайминг</h3><p>Где ячейки оказываются на кристалле и успевает ли тактовая частота.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/placement/">19 Размещение: где оказываются ячейки</a></li><li><a href="/ru/learn/logic-depth/">20 Глубина логики</a></li><li><a href="/ru/learn/timing-slack/">21 Тайминг и запас</a></li></ol></li><li><h3>Модуль 8 · Битстрим</h3><p>Как оттрассированный дизайн становится битами, которые загружает чип, без Vivado.</p><ol><li><a href="/ru/learn/fasm-to-frames/">22 От FASM к фреймам</a></li><li><a href="/ru/learn/a-build-receipt/">23 Сборка по пунктам</a></li><li><a href="/ru/learn/read-a-bitstream/">24 Читаем битстрим</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Спросите чип, кто он, загрузите биты и проверьте их.</p><ol><li><a href="/ru/learn/jtag-asks-the-chip/">25 JTAG спрашивает чип</a></li><li><a href="/ru/learn/same-bits-on-the-board/">26 Те же биты — на плате</a></li><li><a href="/ru/learn/capstone/">27 Итог: переугадайте yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/plan-before-code/index.html b/apps/website/public/ru/learn/plan-before-code/index.html
index 7e2bf278fb..d2b8f927a8 100644
--- a/apps/website/public/ru/learn/plan-before-code/index.html
+++ b/apps/website/public/ru/learn/plan-before-code/index.html
@@ -193,5 +193,6 @@ <h1>Сначала план</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Зачем проверять</h3><p>Дизайны, которые компилируются и ошибаются; модель, которая выносит вердикт; план, записанный до кода.</p><ol><li><a href="/ru/learn/bugs-that-compile/">01 Баги, которые компилируются</a></li><li><a href="/ru/learn/golden-model/">02 Золотая модель</a></li><li class="on" aria-current="page"><a href="/ru/learn/plan-before-code/">03 Сначала план</a></li></ol></li><li><h3>Модуль 2 · Тестбенчи</h3><p>Стимулы, проверки и вердикт, записанные как один spec рядом с дизайном, который они судят.</p><ol><li><a href="/ru/learn/anatomy-of-a-tb/">04 Анатомия тестбенча</a></li><li><a href="/ru/learn/self-checking/">05 Самопроверка вместо всматривания</a></li><li><a href="/ru/learn/directed-vs-random/">06 Направленные против случайных</a></li></ol></li><li><h3>Модуль 3 · Временные диаграммы</h3><p>Трасса каждого сигнала, прочитанная так, как её читает инженер по железу, и два прогона, сравнённые между собой.</p><ol><li><a href="/ru/learn/vcd-format/">07 Формат VCD</a></li><li><a href="/ru/learn/debug-with-waves/">08 Отладка по волнам</a></li><li><a href="/ru/learn/compare-traces/">09 Сравнение двух трасс</a></li></ol></li><li><h3>Модуль 4 · Векторы соответствия</h3><p>Случаи с ответом, записанным рядом, там, откуда их достанет компилятор.</p><ol><li><a href="/ru/learn/what-a-vector-is/">10 Что такое вектор</a></li><li><a href="/ru/learn/golden-ruler-vectors/">11 Векторы как эталонная линейка</a></li><li><a href="/ru/learn/end-to-end-scenarios/">12 Сквозные сценарии</a></li></ol></li><li><h3>Модуль 5 · Косимуляция</h3><p>Spec, симулятор и плата сходятся в одном ответе на стенде Artix-7 XC7A200T, и что делать, когда не сходятся.</p><ol><li><a href="/ru/learn/three-way-agreement/">13 Тройное согласие</a></li><li><a href="/ru/learn/simulator-internals/">14 Внутри симулятора</a></li><li><a href="/ru/learn/when-they-disagree/">15 Когда они не сходятся</a></li></ol></li><li><h3>Модуль 6 · Покрытие</h3><p>Чего коснулись тесты: строки, переключения, состояния — и что прячет это число.</p><ol><li><a href="/ru/learn/line-and-toggle/">16 Строки и переключения</a></li><li><a href="/ru/learn/fsm-coverage/">17 Покрытие автоматов</a></li><li><a href="/ru/learn/coverage-lies/">18 Покрытие лжёт</a></li></ol></li><li><h3>Модуль 7 · Формальные методы</h3><p>Ассерты, верные каждый такт; ограниченный поиск контрпримера; и почему доказательству нужна индукция.</p><ol><li><a href="/ru/learn/assertions/">19 Ассерты</a></li><li><a href="/ru/learn/bounded-model-checking/">20 Ограниченная проверка моделей</a></li><li><a href="/ru/learn/induction/">21 Индукция</a></li></ol></li><li><h3>Модуль 8 · Мутации</h3><p>Ломайте дизайн нарочно и считайте, что заметили тесты.</p><ol><li><a href="/ru/learn/kill-the-mutant/">22 Убить мутанта</a></li><li><a href="/ru/learn/surviving-mutants/">23 Выжившие мутанты</a></li><li><a href="/ru/learn/test-strength/">24 Сила тестов</a></li></ol></li><li><h3>Модуль 9 · Приёмка</h3><p>Одна команда, все квитанции, чистый вердикт, который можно показать.</p><ol><li><a href="/ru/learn/build-verify/">25 Сборка и проверка</a></li><li><a href="/ru/learn/board-selftest/">26 Стенд проверяет себя</a></li><li><a href="/ru/learn/verify-capstone/">27 Капстоун: проверить модуль целиком</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">Этот курс сам является spec t27: прочитайте его</a></p><p>Виджеты сохраняют свои английские слова: каждый живёт в собственном spec t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/pointers-and-flags/index.html b/apps/website/public/ru/learn/pointers-and-flags/index.html
index bef97828de..4aee38fa66 100644
--- a/apps/website/public/ru/learn/pointers-and-flags/index.html
+++ b/apps/website/public/ru/learn/pointers-and-flags/index.html
@@ -434,5 +434,6 @@ <h1>Указатели и флаги</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое тактовый сигнал</h3><p>Один фронт, один мир: всё, что делит фронт тактового сигнала, делит и мир; период и джиттер настоящего фронта и место, где тактовый сигнал входит в плату.</p><ol><li><a href="/ru/learn/one-edge-one-world/">01 Один фронт, один мир</a></li><li><a href="/ru/learn/period-and-jitter/">02 Период, фронт и джиттер</a></li><li><a href="/ru/learn/where-the-clock-enters/">03 Где входит тактовый сигнал</a></li></ol></li><li><h3>Модуль 2 · Деревья тактового сигнала</h3><p>Перекос и задержка прохождения, глобальная сеть буферов и ловушка стробирования клока логикой.</p><ol><li><a href="/ru/learn/skew-and-insertion/">04 Перекос и задержка прохождения</a></li><li><a href="/ru/learn/global-buffers/">05 Глобальный буфер</a></li><li><a href="/ru/learn/gated-clock-trap/">06 Ловушка стробированного клока</a></li></ol></li><li><h3>Модуль 3 · PLL и MMCM</h3><p>Умножить и поделить один тактовый сигнал в другой, сдвинуть его фазу шагами VCO и какие тактовые сигналы анализатор считает родственными.</p><ol><li><a href="/ru/learn/multiply-and-divide/">07 Умножить и поделить</a></li><li><a href="/ru/learn/phase-and-lock/">08 Фаза и захват</a></li><li><a href="/ru/learn/related-or-not/">09 Родственные или чужие</a></li></ol></li><li><h3>Модуль 4 · Сбросы</h3><p>Ассертировать асинхронно, снимать синхронно: три вида сброса, конвейер снятия и дерево, которое растит сброс.</p><ol><li><a href="/ru/learn/sync-or-async/">10 Синхронный или асинхронный</a></li><li><a href="/ru/learn/reset-release/">11 Конвейер снятия</a></li><li><a href="/ru/learn/reset-fanout/">12 Разветвление сброса</a></li></ol></li><li><h3>Модуль 5 · Метастабильность</h3><p>Окно setup-hold, среднее время между отказами в целочисленной арифметике и два триггера, которые это чинят.</p><ol><li><a href="/ru/learn/setup-hold-window/">13 Окно setup-hold</a></li><li><a href="/ru/learn/mtbf/">14 Среднее время между отказами</a></li><li><a href="/ru/learn/two-flop-sync/">15 Двухтриггерный синхронизатор</a></li></ol></li><li><h3>Модуль 6 · Пересечение многих бит</h3><p>Почему двоичная шина рвётся, почему код Грея нет, и рукопожатие, переносящее импульс между мирами.</p><ol><li><a href="/ru/learn/why-binary-fails/">16 Почему двоичный код рвётся</a></li><li><a href="/ru/learn/gray-code/">17 Код Грея</a></li><li><a href="/ru/learn/handshakes-and-pulses/">18 Рукопожатия и импульсы</a></li></ol></li><li><h3>Модуль 7 · Асинхронный FIFO</h3><p>Указатели, флаги и глубина: буфер, переносящий поток между двумя тактовыми сигналами.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/pointers-and-flags/">19 Указатели и флаги</a></li><li><a href="/ru/learn/depth-sizing/">20 Выбор глубины</a></li><li><a href="/ru/learn/fifo-testbench/">21 Тестбенч FIFO</a></li></ol></li><li><h3>Модуль 8 · Ограничения</h3><p>Строки, сообщающие анализатору, что такое тактовый сигнал, какие пути не проверять и чему должны соответствовать выводы.</p><ol><li><a href="/ru/learn/create-clock/">22 Строка create_clock</a></li><li><a href="/ru/learn/false-path-and-max-delay/">23 False path и max delay</a></li><li><a href="/ru/learn/io-timing/">24 Тайминг выводов</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Отчёт CDC, одно пересечение, захваченное на триггерах, и дифф битстрима, замыкающий курс.</p><ol><li><a href="/ru/learn/cdc-report/">25 Отчёт CDC</a></li><li><a href="/ru/learn/capture-a-crossing/">26 Захватить пересечение</a></li><li><a href="/ru/learn/capstone-diff/">27 Капстоун: поменяй и сравни</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/polyphase/index.html b/apps/website/public/ru/learn/polyphase/index.html
index f321967223..d7e9ba5b32 100644
--- a/apps/website/public/ru/learn/polyphase/index.html
+++ b/apps/website/public/ru/learn/polyphase/index.html
@@ -1122,5 +1122,6 @@ <h1>Полифазная форма</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Сложение</h3><p>Почему сложение медленное: перенос, который идёт вверх по слову, выделенная цепь в каждом слайсе, префиксные сети и деревья сохранения переноса.</p><ol><li><a href="/ru/learn/the-carry-chain/">01 Цепь переноса</a></li><li><a href="/ru/learn/prefix-adders/">02 Взгляд вперёд: префиксные сумматоры</a></li><li><a href="/ru/learn/carry-save/">03 Три числа сразу</a></li></ol></li><li><h3>Модуль 2 · Умножение</h3><p>Произведение — сумма сдвинутых копий; перекодирование Бута вдвое сокращает их, а слайс DSP48E1 делает остальное одним блоком.</p><ol><li><a href="/ru/learn/shift-and-add/">04 Сдвиг и сложение</a></li><li><a href="/ru/learn/booths-trick/">05 Приём Бута</a></li><li><a href="/ru/learn/the-dsp48e1-slice/">06 Слайс DSP48E1</a></li></ol></li><li><h3>Модуль 3 · Фиксированная точка</h3><p>Где стоит двоичная точка, что округление делает со значением и с его средним и что даёт каждый бит квантователя.</p><ol><li><a href="/ru/learn/q-formats/">07 Форматы Q</a></li><li><a href="/ru/learn/rounding-and-saturation/">08 Округление и насыщение</a></li><li><a href="/ru/learn/quantization-noise/">09 Шум квантования</a></li></ol></li><li><h3>Модуль 4 · Функции в железе</h3><p>Синус, косинус, угол и длина из сдвигов и сложений — и когда таблица оказывается выгоднее.</p><ol><li><a href="/ru/learn/rotating-by-shifts/">10 Поворот сдвигами</a></li><li><a href="/ru/learn/angle-and-magnitude/">11 Угол и модуль</a></li><li><a href="/ru/learn/tables-and-interpolation/">12 Таблицы и интерполяция</a></li></ol></li><li><h3>Модуль 5 · Сигналы и дискретизация</h3><p>Что дискретизация делает с частотой, как генератор строится из сумматора и что измеряет бин ДПФ.</p><ol><li><a href="/ru/learn/sampling-and-aliasing/">13 Дискретизация и наложение спектров</a></li><li><a href="/ru/learn/the-nco/">14 Генератор с числовым управлением</a></li><li><a href="/ru/learn/bins-of-a-spectrum/">15 Бины спектра</a></li></ol></li><li><h3>Модуль 6 · Фильтры FIR</h3><p>Скользящее среднее, расчёт окном sinc с целыми коэффициентами и раскладка отводов по слайсам DSP.</p><ol><li><a href="/ru/learn/the-moving-average/">16 Скользящее среднее</a></li><li><a href="/ru/learn/designing-taps/">17 Расчёт коэффициентов</a></li><li><a href="/ru/learn/folding-taps-onto-slices/">18 Раскладка отводов по слайсам</a></li></ol></li><li><h3>Модуль 7 · Многоскоростная обработка</h3><p>Понижение частоты дискретизации без заворота шума: децимация, фильтр CIC и полифазная форма.</p><ol><li><a href="/ru/learn/lowering-the-rate/">19 Понижение частоты</a></li><li><a href="/ru/learn/the-cic-filter/">20 Фильтр CIC</a></li><li class="on" aria-current="page"><a href="/ru/learn/polyphase/">21 Полифазная форма</a></li></ol></li><li><h3>Модуль 8 · БПФ</h3><p>N log N вместо N^2: бабочки, бит-реверсный порядок входа и биты, которые добавляет каждый этап.</p><ol><li><a href="/ru/learn/the-butterfly/">22 Бабочка</a></li><li><a href="/ru/learn/twiddles-and-bit-reversal/">23 Поворачивающие множители и бит-реверс</a></li><li><a href="/ru/learn/growth-through-the-stages/">24 Рост через этапы</a></li></ol></li><li><h3>Модуль 9 · На стенде</h3><p>От арифметики к плате: коррелятор из спеки модема, бюджет слайсов и тайминга и итоговый фильтр.</p><ol><li><a href="/ru/learn/a-correlator-finds-the-frame/">25 Коррелятор находит кадр</a></li><li><a href="/ru/learn/does-it-fit-and-close-timing/">26 Помещается ли и сходится ли тайминг?</a></li><li><a href="/ru/learn/capstone-a-filter-on-the-board/">27 Итог: фильтр на плате</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/ports-and-kinds/index.html b/apps/website/public/ru/learn/ports-and-kinds/index.html
index 1361bf8829..63e3508438 100644
--- a/apps/website/public/ru/learn/ports-and-kinds/index.html
+++ b/apps/website/public/ru/learn/ports-and-kinds/index.html
@@ -424,5 +424,6 @@ <h1>Порты и типы</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое шина</h3><p>Зачем вообще существует шина: разговор по проводам, с кадрами и адресами, и кому позволено говорить.</p><ol><li><a href="/ru/learn/wires-that-talk/">01 Провода, которые говорят</a></li><li><a href="/ru/learn/framing-a-conversation/">02 Кадрирование разговора</a></li><li><a href="/ru/learn/master-and-servant/">03 Хозяин и слуга</a></li></ol></li><li><h3>Модуль 2 · Разговор по UART</h3><p>Двухпроводная шина без такта: кадр, делитель, задающий скорость, и статус, который опрашивает драйвер.</p><ol><li><a href="/ru/learn/the-frame/">04 Кадр: старт, данные, стоп</a></li><li><a href="/ru/learn/baud-and-the-divisor/">05 Бод и делитель</a></li><li><a href="/ru/learn/status-and-fifos/">06 Статус и FIFO</a></li></ol></li><li><h3>Модуль 3 · SPI под тактом</h3><p>Разговор под тактом: четыре режима, лестница предделителей и выбор кристалла на каждого подчинённого.</p><ol><li><a href="/ru/learn/modes-cpol-cpha/">07 Режимы: CPOL и CPHA</a></li><li><a href="/ru/learn/the-prescaler-ladder/">08 Лестница предделителей</a></li><li><a href="/ru/learn/chip-select-and-widths/">09 Выбор кристалла и ширины</a></li></ol></li><li><h3>Модуль 4 · Регистровая шина APB</h3><p>Шина регистров: PSEL и PENABLE, строобы и ожидания, и сколько адресных бит стоит количество периферии.</p><ol><li><a href="/ru/learn/psel-and-penable/">10 PSEL и PENABLE</a></li><li><a href="/ru/learn/strobes-and-waits/">11 Строобы и ожидания</a></li><li><a href="/ru/learn/address-decode/">12 Дешифрация адреса</a></li></ol></li><li><h3>Модуль 5 · Пять каналов AXI4</h3><p>Пять каналов: адрес, данные и отклик в обе стороны, lite или full, пакеты и идентификаторы.</p><ol><li><a href="/ru/learn/five-channels/">13 Пять каналов</a></li><li><a href="/ru/learn/lite-or-full/">14 Lite или full</a></li><li><a href="/ru/learn/bursts-and-ids/">15 Пакеты и идентификаторы</a></li></ol></li><li><h3>Модуль 6 · Память</h3><p>Что находится на дальнем конце каждой шины: карты памяти, типы портов и задержка, которую должно покрывать ожидание.</p><ol><li><a href="/ru/learn/memory-maps/">16 Карты памяти</a></li><li class="on" aria-current="page"><a href="/ru/learn/ports-and-kinds/">17 Порты и типы</a></li><li><a href="/ru/learn/latency/">18 Задержка</a></li></ol></li><li><h3>Модуль 7 · Мосты</h3><p>Почему в дизайне появляется больше одной шины и пакетный мост, переносящий работу между ними.</p><ol><li><a href="/ru/learn/why-bridges/">19 Зачем мосты</a></li><li><a href="/ru/learn/the-packet-bridge/">20 Пакетный мост</a></li><li><a href="/ru/learn/everything-in-one-bench/">21 Всё в одном стенде</a></li></ol></li><li><h3>Модуль 8 · Ethernet: кадры и тайминг</h3><p>Кадры, контрольная сумма кадра, тайминг RGMII и заранее зарегистрированные шаги настоящего запуска.</p><ol><li><a href="/ru/learn/frames-and-the-fcs/">22 Кадры и контрольная сумма</a></li><li><a href="/ru/learn/rgmii-timing/">23 Тайминг RGMII</a></li><li><a href="/ru/learn/bring-up-on-the-bench/">24 Запуск на стенде</a></li></ol></li><li><h3>Модуль 9 · Стенд</h3><p>Дисциплина, оберегающая настоящее железо: кто держит IO, как взять и вернуть захват, и что запускается дальше.</p><ol><li><a href="/ru/learn/who-holds-the-io/">25 Кто держит IO</a></li><li><a href="/ru/learn/claim-and-release/">26 Захват и возврат</a></li><li><a href="/ru/learn/capstone-what-runs-next/">27 Капстоун: что запускается дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/prefix-adders/index.html b/apps/website/public/ru/learn/prefix-adders/index.html
index 615641ae92..44c9a7b592 100644
--- a/apps/website/public/ru/learn/prefix-adders/index.html
+++ b/apps/website/public/ru/learn/prefix-adders/index.html
@@ -554,5 +554,6 @@ <h1>Взгляд вперёд: префиксные сумматоры</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Сложение</h3><p>Почему сложение медленное: перенос, который идёт вверх по слову, выделенная цепь в каждом слайсе, префиксные сети и деревья сохранения переноса.</p><ol><li><a href="/ru/learn/the-carry-chain/">01 Цепь переноса</a></li><li class="on" aria-current="page"><a href="/ru/learn/prefix-adders/">02 Взгляд вперёд: префиксные сумматоры</a></li><li><a href="/ru/learn/carry-save/">03 Три числа сразу</a></li></ol></li><li><h3>Модуль 2 · Умножение</h3><p>Произведение — сумма сдвинутых копий; перекодирование Бута вдвое сокращает их, а слайс DSP48E1 делает остальное одним блоком.</p><ol><li><a href="/ru/learn/shift-and-add/">04 Сдвиг и сложение</a></li><li><a href="/ru/learn/booths-trick/">05 Приём Бута</a></li><li><a href="/ru/learn/the-dsp48e1-slice/">06 Слайс DSP48E1</a></li></ol></li><li><h3>Модуль 3 · Фиксированная точка</h3><p>Где стоит двоичная точка, что округление делает со значением и с его средним и что даёт каждый бит квантователя.</p><ol><li><a href="/ru/learn/q-formats/">07 Форматы Q</a></li><li><a href="/ru/learn/rounding-and-saturation/">08 Округление и насыщение</a></li><li><a href="/ru/learn/quantization-noise/">09 Шум квантования</a></li></ol></li><li><h3>Модуль 4 · Функции в железе</h3><p>Синус, косинус, угол и длина из сдвигов и сложений — и когда таблица оказывается выгоднее.</p><ol><li><a href="/ru/learn/rotating-by-shifts/">10 Поворот сдвигами</a></li><li><a href="/ru/learn/angle-and-magnitude/">11 Угол и модуль</a></li><li><a href="/ru/learn/tables-and-interpolation/">12 Таблицы и интерполяция</a></li></ol></li><li><h3>Модуль 5 · Сигналы и дискретизация</h3><p>Что дискретизация делает с частотой, как генератор строится из сумматора и что измеряет бин ДПФ.</p><ol><li><a href="/ru/learn/sampling-and-aliasing/">13 Дискретизация и наложение спектров</a></li><li><a href="/ru/learn/the-nco/">14 Генератор с числовым управлением</a></li><li><a href="/ru/learn/bins-of-a-spectrum/">15 Бины спектра</a></li></ol></li><li><h3>Модуль 6 · Фильтры FIR</h3><p>Скользящее среднее, расчёт окном sinc с целыми коэффициентами и раскладка отводов по слайсам DSP.</p><ol><li><a href="/ru/learn/the-moving-average/">16 Скользящее среднее</a></li><li><a href="/ru/learn/designing-taps/">17 Расчёт коэффициентов</a></li><li><a href="/ru/learn/folding-taps-onto-slices/">18 Раскладка отводов по слайсам</a></li></ol></li><li><h3>Модуль 7 · Многоскоростная обработка</h3><p>Понижение частоты дискретизации без заворота шума: децимация, фильтр CIC и полифазная форма.</p><ol><li><a href="/ru/learn/lowering-the-rate/">19 Понижение частоты</a></li><li><a href="/ru/learn/the-cic-filter/">20 Фильтр CIC</a></li><li><a href="/ru/learn/polyphase/">21 Полифазная форма</a></li></ol></li><li><h3>Модуль 8 · БПФ</h3><p>N log N вместо N^2: бабочки, бит-реверсный порядок входа и биты, которые добавляет каждый этап.</p><ol><li><a href="/ru/learn/the-butterfly/">22 Бабочка</a></li><li><a href="/ru/learn/twiddles-and-bit-reversal/">23 Поворачивающие множители и бит-реверс</a></li><li><a href="/ru/learn/growth-through-the-stages/">24 Рост через этапы</a></li></ol></li><li><h3>Модуль 9 · На стенде</h3><p>От арифметики к плате: коррелятор из спеки модема, бюджет слайсов и тайминга и итоговый фильтр.</p><ol><li><a href="/ru/learn/a-correlator-finds-the-frame/">25 Коррелятор находит кадр</a></li><li><a href="/ru/learn/does-it-fit-and-close-timing/">26 Помещается ли и сходится ли тайминг?</a></li><li><a href="/ru/learn/capstone-a-filter-on-the-board/">27 Итог: фильтр на плате</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/psel-and-penable/index.html b/apps/website/public/ru/learn/psel-and-penable/index.html
index ea24f171be..e2e37ffb35 100644
--- a/apps/website/public/ru/learn/psel-and-penable/index.html
+++ b/apps/website/public/ru/learn/psel-and-penable/index.html
@@ -417,5 +417,6 @@ <h1>PSEL и PENABLE</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое шина</h3><p>Зачем вообще существует шина: разговор по проводам, с кадрами и адресами, и кому позволено говорить.</p><ol><li><a href="/ru/learn/wires-that-talk/">01 Провода, которые говорят</a></li><li><a href="/ru/learn/framing-a-conversation/">02 Кадрирование разговора</a></li><li><a href="/ru/learn/master-and-servant/">03 Хозяин и слуга</a></li></ol></li><li><h3>Модуль 2 · Разговор по UART</h3><p>Двухпроводная шина без такта: кадр, делитель, задающий скорость, и статус, который опрашивает драйвер.</p><ol><li><a href="/ru/learn/the-frame/">04 Кадр: старт, данные, стоп</a></li><li><a href="/ru/learn/baud-and-the-divisor/">05 Бод и делитель</a></li><li><a href="/ru/learn/status-and-fifos/">06 Статус и FIFO</a></li></ol></li><li><h3>Модуль 3 · SPI под тактом</h3><p>Разговор под тактом: четыре режима, лестница предделителей и выбор кристалла на каждого подчинённого.</p><ol><li><a href="/ru/learn/modes-cpol-cpha/">07 Режимы: CPOL и CPHA</a></li><li><a href="/ru/learn/the-prescaler-ladder/">08 Лестница предделителей</a></li><li><a href="/ru/learn/chip-select-and-widths/">09 Выбор кристалла и ширины</a></li></ol></li><li><h3>Модуль 4 · Регистровая шина APB</h3><p>Шина регистров: PSEL и PENABLE, строобы и ожидания, и сколько адресных бит стоит количество периферии.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/psel-and-penable/">10 PSEL и PENABLE</a></li><li><a href="/ru/learn/strobes-and-waits/">11 Строобы и ожидания</a></li><li><a href="/ru/learn/address-decode/">12 Дешифрация адреса</a></li></ol></li><li><h3>Модуль 5 · Пять каналов AXI4</h3><p>Пять каналов: адрес, данные и отклик в обе стороны, lite или full, пакеты и идентификаторы.</p><ol><li><a href="/ru/learn/five-channels/">13 Пять каналов</a></li><li><a href="/ru/learn/lite-or-full/">14 Lite или full</a></li><li><a href="/ru/learn/bursts-and-ids/">15 Пакеты и идентификаторы</a></li></ol></li><li><h3>Модуль 6 · Память</h3><p>Что находится на дальнем конце каждой шины: карты памяти, типы портов и задержка, которую должно покрывать ожидание.</p><ol><li><a href="/ru/learn/memory-maps/">16 Карты памяти</a></li><li><a href="/ru/learn/ports-and-kinds/">17 Порты и типы</a></li><li><a href="/ru/learn/latency/">18 Задержка</a></li></ol></li><li><h3>Модуль 7 · Мосты</h3><p>Почему в дизайне появляется больше одной шины и пакетный мост, переносящий работу между ними.</p><ol><li><a href="/ru/learn/why-bridges/">19 Зачем мосты</a></li><li><a href="/ru/learn/the-packet-bridge/">20 Пакетный мост</a></li><li><a href="/ru/learn/everything-in-one-bench/">21 Всё в одном стенде</a></li></ol></li><li><h3>Модуль 8 · Ethernet: кадры и тайминг</h3><p>Кадры, контрольная сумма кадра, тайминг RGMII и заранее зарегистрированные шаги настоящего запуска.</p><ol><li><a href="/ru/learn/frames-and-the-fcs/">22 Кадры и контрольная сумма</a></li><li><a href="/ru/learn/rgmii-timing/">23 Тайминг RGMII</a></li><li><a href="/ru/learn/bring-up-on-the-bench/">24 Запуск на стенде</a></li></ol></li><li><h3>Модуль 9 · Стенд</h3><p>Дисциплина, оберегающая настоящее железо: кто держит IO, как взять и вернуть захват, и что запускается дальше.</p><ol><li><a href="/ru/learn/who-holds-the-io/">25 Кто держит IO</a></li><li><a href="/ru/learn/claim-and-release/">26 Захват и возврат</a></li><li><a href="/ru/learn/capstone-what-runs-next/">27 Капстоун: что запускается дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/pub-or-private/index.html b/apps/website/public/ru/learn/pub-or-private/index.html
index 68e24ce2cc..86dc0f4753 100644
--- a/apps/website/public/ru/learn/pub-or-private/index.html
+++ b/apps/website/public/ru/learn/pub-or-private/index.html
@@ -83,5 +83,6 @@ <h1>pub или закрытое имя</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Файл спеки</h3><p>Что лежит в файле .t27: строка module, строки прозы и комментарии.</p><ol><li><a href="/ru/learn/what-a-spec-is/">01 Что такое спека</a></li><li><a href="/ru/learn/the-module-line/">02 Строка module</a></li><li><a href="/ru/learn/comments-and-prose/">03 Комментарии и строки прозы</a></li></ol></li><li><h3>Модуль 2 · Значения и типы</h3><p>Константы, ширина целого числа, а также true, false и текст.</p><ol><li><a href="/ru/learn/constants/">04 Константы</a></li><li><a href="/ru/learn/whole-numbers/">05 Целые числа и их ширина</a></li><li><a href="/ru/learn/true-false-and-text/">06 Истина, ложь и текст</a></li></ol></li><li><h3>Модуль 3 · Массивы, триты и выражения</h3><p>Списки значений, три значения трита и что вычисляют операторы.</p><ol><li><a href="/ru/learn/arrays/">07 Массивы</a></li><li><a href="/ru/learn/trits/">08 Триты: минус один, ноль, один</a></li><li><a href="/ru/learn/expressions/">09 Выражения</a></li></ol></li><li><h3>Модуль 4 · Тесты и функции</h3><p>Блок test, несколько маленьких тестов в одной спеке и функция с типизированными входами.</p><ol><li><a href="/ru/learn/a-test-block/">10 Блок test</a></li><li><a href="/ru/learn/many-tests/">11 Много тестов в одной спеке</a></li><li><a href="/ru/learn/functions/">12 Функции</a></li></ol></li><li><h3>Модуль 5 · Внутри функции</h3><p>Имена через let и var, выбор через if и switch и циклы.</p><ol><li><a href="/ru/learn/local-names/">13 Локальные имена: let и var</a></li><li><a href="/ru/learn/if-and-else/">14 Выбор: if, else и switch</a></li><li><a href="/ru/learn/loops/">15 Циклы: while и for</a></li></ol></li><li><h3>Модуль 6 · Видимость и свои типы</h3><p>Что pub открывает другим модулям, и структуры и перечисления, которые вы объявляете.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/pub-or-private/">16 pub или закрытое имя</a></li><li><a href="/ru/learn/structs/">17 Структуры</a></li><li><a href="/ru/learn/enums/">18 Перечисления</a></li></ol></li><li><h3>Модуль 7 · Модули, правила и gen-ts</h3><p>Чтение другого модуля, правило, которое выполняется всегда, и спека, превращённая в TypeScript.</p><ol><li><a href="/ru/learn/use-other-modules/">19 Другие модули</a></li><li><a href="/ru/learn/invariants/">20 Инварианты и замеры</a></li><li><a href="/ru/learn/gen-ts/">21 gen-ts: спека превращается в TypeScript</a></li></ol></li><li><h3>Модуль 8 · Компилятор и tri</h3><p>Бэкенды, которые пишет t27c, стадии, в которых он читает спеку, и команда tri.</p><ol><li><a href="/ru/learn/seven-backends/">22 Семь бэкендов</a></li><li><a href="/ru/learn/stages-of-t27c/">23 Стадии t27c</a></li><li><a href="/ru/learn/tri-commands/">24 tri: командная строка</a></li></ol></li><li><h3>Модуль 9 · Ошибки, программа и что дальше</h3><p>Как читать ошибку компилятора, одна маленькая законченная программа и какой курс пройти дальше.</p><ol><li><a href="/ru/learn/reading-errors/">25 Как читать ошибки компилятора</a></li><li><a href="/ru/learn/a-small-program/">26 Маленькая законченная программа</a></li><li><a href="/ru/learn/where-next/">27 Куда дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/q-formats/index.html b/apps/website/public/ru/learn/q-formats/index.html
index 8f8e356d4c..e3c0d11aae 100644
--- a/apps/website/public/ru/learn/q-formats/index.html
+++ b/apps/website/public/ru/learn/q-formats/index.html
@@ -346,5 +346,6 @@ <h1>Форматы Q</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Сложение</h3><p>Почему сложение медленное: перенос, который идёт вверх по слову, выделенная цепь в каждом слайсе, префиксные сети и деревья сохранения переноса.</p><ol><li><a href="/ru/learn/the-carry-chain/">01 Цепь переноса</a></li><li><a href="/ru/learn/prefix-adders/">02 Взгляд вперёд: префиксные сумматоры</a></li><li><a href="/ru/learn/carry-save/">03 Три числа сразу</a></li></ol></li><li><h3>Модуль 2 · Умножение</h3><p>Произведение — сумма сдвинутых копий; перекодирование Бута вдвое сокращает их, а слайс DSP48E1 делает остальное одним блоком.</p><ol><li><a href="/ru/learn/shift-and-add/">04 Сдвиг и сложение</a></li><li><a href="/ru/learn/booths-trick/">05 Приём Бута</a></li><li><a href="/ru/learn/the-dsp48e1-slice/">06 Слайс DSP48E1</a></li></ol></li><li><h3>Модуль 3 · Фиксированная точка</h3><p>Где стоит двоичная точка, что округление делает со значением и с его средним и что даёт каждый бит квантователя.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/q-formats/">07 Форматы Q</a></li><li><a href="/ru/learn/rounding-and-saturation/">08 Округление и насыщение</a></li><li><a href="/ru/learn/quantization-noise/">09 Шум квантования</a></li></ol></li><li><h3>Модуль 4 · Функции в железе</h3><p>Синус, косинус, угол и длина из сдвигов и сложений — и когда таблица оказывается выгоднее.</p><ol><li><a href="/ru/learn/rotating-by-shifts/">10 Поворот сдвигами</a></li><li><a href="/ru/learn/angle-and-magnitude/">11 Угол и модуль</a></li><li><a href="/ru/learn/tables-and-interpolation/">12 Таблицы и интерполяция</a></li></ol></li><li><h3>Модуль 5 · Сигналы и дискретизация</h3><p>Что дискретизация делает с частотой, как генератор строится из сумматора и что измеряет бин ДПФ.</p><ol><li><a href="/ru/learn/sampling-and-aliasing/">13 Дискретизация и наложение спектров</a></li><li><a href="/ru/learn/the-nco/">14 Генератор с числовым управлением</a></li><li><a href="/ru/learn/bins-of-a-spectrum/">15 Бины спектра</a></li></ol></li><li><h3>Модуль 6 · Фильтры FIR</h3><p>Скользящее среднее, расчёт окном sinc с целыми коэффициентами и раскладка отводов по слайсам DSP.</p><ol><li><a href="/ru/learn/the-moving-average/">16 Скользящее среднее</a></li><li><a href="/ru/learn/designing-taps/">17 Расчёт коэффициентов</a></li><li><a href="/ru/learn/folding-taps-onto-slices/">18 Раскладка отводов по слайсам</a></li></ol></li><li><h3>Модуль 7 · Многоскоростная обработка</h3><p>Понижение частоты дискретизации без заворота шума: децимация, фильтр CIC и полифазная форма.</p><ol><li><a href="/ru/learn/lowering-the-rate/">19 Понижение частоты</a></li><li><a href="/ru/learn/the-cic-filter/">20 Фильтр CIC</a></li><li><a href="/ru/learn/polyphase/">21 Полифазная форма</a></li></ol></li><li><h3>Модуль 8 · БПФ</h3><p>N log N вместо N^2: бабочки, бит-реверсный порядок входа и биты, которые добавляет каждый этап.</p><ol><li><a href="/ru/learn/the-butterfly/">22 Бабочка</a></li><li><a href="/ru/learn/twiddles-and-bit-reversal/">23 Поворачивающие множители и бит-реверс</a></li><li><a href="/ru/learn/growth-through-the-stages/">24 Рост через этапы</a></li></ol></li><li><h3>Модуль 9 · На стенде</h3><p>От арифметики к плате: коррелятор из спеки модема, бюджет слайсов и тайминга и итоговый фильтр.</p><ol><li><a href="/ru/learn/a-correlator-finds-the-frame/">25 Коррелятор находит кадр</a></li><li><a href="/ru/learn/does-it-fit-and-close-timing/">26 Помещается ли и сходится ли тайминг?</a></li><li><a href="/ru/learn/capstone-a-filter-on-the-board/">27 Итог: фильтр на плате</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/quantization-noise/index.html b/apps/website/public/ru/learn/quantization-noise/index.html
index 5f95de1f2f..b9cc0c21f3 100644
--- a/apps/website/public/ru/learn/quantization-noise/index.html
+++ b/apps/website/public/ru/learn/quantization-noise/index.html
@@ -1122,5 +1122,6 @@ <h1>Шум квантования</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Сложение</h3><p>Почему сложение медленное: перенос, который идёт вверх по слову, выделенная цепь в каждом слайсе, префиксные сети и деревья сохранения переноса.</p><ol><li><a href="/ru/learn/the-carry-chain/">01 Цепь переноса</a></li><li><a href="/ru/learn/prefix-adders/">02 Взгляд вперёд: префиксные сумматоры</a></li><li><a href="/ru/learn/carry-save/">03 Три числа сразу</a></li></ol></li><li><h3>Модуль 2 · Умножение</h3><p>Произведение — сумма сдвинутых копий; перекодирование Бута вдвое сокращает их, а слайс DSP48E1 делает остальное одним блоком.</p><ol><li><a href="/ru/learn/shift-and-add/">04 Сдвиг и сложение</a></li><li><a href="/ru/learn/booths-trick/">05 Приём Бута</a></li><li><a href="/ru/learn/the-dsp48e1-slice/">06 Слайс DSP48E1</a></li></ol></li><li><h3>Модуль 3 · Фиксированная точка</h3><p>Где стоит двоичная точка, что округление делает со значением и с его средним и что даёт каждый бит квантователя.</p><ol><li><a href="/ru/learn/q-formats/">07 Форматы Q</a></li><li><a href="/ru/learn/rounding-and-saturation/">08 Округление и насыщение</a></li><li class="on" aria-current="page"><a href="/ru/learn/quantization-noise/">09 Шум квантования</a></li></ol></li><li><h3>Модуль 4 · Функции в железе</h3><p>Синус, косинус, угол и длина из сдвигов и сложений — и когда таблица оказывается выгоднее.</p><ol><li><a href="/ru/learn/rotating-by-shifts/">10 Поворот сдвигами</a></li><li><a href="/ru/learn/angle-and-magnitude/">11 Угол и модуль</a></li><li><a href="/ru/learn/tables-and-interpolation/">12 Таблицы и интерполяция</a></li></ol></li><li><h3>Модуль 5 · Сигналы и дискретизация</h3><p>Что дискретизация делает с частотой, как генератор строится из сумматора и что измеряет бин ДПФ.</p><ol><li><a href="/ru/learn/sampling-and-aliasing/">13 Дискретизация и наложение спектров</a></li><li><a href="/ru/learn/the-nco/">14 Генератор с числовым управлением</a></li><li><a href="/ru/learn/bins-of-a-spectrum/">15 Бины спектра</a></li></ol></li><li><h3>Модуль 6 · Фильтры FIR</h3><p>Скользящее среднее, расчёт окном sinc с целыми коэффициентами и раскладка отводов по слайсам DSP.</p><ol><li><a href="/ru/learn/the-moving-average/">16 Скользящее среднее</a></li><li><a href="/ru/learn/designing-taps/">17 Расчёт коэффициентов</a></li><li><a href="/ru/learn/folding-taps-onto-slices/">18 Раскладка отводов по слайсам</a></li></ol></li><li><h3>Модуль 7 · Многоскоростная обработка</h3><p>Понижение частоты дискретизации без заворота шума: децимация, фильтр CIC и полифазная форма.</p><ol><li><a href="/ru/learn/lowering-the-rate/">19 Понижение частоты</a></li><li><a href="/ru/learn/the-cic-filter/">20 Фильтр CIC</a></li><li><a href="/ru/learn/polyphase/">21 Полифазная форма</a></li></ol></li><li><h3>Модуль 8 · БПФ</h3><p>N log N вместо N^2: бабочки, бит-реверсный порядок входа и биты, которые добавляет каждый этап.</p><ol><li><a href="/ru/learn/the-butterfly/">22 Бабочка</a></li><li><a href="/ru/learn/twiddles-and-bit-reversal/">23 Поворачивающие множители и бит-реверс</a></li><li><a href="/ru/learn/growth-through-the-stages/">24 Рост через этапы</a></li></ol></li><li><h3>Модуль 9 · На стенде</h3><p>От арифметики к плате: коррелятор из спеки модема, бюджет слайсов и тайминга и итоговый фильтр.</p><ol><li><a href="/ru/learn/a-correlator-finds-the-frame/">25 Коррелятор находит кадр</a></li><li><a href="/ru/learn/does-it-fit-and-close-timing/">26 Помещается ли и сходится ли тайминг?</a></li><li><a href="/ru/learn/capstone-a-filter-on-the-board/">27 Итог: фильтр на плате</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/questions-in-thirds/index.html b/apps/website/public/ru/learn/questions-in-thirds/index.html
index b7ff0ad424..cc0395c66f 100644
--- a/apps/website/public/ru/learn/questions-in-thirds/index.html
+++ b/apps/website/public/ru/learn/questions-in-thirds/index.html
@@ -769,5 +769,6 @@ <h1>Вопросы по третям</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Три значения</h3><p>Почему три, как сбалансированная троичная система пишет любое число без знака и что дают переворот и отсечение тритов.</p><ol><li><a href="/ru/learn/why-three/">01 Почему три</a></li><li><a href="/ru/learn/balanced-ternary/">02 Сбалансированная троичная</a></li><li><a href="/ru/learn/flip-and-cut/">03 Переворот вместо минуса, отсечение вместо округления</a></li></ol></li><li><h3>Модуль 2 · Логика с «не знаю»</h3><p>Трёхзначные вентили Клини, два вентиля без двоичного двойника и сложение как пара таблиц.</p><ol><li><a href="/ru/learn/kleene-logic/">04 Трёхзначная логика</a></li><li><a href="/ru/learn/consensus-and-any/">05 Консенсус и «любой»</a></li><li><a href="/ru/learn/sum-and-carry/">06 Сумма и перенос</a></li></ol></li><li><h3>Модуль 3 · Складываем триты</h3><p>Полусумматор, полный сумматор и перенос, бегущий влево разряд за разрядом.</p><ol><li><a href="/ru/learn/half-adder/">07 Полусумматор</a></li><li><a href="/ru/learn/full-adder/">08 Полный сумматор</a></li><li><a href="/ru/learn/ripple-carry/">09 Сквозной перенос</a></li></ol></li><li><h3>Модуль 4 · Умножение без умножителя</h3><p>Копия, ноль или переворот: произведение на один трит, умножение столбиком и MAC, который только складывает.</p><ol><li><a href="/ru/learn/times-a-trit/">10 Умножить на один трит</a></li><li><a href="/ru/learn/long-multiplication/">11 Умножение столбиком</a></li><li><a href="/ru/learn/ternary-mac/">12 Троичный MAC</a></li></ol></li><li><h3>Модуль 5 · Триты в двоичной памяти</h3><p>Два бита на трит, пять тритов на байт и сколько бит нужно слову из 27 тритов.</p><ol><li><a href="/ru/learn/two-bits-per-trit/">13 Два бита на трит</a></li><li><a href="/ru/learn/five-trits-a-byte/">14 Пять тритов в байте</a></li><li><a href="/ru/learn/a-word-of-27/">15 Слово из 27 тритов</a></li></ol></li><li><h3>Модуль 6 · Командное слово TRI-27</h3><p>32-битное слово, которое декодирует эмулятор Trinity: его поля, 47 опкодов и 15-битное непосредственное значение.</p><ol><li><a href="/ru/learn/the-instruction-word/">16 Командное слово</a></li><li><a href="/ru/learn/the-opcode-map/">17 Карта опкодов</a></li><li><a href="/ru/learn/the-immediate/">18 Непосредственное значение</a></li></ol></li><li><h3>Модуль 7 · Троичная машина</h3><p>АЛУ из сумматоров этого курса, переход на три стороны и программа, которую можно пройти по шагам.</p><ol><li><a href="/ru/learn/a-ternary-alu/">19 Троичное АЛУ</a></li><li><a href="/ru/learn/three-way-jump/">20 Переход на три стороны</a></li><li><a href="/ru/learn/run-a-program/">21 Запустить программу</a></li></ol></li><li><h3>Модуль 8 · Три ответа</h3><p>Сравнение по старшему различному триту, поиск числа по третям и сортировка трёхзначными сравнениями.</p><ol><li><a href="/ru/learn/compare-at-the-top/">22 Сравнение сверху</a></li><li class="on" aria-current="page"><a href="/ru/learn/questions-in-thirds/">23 Вопросы по третям</a></li><li><a href="/ru/learn/sort-by-three/">24 Сортировка трёхзначными сравнениями</a></li></ol></li><li><h3>Модуль 9 · Троичные нейроны</h3><p>Веса -1, 0, +1: один нейрон, детектор и маленький слой, и нигде нет умножителя.</p><ol><li><a href="/ru/learn/one-ternary-neuron/">25 Один троичный нейрон</a></li><li><a href="/ru/learn/a-bar-detector/">26 Детектор полосы</a></li><li><a href="/ru/learn/a-tiny-layer/">27 Маленький слой</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/read-a-bitstream/index.html b/apps/website/public/ru/learn/read-a-bitstream/index.html
index f4987ae826..1a5a6b6a5b 100644
--- a/apps/website/public/ru/learn/read-a-bitstream/index.html
+++ b/apps/website/public/ru/learn/read-a-bitstream/index.html
@@ -110,5 +110,6 @@ <h1>Читаем битстрим</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Чип</h3><p>Что такое FPGA, какой чип у нас и как его выводы встречаются с платой.</p><ol><li><a href="/ru/learn/fpga-is-a-grid/">01 FPGA — это сетка, которую вы настраиваете</a></li><li><a href="/ru/learn/which-chip/">02 Какой это чип?</a></li><li><a href="/ru/learn/pins-and-package/">03 Выводы, шарики и корпус</a></li></ol></li><li><h3>Модуль 2 · Числа в железе</h3><p>Биты, триты и форматы чисел, и сколько стоит арифметика в логике.</p><ol><li><a href="/ru/learn/bits-and-trits/">04 Биты, триты и GF16</a></li><li><a href="/ru/learn/what-arithmetic-costs/">05 Сколько стоит арифметика</a></li><li><a href="/ru/learn/how-much-fits/">06 Сколько помещается на чип</a></li></ol></li><li><h3>Модуль 3 · Ваша программа на t27</h3><p>Напишите спеку, проверьте её и увидьте, почему «компилируется» не значит «верно».</p><ol><li><a href="/ru/learn/hello-t27/">07 Привет, t27</a></li><li><a href="/ru/learn/tests-are-the-spec/">08 Тесты — часть спеки</a></li><li><a href="/ru/learn/compiles-is-not-correct/">09 Компилируется — не значит верно</a></li></ol></li><li><h3>Модуль 4 · Внутри t27c</h3><p>Как компилятор читает спеку и что он пишет, включая нативный t27b.</p><ol><li><a href="/ru/learn/compiler-stages/">10 Стадии t27c</a></li><li><a href="/ru/learn/t27c-on-a-real-machine/">11 t27c на настоящей машине</a></li><li><a href="/ru/learn/seven-backends-and-t27b/">12 Семь бэкендов и t27b</a></li></ol></li><li><h3>Модуль 5 · От спеки к железу</h3><p>Verilog, который пишет t27c, ячейки, в которые он превращается, и что делает одна LUT.</p><ol><li><a href="/ru/learn/spec-to-verilog/">13 От спеки к Verilog</a></li><li><a href="/ru/learn/count-the-cells/">14 Посчитайте ячейки</a></li><li><a href="/ru/learn/one-lut-up-close/">15 Одна LUT вблизи</a></li></ol></li><li><h3>Модуль 6 · Читаем синтез</h3><p>Что yosys сообщает о вашем дизайне и какие предупреждения важны.</p><ol><li><a href="/ru/learn/synthesis-warnings/">16 Важные предупреждения синтеза</a></li><li><a href="/ru/learn/state-machines/">17 Конечные автоматы</a></li><li><a href="/ru/learn/where-the-luts-go/">18 Куда уходят LUT</a></li></ol></li><li><h3>Модуль 7 · Размещение, трассировка, тайминг</h3><p>Где ячейки оказываются на кристалле и успевает ли тактовая частота.</p><ol><li><a href="/ru/learn/placement/">19 Размещение: где оказываются ячейки</a></li><li><a href="/ru/learn/logic-depth/">20 Глубина логики</a></li><li><a href="/ru/learn/timing-slack/">21 Тайминг и запас</a></li></ol></li><li><h3>Модуль 8 · Битстрим</h3><p>Как оттрассированный дизайн становится битами, которые загружает чип, без Vivado.</p><ol><li><a href="/ru/learn/fasm-to-frames/">22 От FASM к фреймам</a></li><li><a href="/ru/learn/a-build-receipt/">23 Сборка по пунктам</a></li><li class="on" aria-current="page"><a href="/ru/learn/read-a-bitstream/">24 Читаем битстрим</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Спросите чип, кто он, загрузите биты и проверьте их.</p><ol><li><a href="/ru/learn/jtag-asks-the-chip/">25 JTAG спрашивает чип</a></li><li><a href="/ru/learn/same-bits-on-the-board/">26 Те же биты — на плате</a></li><li><a href="/ru/learn/capstone/">27 Итог: переугадайте yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/reading-errors/index.html b/apps/website/public/ru/learn/reading-errors/index.html
index 89d2d71c19..27548b9d42 100644
--- a/apps/website/public/ru/learn/reading-errors/index.html
+++ b/apps/website/public/ru/learn/reading-errors/index.html
@@ -81,5 +81,6 @@ <h1>Как читать ошибки компилятора</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Файл спеки</h3><p>Что лежит в файле .t27: строка module, строки прозы и комментарии.</p><ol><li><a href="/ru/learn/what-a-spec-is/">01 Что такое спека</a></li><li><a href="/ru/learn/the-module-line/">02 Строка module</a></li><li><a href="/ru/learn/comments-and-prose/">03 Комментарии и строки прозы</a></li></ol></li><li><h3>Модуль 2 · Значения и типы</h3><p>Константы, ширина целого числа, а также true, false и текст.</p><ol><li><a href="/ru/learn/constants/">04 Константы</a></li><li><a href="/ru/learn/whole-numbers/">05 Целые числа и их ширина</a></li><li><a href="/ru/learn/true-false-and-text/">06 Истина, ложь и текст</a></li></ol></li><li><h3>Модуль 3 · Массивы, триты и выражения</h3><p>Списки значений, три значения трита и что вычисляют операторы.</p><ol><li><a href="/ru/learn/arrays/">07 Массивы</a></li><li><a href="/ru/learn/trits/">08 Триты: минус один, ноль, один</a></li><li><a href="/ru/learn/expressions/">09 Выражения</a></li></ol></li><li><h3>Модуль 4 · Тесты и функции</h3><p>Блок test, несколько маленьких тестов в одной спеке и функция с типизированными входами.</p><ol><li><a href="/ru/learn/a-test-block/">10 Блок test</a></li><li><a href="/ru/learn/many-tests/">11 Много тестов в одной спеке</a></li><li><a href="/ru/learn/functions/">12 Функции</a></li></ol></li><li><h3>Модуль 5 · Внутри функции</h3><p>Имена через let и var, выбор через if и switch и циклы.</p><ol><li><a href="/ru/learn/local-names/">13 Локальные имена: let и var</a></li><li><a href="/ru/learn/if-and-else/">14 Выбор: if, else и switch</a></li><li><a href="/ru/learn/loops/">15 Циклы: while и for</a></li></ol></li><li><h3>Модуль 6 · Видимость и свои типы</h3><p>Что pub открывает другим модулям, и структуры и перечисления, которые вы объявляете.</p><ol><li><a href="/ru/learn/pub-or-private/">16 pub или закрытое имя</a></li><li><a href="/ru/learn/structs/">17 Структуры</a></li><li><a href="/ru/learn/enums/">18 Перечисления</a></li></ol></li><li><h3>Модуль 7 · Модули, правила и gen-ts</h3><p>Чтение другого модуля, правило, которое выполняется всегда, и спека, превращённая в TypeScript.</p><ol><li><a href="/ru/learn/use-other-modules/">19 Другие модули</a></li><li><a href="/ru/learn/invariants/">20 Инварианты и замеры</a></li><li><a href="/ru/learn/gen-ts/">21 gen-ts: спека превращается в TypeScript</a></li></ol></li><li><h3>Модуль 8 · Компилятор и tri</h3><p>Бэкенды, которые пишет t27c, стадии, в которых он читает спеку, и команда tri.</p><ol><li><a href="/ru/learn/seven-backends/">22 Семь бэкендов</a></li><li><a href="/ru/learn/stages-of-t27c/">23 Стадии t27c</a></li><li><a href="/ru/learn/tri-commands/">24 tri: командная строка</a></li></ol></li><li><h3>Модуль 9 · Ошибки, программа и что дальше</h3><p>Как читать ошибку компилятора, одна маленькая законченная программа и какой курс пройти дальше.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/reading-errors/">25 Как читать ошибки компилятора</a></li><li><a href="/ru/learn/a-small-program/">26 Маленькая законченная программа</a></li><li><a href="/ru/learn/where-next/">27 Куда дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/related-or-not/index.html b/apps/website/public/ru/learn/related-or-not/index.html
index 6ecef56de6..9b1a46e48f 100644
--- a/apps/website/public/ru/learn/related-or-not/index.html
+++ b/apps/website/public/ru/learn/related-or-not/index.html
@@ -448,5 +448,6 @@ <h1>Родственные или чужие</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое тактовый сигнал</h3><p>Один фронт, один мир: всё, что делит фронт тактового сигнала, делит и мир; период и джиттер настоящего фронта и место, где тактовый сигнал входит в плату.</p><ol><li><a href="/ru/learn/one-edge-one-world/">01 Один фронт, один мир</a></li><li><a href="/ru/learn/period-and-jitter/">02 Период, фронт и джиттер</a></li><li><a href="/ru/learn/where-the-clock-enters/">03 Где входит тактовый сигнал</a></li></ol></li><li><h3>Модуль 2 · Деревья тактового сигнала</h3><p>Перекос и задержка прохождения, глобальная сеть буферов и ловушка стробирования клока логикой.</p><ol><li><a href="/ru/learn/skew-and-insertion/">04 Перекос и задержка прохождения</a></li><li><a href="/ru/learn/global-buffers/">05 Глобальный буфер</a></li><li><a href="/ru/learn/gated-clock-trap/">06 Ловушка стробированного клока</a></li></ol></li><li><h3>Модуль 3 · PLL и MMCM</h3><p>Умножить и поделить один тактовый сигнал в другой, сдвинуть его фазу шагами VCO и какие тактовые сигналы анализатор считает родственными.</p><ol><li><a href="/ru/learn/multiply-and-divide/">07 Умножить и поделить</a></li><li><a href="/ru/learn/phase-and-lock/">08 Фаза и захват</a></li><li class="on" aria-current="page"><a href="/ru/learn/related-or-not/">09 Родственные или чужие</a></li></ol></li><li><h3>Модуль 4 · Сбросы</h3><p>Ассертировать асинхронно, снимать синхронно: три вида сброса, конвейер снятия и дерево, которое растит сброс.</p><ol><li><a href="/ru/learn/sync-or-async/">10 Синхронный или асинхронный</a></li><li><a href="/ru/learn/reset-release/">11 Конвейер снятия</a></li><li><a href="/ru/learn/reset-fanout/">12 Разветвление сброса</a></li></ol></li><li><h3>Модуль 5 · Метастабильность</h3><p>Окно setup-hold, среднее время между отказами в целочисленной арифметике и два триггера, которые это чинят.</p><ol><li><a href="/ru/learn/setup-hold-window/">13 Окно setup-hold</a></li><li><a href="/ru/learn/mtbf/">14 Среднее время между отказами</a></li><li><a href="/ru/learn/two-flop-sync/">15 Двухтриггерный синхронизатор</a></li></ol></li><li><h3>Модуль 6 · Пересечение многих бит</h3><p>Почему двоичная шина рвётся, почему код Грея нет, и рукопожатие, переносящее импульс между мирами.</p><ol><li><a href="/ru/learn/why-binary-fails/">16 Почему двоичный код рвётся</a></li><li><a href="/ru/learn/gray-code/">17 Код Грея</a></li><li><a href="/ru/learn/handshakes-and-pulses/">18 Рукопожатия и импульсы</a></li></ol></li><li><h3>Модуль 7 · Асинхронный FIFO</h3><p>Указатели, флаги и глубина: буфер, переносящий поток между двумя тактовыми сигналами.</p><ol><li><a href="/ru/learn/pointers-and-flags/">19 Указатели и флаги</a></li><li><a href="/ru/learn/depth-sizing/">20 Выбор глубины</a></li><li><a href="/ru/learn/fifo-testbench/">21 Тестбенч FIFO</a></li></ol></li><li><h3>Модуль 8 · Ограничения</h3><p>Строки, сообщающие анализатору, что такое тактовый сигнал, какие пути не проверять и чему должны соответствовать выводы.</p><ol><li><a href="/ru/learn/create-clock/">22 Строка create_clock</a></li><li><a href="/ru/learn/false-path-and-max-delay/">23 False path и max delay</a></li><li><a href="/ru/learn/io-timing/">24 Тайминг выводов</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Отчёт CDC, одно пересечение, захваченное на триггерах, и дифф битстрима, замыкающий курс.</p><ol><li><a href="/ru/learn/cdc-report/">25 Отчёт CDC</a></li><li><a href="/ru/learn/capture-a-crossing/">26 Захватить пересечение</a></li><li><a href="/ru/learn/capstone-diff/">27 Капстоун: поменяй и сравни</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/reset-fanout/index.html b/apps/website/public/ru/learn/reset-fanout/index.html
index e89f1b17a1..1ed7168b72 100644
--- a/apps/website/public/ru/learn/reset-fanout/index.html
+++ b/apps/website/public/ru/learn/reset-fanout/index.html
@@ -294,5 +294,6 @@ <h1>Разветвление сброса</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое тактовый сигнал</h3><p>Один фронт, один мир: всё, что делит фронт тактового сигнала, делит и мир; период и джиттер настоящего фронта и место, где тактовый сигнал входит в плату.</p><ol><li><a href="/ru/learn/one-edge-one-world/">01 Один фронт, один мир</a></li><li><a href="/ru/learn/period-and-jitter/">02 Период, фронт и джиттер</a></li><li><a href="/ru/learn/where-the-clock-enters/">03 Где входит тактовый сигнал</a></li></ol></li><li><h3>Модуль 2 · Деревья тактового сигнала</h3><p>Перекос и задержка прохождения, глобальная сеть буферов и ловушка стробирования клока логикой.</p><ol><li><a href="/ru/learn/skew-and-insertion/">04 Перекос и задержка прохождения</a></li><li><a href="/ru/learn/global-buffers/">05 Глобальный буфер</a></li><li><a href="/ru/learn/gated-clock-trap/">06 Ловушка стробированного клока</a></li></ol></li><li><h3>Модуль 3 · PLL и MMCM</h3><p>Умножить и поделить один тактовый сигнал в другой, сдвинуть его фазу шагами VCO и какие тактовые сигналы анализатор считает родственными.</p><ol><li><a href="/ru/learn/multiply-and-divide/">07 Умножить и поделить</a></li><li><a href="/ru/learn/phase-and-lock/">08 Фаза и захват</a></li><li><a href="/ru/learn/related-or-not/">09 Родственные или чужие</a></li></ol></li><li><h3>Модуль 4 · Сбросы</h3><p>Ассертировать асинхронно, снимать синхронно: три вида сброса, конвейер снятия и дерево, которое растит сброс.</p><ol><li><a href="/ru/learn/sync-or-async/">10 Синхронный или асинхронный</a></li><li><a href="/ru/learn/reset-release/">11 Конвейер снятия</a></li><li class="on" aria-current="page"><a href="/ru/learn/reset-fanout/">12 Разветвление сброса</a></li></ol></li><li><h3>Модуль 5 · Метастабильность</h3><p>Окно setup-hold, среднее время между отказами в целочисленной арифметике и два триггера, которые это чинят.</p><ol><li><a href="/ru/learn/setup-hold-window/">13 Окно setup-hold</a></li><li><a href="/ru/learn/mtbf/">14 Среднее время между отказами</a></li><li><a href="/ru/learn/two-flop-sync/">15 Двухтриггерный синхронизатор</a></li></ol></li><li><h3>Модуль 6 · Пересечение многих бит</h3><p>Почему двоичная шина рвётся, почему код Грея нет, и рукопожатие, переносящее импульс между мирами.</p><ol><li><a href="/ru/learn/why-binary-fails/">16 Почему двоичный код рвётся</a></li><li><a href="/ru/learn/gray-code/">17 Код Грея</a></li><li><a href="/ru/learn/handshakes-and-pulses/">18 Рукопожатия и импульсы</a></li></ol></li><li><h3>Модуль 7 · Асинхронный FIFO</h3><p>Указатели, флаги и глубина: буфер, переносящий поток между двумя тактовыми сигналами.</p><ol><li><a href="/ru/learn/pointers-and-flags/">19 Указатели и флаги</a></li><li><a href="/ru/learn/depth-sizing/">20 Выбор глубины</a></li><li><a href="/ru/learn/fifo-testbench/">21 Тестбенч FIFO</a></li></ol></li><li><h3>Модуль 8 · Ограничения</h3><p>Строки, сообщающие анализатору, что такое тактовый сигнал, какие пути не проверять и чему должны соответствовать выводы.</p><ol><li><a href="/ru/learn/create-clock/">22 Строка create_clock</a></li><li><a href="/ru/learn/false-path-and-max-delay/">23 False path и max delay</a></li><li><a href="/ru/learn/io-timing/">24 Тайминг выводов</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Отчёт CDC, одно пересечение, захваченное на триггерах, и дифф битстрима, замыкающий курс.</p><ol><li><a href="/ru/learn/cdc-report/">25 Отчёт CDC</a></li><li><a href="/ru/learn/capture-a-crossing/">26 Захватить пересечение</a></li><li><a href="/ru/learn/capstone-diff/">27 Капстоун: поменяй и сравни</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/reset-release/index.html b/apps/website/public/ru/learn/reset-release/index.html
index dfd0647bd9..610b31c184 100644
--- a/apps/website/public/ru/learn/reset-release/index.html
+++ b/apps/website/public/ru/learn/reset-release/index.html
@@ -329,5 +329,6 @@ <h1>Конвейер снятия</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое тактовый сигнал</h3><p>Один фронт, один мир: всё, что делит фронт тактового сигнала, делит и мир; период и джиттер настоящего фронта и место, где тактовый сигнал входит в плату.</p><ol><li><a href="/ru/learn/one-edge-one-world/">01 Один фронт, один мир</a></li><li><a href="/ru/learn/period-and-jitter/">02 Период, фронт и джиттер</a></li><li><a href="/ru/learn/where-the-clock-enters/">03 Где входит тактовый сигнал</a></li></ol></li><li><h3>Модуль 2 · Деревья тактового сигнала</h3><p>Перекос и задержка прохождения, глобальная сеть буферов и ловушка стробирования клока логикой.</p><ol><li><a href="/ru/learn/skew-and-insertion/">04 Перекос и задержка прохождения</a></li><li><a href="/ru/learn/global-buffers/">05 Глобальный буфер</a></li><li><a href="/ru/learn/gated-clock-trap/">06 Ловушка стробированного клока</a></li></ol></li><li><h3>Модуль 3 · PLL и MMCM</h3><p>Умножить и поделить один тактовый сигнал в другой, сдвинуть его фазу шагами VCO и какие тактовые сигналы анализатор считает родственными.</p><ol><li><a href="/ru/learn/multiply-and-divide/">07 Умножить и поделить</a></li><li><a href="/ru/learn/phase-and-lock/">08 Фаза и захват</a></li><li><a href="/ru/learn/related-or-not/">09 Родственные или чужие</a></li></ol></li><li><h3>Модуль 4 · Сбросы</h3><p>Ассертировать асинхронно, снимать синхронно: три вида сброса, конвейер снятия и дерево, которое растит сброс.</p><ol><li><a href="/ru/learn/sync-or-async/">10 Синхронный или асинхронный</a></li><li class="on" aria-current="page"><a href="/ru/learn/reset-release/">11 Конвейер снятия</a></li><li><a href="/ru/learn/reset-fanout/">12 Разветвление сброса</a></li></ol></li><li><h3>Модуль 5 · Метастабильность</h3><p>Окно setup-hold, среднее время между отказами в целочисленной арифметике и два триггера, которые это чинят.</p><ol><li><a href="/ru/learn/setup-hold-window/">13 Окно setup-hold</a></li><li><a href="/ru/learn/mtbf/">14 Среднее время между отказами</a></li><li><a href="/ru/learn/two-flop-sync/">15 Двухтриггерный синхронизатор</a></li></ol></li><li><h3>Модуль 6 · Пересечение многих бит</h3><p>Почему двоичная шина рвётся, почему код Грея нет, и рукопожатие, переносящее импульс между мирами.</p><ol><li><a href="/ru/learn/why-binary-fails/">16 Почему двоичный код рвётся</a></li><li><a href="/ru/learn/gray-code/">17 Код Грея</a></li><li><a href="/ru/learn/handshakes-and-pulses/">18 Рукопожатия и импульсы</a></li></ol></li><li><h3>Модуль 7 · Асинхронный FIFO</h3><p>Указатели, флаги и глубина: буфер, переносящий поток между двумя тактовыми сигналами.</p><ol><li><a href="/ru/learn/pointers-and-flags/">19 Указатели и флаги</a></li><li><a href="/ru/learn/depth-sizing/">20 Выбор глубины</a></li><li><a href="/ru/learn/fifo-testbench/">21 Тестбенч FIFO</a></li></ol></li><li><h3>Модуль 8 · Ограничения</h3><p>Строки, сообщающие анализатору, что такое тактовый сигнал, какие пути не проверять и чему должны соответствовать выводы.</p><ol><li><a href="/ru/learn/create-clock/">22 Строка create_clock</a></li><li><a href="/ru/learn/false-path-and-max-delay/">23 False path и max delay</a></li><li><a href="/ru/learn/io-timing/">24 Тайминг выводов</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Отчёт CDC, одно пересечение, захваченное на триггерах, и дифф битстрима, замыкающий курс.</p><ol><li><a href="/ru/learn/cdc-report/">25 Отчёт CDC</a></li><li><a href="/ru/learn/capture-a-crossing/">26 Захватить пересечение</a></li><li><a href="/ru/learn/capstone-diff/">27 Капстоун: поменяй и сравни</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/rgmii-timing/index.html b/apps/website/public/ru/learn/rgmii-timing/index.html
index 40b0a6ae0f..5be62bb460 100644
--- a/apps/website/public/ru/learn/rgmii-timing/index.html
+++ b/apps/website/public/ru/learn/rgmii-timing/index.html
@@ -186,5 +186,6 @@ <h1>Тайминг RGMII</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое шина</h3><p>Зачем вообще существует шина: разговор по проводам, с кадрами и адресами, и кому позволено говорить.</p><ol><li><a href="/ru/learn/wires-that-talk/">01 Провода, которые говорят</a></li><li><a href="/ru/learn/framing-a-conversation/">02 Кадрирование разговора</a></li><li><a href="/ru/learn/master-and-servant/">03 Хозяин и слуга</a></li></ol></li><li><h3>Модуль 2 · Разговор по UART</h3><p>Двухпроводная шина без такта: кадр, делитель, задающий скорость, и статус, который опрашивает драйвер.</p><ol><li><a href="/ru/learn/the-frame/">04 Кадр: старт, данные, стоп</a></li><li><a href="/ru/learn/baud-and-the-divisor/">05 Бод и делитель</a></li><li><a href="/ru/learn/status-and-fifos/">06 Статус и FIFO</a></li></ol></li><li><h3>Модуль 3 · SPI под тактом</h3><p>Разговор под тактом: четыре режима, лестница предделителей и выбор кристалла на каждого подчинённого.</p><ol><li><a href="/ru/learn/modes-cpol-cpha/">07 Режимы: CPOL и CPHA</a></li><li><a href="/ru/learn/the-prescaler-ladder/">08 Лестница предделителей</a></li><li><a href="/ru/learn/chip-select-and-widths/">09 Выбор кристалла и ширины</a></li></ol></li><li><h3>Модуль 4 · Регистровая шина APB</h3><p>Шина регистров: PSEL и PENABLE, строобы и ожидания, и сколько адресных бит стоит количество периферии.</p><ol><li><a href="/ru/learn/psel-and-penable/">10 PSEL и PENABLE</a></li><li><a href="/ru/learn/strobes-and-waits/">11 Строобы и ожидания</a></li><li><a href="/ru/learn/address-decode/">12 Дешифрация адреса</a></li></ol></li><li><h3>Модуль 5 · Пять каналов AXI4</h3><p>Пять каналов: адрес, данные и отклик в обе стороны, lite или full, пакеты и идентификаторы.</p><ol><li><a href="/ru/learn/five-channels/">13 Пять каналов</a></li><li><a href="/ru/learn/lite-or-full/">14 Lite или full</a></li><li><a href="/ru/learn/bursts-and-ids/">15 Пакеты и идентификаторы</a></li></ol></li><li><h3>Модуль 6 · Память</h3><p>Что находится на дальнем конце каждой шины: карты памяти, типы портов и задержка, которую должно покрывать ожидание.</p><ol><li><a href="/ru/learn/memory-maps/">16 Карты памяти</a></li><li><a href="/ru/learn/ports-and-kinds/">17 Порты и типы</a></li><li><a href="/ru/learn/latency/">18 Задержка</a></li></ol></li><li><h3>Модуль 7 · Мосты</h3><p>Почему в дизайне появляется больше одной шины и пакетный мост, переносящий работу между ними.</p><ol><li><a href="/ru/learn/why-bridges/">19 Зачем мосты</a></li><li><a href="/ru/learn/the-packet-bridge/">20 Пакетный мост</a></li><li><a href="/ru/learn/everything-in-one-bench/">21 Всё в одном стенде</a></li></ol></li><li><h3>Модуль 8 · Ethernet: кадры и тайминг</h3><p>Кадры, контрольная сумма кадра, тайминг RGMII и заранее зарегистрированные шаги настоящего запуска.</p><ol><li><a href="/ru/learn/frames-and-the-fcs/">22 Кадры и контрольная сумма</a></li><li class="on" aria-current="page"><a href="/ru/learn/rgmii-timing/">23 Тайминг RGMII</a></li><li><a href="/ru/learn/bring-up-on-the-bench/">24 Запуск на стенде</a></li></ol></li><li><h3>Модуль 9 · Стенд</h3><p>Дисциплина, оберегающая настоящее железо: кто держит IO, как взять и вернуть захват, и что запускается дальше.</p><ol><li><a href="/ru/learn/who-holds-the-io/">25 Кто держит IO</a></li><li><a href="/ru/learn/claim-and-release/">26 Захват и возврат</a></li><li><a href="/ru/learn/capstone-what-runs-next/">27 Капстоун: что запускается дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/ripple-carry/index.html b/apps/website/public/ru/learn/ripple-carry/index.html
index 1fdec44a27..805fc9ccfc 100644
--- a/apps/website/public/ru/learn/ripple-carry/index.html
+++ b/apps/website/public/ru/learn/ripple-carry/index.html
@@ -159,5 +159,6 @@ <h1>Сквозной перенос</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Три значения</h3><p>Почему три, как сбалансированная троичная система пишет любое число без знака и что дают переворот и отсечение тритов.</p><ol><li><a href="/ru/learn/why-three/">01 Почему три</a></li><li><a href="/ru/learn/balanced-ternary/">02 Сбалансированная троичная</a></li><li><a href="/ru/learn/flip-and-cut/">03 Переворот вместо минуса, отсечение вместо округления</a></li></ol></li><li><h3>Модуль 2 · Логика с «не знаю»</h3><p>Трёхзначные вентили Клини, два вентиля без двоичного двойника и сложение как пара таблиц.</p><ol><li><a href="/ru/learn/kleene-logic/">04 Трёхзначная логика</a></li><li><a href="/ru/learn/consensus-and-any/">05 Консенсус и «любой»</a></li><li><a href="/ru/learn/sum-and-carry/">06 Сумма и перенос</a></li></ol></li><li><h3>Модуль 3 · Складываем триты</h3><p>Полусумматор, полный сумматор и перенос, бегущий влево разряд за разрядом.</p><ol><li><a href="/ru/learn/half-adder/">07 Полусумматор</a></li><li><a href="/ru/learn/full-adder/">08 Полный сумматор</a></li><li class="on" aria-current="page"><a href="/ru/learn/ripple-carry/">09 Сквозной перенос</a></li></ol></li><li><h3>Модуль 4 · Умножение без умножителя</h3><p>Копия, ноль или переворот: произведение на один трит, умножение столбиком и MAC, который только складывает.</p><ol><li><a href="/ru/learn/times-a-trit/">10 Умножить на один трит</a></li><li><a href="/ru/learn/long-multiplication/">11 Умножение столбиком</a></li><li><a href="/ru/learn/ternary-mac/">12 Троичный MAC</a></li></ol></li><li><h3>Модуль 5 · Триты в двоичной памяти</h3><p>Два бита на трит, пять тритов на байт и сколько бит нужно слову из 27 тритов.</p><ol><li><a href="/ru/learn/two-bits-per-trit/">13 Два бита на трит</a></li><li><a href="/ru/learn/five-trits-a-byte/">14 Пять тритов в байте</a></li><li><a href="/ru/learn/a-word-of-27/">15 Слово из 27 тритов</a></li></ol></li><li><h3>Модуль 6 · Командное слово TRI-27</h3><p>32-битное слово, которое декодирует эмулятор Trinity: его поля, 47 опкодов и 15-битное непосредственное значение.</p><ol><li><a href="/ru/learn/the-instruction-word/">16 Командное слово</a></li><li><a href="/ru/learn/the-opcode-map/">17 Карта опкодов</a></li><li><a href="/ru/learn/the-immediate/">18 Непосредственное значение</a></li></ol></li><li><h3>Модуль 7 · Троичная машина</h3><p>АЛУ из сумматоров этого курса, переход на три стороны и программа, которую можно пройти по шагам.</p><ol><li><a href="/ru/learn/a-ternary-alu/">19 Троичное АЛУ</a></li><li><a href="/ru/learn/three-way-jump/">20 Переход на три стороны</a></li><li><a href="/ru/learn/run-a-program/">21 Запустить программу</a></li></ol></li><li><h3>Модуль 8 · Три ответа</h3><p>Сравнение по старшему различному триту, поиск числа по третям и сортировка трёхзначными сравнениями.</p><ol><li><a href="/ru/learn/compare-at-the-top/">22 Сравнение сверху</a></li><li><a href="/ru/learn/questions-in-thirds/">23 Вопросы по третям</a></li><li><a href="/ru/learn/sort-by-three/">24 Сортировка трёхзначными сравнениями</a></li></ol></li><li><h3>Модуль 9 · Троичные нейроны</h3><p>Веса -1, 0, +1: один нейрон, детектор и маленький слой, и нигде нет умножителя.</p><ol><li><a href="/ru/learn/one-ternary-neuron/">25 Один троичный нейрон</a></li><li><a href="/ru/learn/a-bar-detector/">26 Детектор полосы</a></li><li><a href="/ru/learn/a-tiny-layer/">27 Маленький слой</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/rotating-by-shifts/index.html b/apps/website/public/ru/learn/rotating-by-shifts/index.html
index 143b9b2fe6..988eda11a8 100644
--- a/apps/website/public/ru/learn/rotating-by-shifts/index.html
+++ b/apps/website/public/ru/learn/rotating-by-shifts/index.html
@@ -439,5 +439,6 @@ <h1>Поворот сдвигами</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Сложение</h3><p>Почему сложение медленное: перенос, который идёт вверх по слову, выделенная цепь в каждом слайсе, префиксные сети и деревья сохранения переноса.</p><ol><li><a href="/ru/learn/the-carry-chain/">01 Цепь переноса</a></li><li><a href="/ru/learn/prefix-adders/">02 Взгляд вперёд: префиксные сумматоры</a></li><li><a href="/ru/learn/carry-save/">03 Три числа сразу</a></li></ol></li><li><h3>Модуль 2 · Умножение</h3><p>Произведение — сумма сдвинутых копий; перекодирование Бута вдвое сокращает их, а слайс DSP48E1 делает остальное одним блоком.</p><ol><li><a href="/ru/learn/shift-and-add/">04 Сдвиг и сложение</a></li><li><a href="/ru/learn/booths-trick/">05 Приём Бута</a></li><li><a href="/ru/learn/the-dsp48e1-slice/">06 Слайс DSP48E1</a></li></ol></li><li><h3>Модуль 3 · Фиксированная точка</h3><p>Где стоит двоичная точка, что округление делает со значением и с его средним и что даёт каждый бит квантователя.</p><ol><li><a href="/ru/learn/q-formats/">07 Форматы Q</a></li><li><a href="/ru/learn/rounding-and-saturation/">08 Округление и насыщение</a></li><li><a href="/ru/learn/quantization-noise/">09 Шум квантования</a></li></ol></li><li><h3>Модуль 4 · Функции в железе</h3><p>Синус, косинус, угол и длина из сдвигов и сложений — и когда таблица оказывается выгоднее.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/rotating-by-shifts/">10 Поворот сдвигами</a></li><li><a href="/ru/learn/angle-and-magnitude/">11 Угол и модуль</a></li><li><a href="/ru/learn/tables-and-interpolation/">12 Таблицы и интерполяция</a></li></ol></li><li><h3>Модуль 5 · Сигналы и дискретизация</h3><p>Что дискретизация делает с частотой, как генератор строится из сумматора и что измеряет бин ДПФ.</p><ol><li><a href="/ru/learn/sampling-and-aliasing/">13 Дискретизация и наложение спектров</a></li><li><a href="/ru/learn/the-nco/">14 Генератор с числовым управлением</a></li><li><a href="/ru/learn/bins-of-a-spectrum/">15 Бины спектра</a></li></ol></li><li><h3>Модуль 6 · Фильтры FIR</h3><p>Скользящее среднее, расчёт окном sinc с целыми коэффициентами и раскладка отводов по слайсам DSP.</p><ol><li><a href="/ru/learn/the-moving-average/">16 Скользящее среднее</a></li><li><a href="/ru/learn/designing-taps/">17 Расчёт коэффициентов</a></li><li><a href="/ru/learn/folding-taps-onto-slices/">18 Раскладка отводов по слайсам</a></li></ol></li><li><h3>Модуль 7 · Многоскоростная обработка</h3><p>Понижение частоты дискретизации без заворота шума: децимация, фильтр CIC и полифазная форма.</p><ol><li><a href="/ru/learn/lowering-the-rate/">19 Понижение частоты</a></li><li><a href="/ru/learn/the-cic-filter/">20 Фильтр CIC</a></li><li><a href="/ru/learn/polyphase/">21 Полифазная форма</a></li></ol></li><li><h3>Модуль 8 · БПФ</h3><p>N log N вместо N^2: бабочки, бит-реверсный порядок входа и биты, которые добавляет каждый этап.</p><ol><li><a href="/ru/learn/the-butterfly/">22 Бабочка</a></li><li><a href="/ru/learn/twiddles-and-bit-reversal/">23 Поворачивающие множители и бит-реверс</a></li><li><a href="/ru/learn/growth-through-the-stages/">24 Рост через этапы</a></li></ol></li><li><h3>Модуль 9 · На стенде</h3><p>От арифметики к плате: коррелятор из спеки модема, бюджет слайсов и тайминга и итоговый фильтр.</p><ol><li><a href="/ru/learn/a-correlator-finds-the-frame/">25 Коррелятор находит кадр</a></li><li><a href="/ru/learn/does-it-fit-and-close-timing/">26 Помещается ли и сходится ли тайминг?</a></li><li><a href="/ru/learn/capstone-a-filter-on-the-board/">27 Итог: фильтр на плате</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/rounding-and-saturation/index.html b/apps/website/public/ru/learn/rounding-and-saturation/index.html
index 3588206a6a..080f96b013 100644
--- a/apps/website/public/ru/learn/rounding-and-saturation/index.html
+++ b/apps/website/public/ru/learn/rounding-and-saturation/index.html
@@ -346,5 +346,6 @@ <h1>Округление и насыщение</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Сложение</h3><p>Почему сложение медленное: перенос, который идёт вверх по слову, выделенная цепь в каждом слайсе, префиксные сети и деревья сохранения переноса.</p><ol><li><a href="/ru/learn/the-carry-chain/">01 Цепь переноса</a></li><li><a href="/ru/learn/prefix-adders/">02 Взгляд вперёд: префиксные сумматоры</a></li><li><a href="/ru/learn/carry-save/">03 Три числа сразу</a></li></ol></li><li><h3>Модуль 2 · Умножение</h3><p>Произведение — сумма сдвинутых копий; перекодирование Бута вдвое сокращает их, а слайс DSP48E1 делает остальное одним блоком.</p><ol><li><a href="/ru/learn/shift-and-add/">04 Сдвиг и сложение</a></li><li><a href="/ru/learn/booths-trick/">05 Приём Бута</a></li><li><a href="/ru/learn/the-dsp48e1-slice/">06 Слайс DSP48E1</a></li></ol></li><li><h3>Модуль 3 · Фиксированная точка</h3><p>Где стоит двоичная точка, что округление делает со значением и с его средним и что даёт каждый бит квантователя.</p><ol><li><a href="/ru/learn/q-formats/">07 Форматы Q</a></li><li class="on" aria-current="page"><a href="/ru/learn/rounding-and-saturation/">08 Округление и насыщение</a></li><li><a href="/ru/learn/quantization-noise/">09 Шум квантования</a></li></ol></li><li><h3>Модуль 4 · Функции в железе</h3><p>Синус, косинус, угол и длина из сдвигов и сложений — и когда таблица оказывается выгоднее.</p><ol><li><a href="/ru/learn/rotating-by-shifts/">10 Поворот сдвигами</a></li><li><a href="/ru/learn/angle-and-magnitude/">11 Угол и модуль</a></li><li><a href="/ru/learn/tables-and-interpolation/">12 Таблицы и интерполяция</a></li></ol></li><li><h3>Модуль 5 · Сигналы и дискретизация</h3><p>Что дискретизация делает с частотой, как генератор строится из сумматора и что измеряет бин ДПФ.</p><ol><li><a href="/ru/learn/sampling-and-aliasing/">13 Дискретизация и наложение спектров</a></li><li><a href="/ru/learn/the-nco/">14 Генератор с числовым управлением</a></li><li><a href="/ru/learn/bins-of-a-spectrum/">15 Бины спектра</a></li></ol></li><li><h3>Модуль 6 · Фильтры FIR</h3><p>Скользящее среднее, расчёт окном sinc с целыми коэффициентами и раскладка отводов по слайсам DSP.</p><ol><li><a href="/ru/learn/the-moving-average/">16 Скользящее среднее</a></li><li><a href="/ru/learn/designing-taps/">17 Расчёт коэффициентов</a></li><li><a href="/ru/learn/folding-taps-onto-slices/">18 Раскладка отводов по слайсам</a></li></ol></li><li><h3>Модуль 7 · Многоскоростная обработка</h3><p>Понижение частоты дискретизации без заворота шума: децимация, фильтр CIC и полифазная форма.</p><ol><li><a href="/ru/learn/lowering-the-rate/">19 Понижение частоты</a></li><li><a href="/ru/learn/the-cic-filter/">20 Фильтр CIC</a></li><li><a href="/ru/learn/polyphase/">21 Полифазная форма</a></li></ol></li><li><h3>Модуль 8 · БПФ</h3><p>N log N вместо N^2: бабочки, бит-реверсный порядок входа и биты, которые добавляет каждый этап.</p><ol><li><a href="/ru/learn/the-butterfly/">22 Бабочка</a></li><li><a href="/ru/learn/twiddles-and-bit-reversal/">23 Поворачивающие множители и бит-реверс</a></li><li><a href="/ru/learn/growth-through-the-stages/">24 Рост через этапы</a></li></ol></li><li><h3>Модуль 9 · На стенде</h3><p>От арифметики к плате: коррелятор из спеки модема, бюджет слайсов и тайминга и итоговый фильтр.</p><ol><li><a href="/ru/learn/a-correlator-finds-the-frame/">25 Коррелятор находит кадр</a></li><li><a href="/ru/learn/does-it-fit-and-close-timing/">26 Помещается ли и сходится ли тайминг?</a></li><li><a href="/ru/learn/capstone-a-filter-on-the-board/">27 Итог: фильтр на плате</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/run-a-program/index.html b/apps/website/public/ru/learn/run-a-program/index.html
index 4299b3a2fe..6e1a973c47 100644
--- a/apps/website/public/ru/learn/run-a-program/index.html
+++ b/apps/website/public/ru/learn/run-a-program/index.html
@@ -278,5 +278,6 @@ <h1>Запустить программу</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Три значения</h3><p>Почему три, как сбалансированная троичная система пишет любое число без знака и что дают переворот и отсечение тритов.</p><ol><li><a href="/ru/learn/why-three/">01 Почему три</a></li><li><a href="/ru/learn/balanced-ternary/">02 Сбалансированная троичная</a></li><li><a href="/ru/learn/flip-and-cut/">03 Переворот вместо минуса, отсечение вместо округления</a></li></ol></li><li><h3>Модуль 2 · Логика с «не знаю»</h3><p>Трёхзначные вентили Клини, два вентиля без двоичного двойника и сложение как пара таблиц.</p><ol><li><a href="/ru/learn/kleene-logic/">04 Трёхзначная логика</a></li><li><a href="/ru/learn/consensus-and-any/">05 Консенсус и «любой»</a></li><li><a href="/ru/learn/sum-and-carry/">06 Сумма и перенос</a></li></ol></li><li><h3>Модуль 3 · Складываем триты</h3><p>Полусумматор, полный сумматор и перенос, бегущий влево разряд за разрядом.</p><ol><li><a href="/ru/learn/half-adder/">07 Полусумматор</a></li><li><a href="/ru/learn/full-adder/">08 Полный сумматор</a></li><li><a href="/ru/learn/ripple-carry/">09 Сквозной перенос</a></li></ol></li><li><h3>Модуль 4 · Умножение без умножителя</h3><p>Копия, ноль или переворот: произведение на один трит, умножение столбиком и MAC, который только складывает.</p><ol><li><a href="/ru/learn/times-a-trit/">10 Умножить на один трит</a></li><li><a href="/ru/learn/long-multiplication/">11 Умножение столбиком</a></li><li><a href="/ru/learn/ternary-mac/">12 Троичный MAC</a></li></ol></li><li><h3>Модуль 5 · Триты в двоичной памяти</h3><p>Два бита на трит, пять тритов на байт и сколько бит нужно слову из 27 тритов.</p><ol><li><a href="/ru/learn/two-bits-per-trit/">13 Два бита на трит</a></li><li><a href="/ru/learn/five-trits-a-byte/">14 Пять тритов в байте</a></li><li><a href="/ru/learn/a-word-of-27/">15 Слово из 27 тритов</a></li></ol></li><li><h3>Модуль 6 · Командное слово TRI-27</h3><p>32-битное слово, которое декодирует эмулятор Trinity: его поля, 47 опкодов и 15-битное непосредственное значение.</p><ol><li><a href="/ru/learn/the-instruction-word/">16 Командное слово</a></li><li><a href="/ru/learn/the-opcode-map/">17 Карта опкодов</a></li><li><a href="/ru/learn/the-immediate/">18 Непосредственное значение</a></li></ol></li><li><h3>Модуль 7 · Троичная машина</h3><p>АЛУ из сумматоров этого курса, переход на три стороны и программа, которую можно пройти по шагам.</p><ol><li><a href="/ru/learn/a-ternary-alu/">19 Троичное АЛУ</a></li><li><a href="/ru/learn/three-way-jump/">20 Переход на три стороны</a></li><li class="on" aria-current="page"><a href="/ru/learn/run-a-program/">21 Запустить программу</a></li></ol></li><li><h3>Модуль 8 · Три ответа</h3><p>Сравнение по старшему различному триту, поиск числа по третям и сортировка трёхзначными сравнениями.</p><ol><li><a href="/ru/learn/compare-at-the-top/">22 Сравнение сверху</a></li><li><a href="/ru/learn/questions-in-thirds/">23 Вопросы по третям</a></li><li><a href="/ru/learn/sort-by-three/">24 Сортировка трёхзначными сравнениями</a></li></ol></li><li><h3>Модуль 9 · Троичные нейроны</h3><p>Веса -1, 0, +1: один нейрон, детектор и маленький слой, и нигде нет умножителя.</p><ol><li><a href="/ru/learn/one-ternary-neuron/">25 Один троичный нейрон</a></li><li><a href="/ru/learn/a-bar-detector/">26 Детектор полосы</a></li><li><a href="/ru/learn/a-tiny-layer/">27 Маленький слой</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/same-bits-on-the-board/index.html b/apps/website/public/ru/learn/same-bits-on-the-board/index.html
index 6c01ccd674..e4929a0cf5 100644
--- a/apps/website/public/ru/learn/same-bits-on-the-board/index.html
+++ b/apps/website/public/ru/learn/same-bits-on-the-board/index.html
@@ -110,5 +110,6 @@ <h1>Те же биты — на плате</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Чип</h3><p>Что такое FPGA, какой чип у нас и как его выводы встречаются с платой.</p><ol><li><a href="/ru/learn/fpga-is-a-grid/">01 FPGA — это сетка, которую вы настраиваете</a></li><li><a href="/ru/learn/which-chip/">02 Какой это чип?</a></li><li><a href="/ru/learn/pins-and-package/">03 Выводы, шарики и корпус</a></li></ol></li><li><h3>Модуль 2 · Числа в железе</h3><p>Биты, триты и форматы чисел, и сколько стоит арифметика в логике.</p><ol><li><a href="/ru/learn/bits-and-trits/">04 Биты, триты и GF16</a></li><li><a href="/ru/learn/what-arithmetic-costs/">05 Сколько стоит арифметика</a></li><li><a href="/ru/learn/how-much-fits/">06 Сколько помещается на чип</a></li></ol></li><li><h3>Модуль 3 · Ваша программа на t27</h3><p>Напишите спеку, проверьте её и увидьте, почему «компилируется» не значит «верно».</p><ol><li><a href="/ru/learn/hello-t27/">07 Привет, t27</a></li><li><a href="/ru/learn/tests-are-the-spec/">08 Тесты — часть спеки</a></li><li><a href="/ru/learn/compiles-is-not-correct/">09 Компилируется — не значит верно</a></li></ol></li><li><h3>Модуль 4 · Внутри t27c</h3><p>Как компилятор читает спеку и что он пишет, включая нативный t27b.</p><ol><li><a href="/ru/learn/compiler-stages/">10 Стадии t27c</a></li><li><a href="/ru/learn/t27c-on-a-real-machine/">11 t27c на настоящей машине</a></li><li><a href="/ru/learn/seven-backends-and-t27b/">12 Семь бэкендов и t27b</a></li></ol></li><li><h3>Модуль 5 · От спеки к железу</h3><p>Verilog, который пишет t27c, ячейки, в которые он превращается, и что делает одна LUT.</p><ol><li><a href="/ru/learn/spec-to-verilog/">13 От спеки к Verilog</a></li><li><a href="/ru/learn/count-the-cells/">14 Посчитайте ячейки</a></li><li><a href="/ru/learn/one-lut-up-close/">15 Одна LUT вблизи</a></li></ol></li><li><h3>Модуль 6 · Читаем синтез</h3><p>Что yosys сообщает о вашем дизайне и какие предупреждения важны.</p><ol><li><a href="/ru/learn/synthesis-warnings/">16 Важные предупреждения синтеза</a></li><li><a href="/ru/learn/state-machines/">17 Конечные автоматы</a></li><li><a href="/ru/learn/where-the-luts-go/">18 Куда уходят LUT</a></li></ol></li><li><h3>Модуль 7 · Размещение, трассировка, тайминг</h3><p>Где ячейки оказываются на кристалле и успевает ли тактовая частота.</p><ol><li><a href="/ru/learn/placement/">19 Размещение: где оказываются ячейки</a></li><li><a href="/ru/learn/logic-depth/">20 Глубина логики</a></li><li><a href="/ru/learn/timing-slack/">21 Тайминг и запас</a></li></ol></li><li><h3>Модуль 8 · Битстрим</h3><p>Как оттрассированный дизайн становится битами, которые загружает чип, без Vivado.</p><ol><li><a href="/ru/learn/fasm-to-frames/">22 От FASM к фреймам</a></li><li><a href="/ru/learn/a-build-receipt/">23 Сборка по пунктам</a></li><li><a href="/ru/learn/read-a-bitstream/">24 Читаем битстрим</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Спросите чип, кто он, загрузите биты и проверьте их.</p><ol><li><a href="/ru/learn/jtag-asks-the-chip/">25 JTAG спрашивает чип</a></li><li class="on" aria-current="page"><a href="/ru/learn/same-bits-on-the-board/">26 Те же биты — на плате</a></li><li><a href="/ru/learn/capstone/">27 Итог: переугадайте yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/same-neuron-new-weights/index.html b/apps/website/public/ru/learn/same-neuron-new-weights/index.html
index f4f70d79cd..86ccdb8acf 100644
--- a/apps/website/public/ru/learn/same-neuron-new-weights/index.html
+++ b/apps/website/public/ru/learn/same-neuron-new-weights/index.html
@@ -135,5 +135,6 @@ <h1>Тот же нейрон, другие веса</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Лаборатория: наши исследования</h3><p>Свой формат чисел, честная таблица результатов и таблицы модели, перемноженные на плате.</p><ol><li><a href="/ru/learn/t27-number-format/">01 T27: наш собственный формат чисел</a></li><li><a href="/ru/learn/an-honest-scoreboard/">02 Честная таблица результатов</a></li><li><a href="/ru/learn/a-model-on-the-board/">03 Таблицы модели на плате</a></li></ol></li><li><h3>Модуль 2 · ИИ-числа: блок MX</h3><p>Как ИИ-чипы хранят веса в нескольких битах: один общий масштаб на блок, сам байт масштаба и что один выброс делает с соседями.</p><ol><li><a href="/ru/learn/one-scale-per-block/">04 Один масштаб на блок весов</a></li><li><a href="/ru/learn/scale-byte-native/">05 Байт масштаба на настоящей машине</a></li><li><a href="/ru/learn/one-outlier/">06 Один выброс, много нулей</a></li></ol></li><li><h3>Модуль 3 · Тернарные веса</h3><p>Веса, которые бывают только минус масштаб, ноль или плюс масштаб, пять правил, которые должен пройти тернарный алфавит, и прогон тестов, который ничего не проверил.</p><ol><li><a href="/ru/learn/phi-scaled-ternary/">07 Веса минус фи, ноль и плюс фи</a></li><li><a href="/ru/learn/a-pass-that-checked-nothing/">08 Прогон, который ничего не проверил</a></li><li><a href="/ru/learn/five-rules-for-weights/">09 Пять правил для алфавита весов</a></li></ol></li><li><h3>Модуль 4 · Тернарный формат Ternary Network Float</h3><p>Правило, которое компилятор проверяет до запуска любого теста, и 17-битное число с плавающей точкой, порядок которого — четыре сбалансированных трита.</p><ol><li><a href="/ru/learn/blocked-is-not-failed/">10 Заблокирован — не значит провален</a></li><li><a href="/ru/learn/four-trit-exponent/">11 Число, у которого порядок — четыре трита</a></li><li><a href="/ru/learn/negate-twice/">12 Дважды сменить знак — получить число обратно</a></li></ol></li><li><h3>Модуль 5 · Арифметика чисел со знаком</h3><p>Умножить два числа со знаком, сложить их, когда знаки разные, и сделать то и другое сразу в умножении с накоплением.</p><ol><li><a href="/ru/learn/a-sign-is-one-bit/">13 Знак — это один бит</a></li><li><a href="/ru/learn/opposite-signs-subtract/">14 Разные знаки — вычитание</a></li><li><a href="/ru/learn/multiply-then-add/">15 Умножить, потом сложить</a></li></ol></li><li><h3>Модуль 6 · Части нейрона</h3><p>ReLU с изломом в нуле, степень двойки для softmax и argmax, который называет ответ.</p><ol><li><a href="/ru/learn/a-bend-at-zero/">16 Излом в нуле</a></li><li><a href="/ru/learn/two-to-the-x/">17 Два в степени x</a></li><li><a href="/ru/learn/pick-the-largest/">18 Выбрать наибольшее</a></li></ol></li><li><h3>Модуль 7 · Обучение на ошибке</h3><p>Потеря, которая оценивает неверную догадку в битах, один шаг, который сдвигает вес против градиента, и скрытый слой, который нужен XOR.</p><ol><li><a href="/ru/learn/a-loss-in-bits/">19 Потеря в битах</a></li><li><a href="/ru/learn/one-step-downhill/">20 Один шаг вниз по склону</a></li><li><a href="/ru/learn/xor-needs-a-bend/">21 XOR нужен излом</a></li></ol></li><li><h3>Модуль 8 · BitNet: тернарные сети</h3><p>Порог, который сжимает сумму обратно до трёх значений, один нейрон, который становится другой функцией при смене весов, и нейрон, который читает входы по 27 тритов за раз.</p><ol><li><a href="/ru/learn/back-to-three-values/">22 Обратно в три значения</a></li><li class="on" aria-current="page"><a href="/ru/learn/same-neuron-new-weights/">23 Тот же нейрон, другие веса</a></li><li><a href="/ru/learn/a-neuron-in-chunks/">24 Нейрон по кускам</a></li></ol></li><li><h3>Модуль 9 · Тернарный MAC как чип</h3><p>Скалярное произведение 27 тритов как провода без регистра, та же сумма, которую регистр копит на каждом такте, и небольшая целая сеть в конце курса.</p><ol><li><a href="/ru/learn/a-dot-product-in-wires/">25 Скалярное произведение из проводов</a></li><li><a href="/ru/learn/add-it-up-every-clock/">26 Складывать на каждом такте</a></li><li><a href="/ru/learn/a-whole-network/">27 Целая сеть</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/sampling-and-aliasing/index.html b/apps/website/public/ru/learn/sampling-and-aliasing/index.html
index 0c3f87dfcf..5b05d88cf6 100644
--- a/apps/website/public/ru/learn/sampling-and-aliasing/index.html
+++ b/apps/website/public/ru/learn/sampling-and-aliasing/index.html
@@ -1122,5 +1122,6 @@ <h1>Дискретизация и наложение спектров</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Сложение</h3><p>Почему сложение медленное: перенос, который идёт вверх по слову, выделенная цепь в каждом слайсе, префиксные сети и деревья сохранения переноса.</p><ol><li><a href="/ru/learn/the-carry-chain/">01 Цепь переноса</a></li><li><a href="/ru/learn/prefix-adders/">02 Взгляд вперёд: префиксные сумматоры</a></li><li><a href="/ru/learn/carry-save/">03 Три числа сразу</a></li></ol></li><li><h3>Модуль 2 · Умножение</h3><p>Произведение — сумма сдвинутых копий; перекодирование Бута вдвое сокращает их, а слайс DSP48E1 делает остальное одним блоком.</p><ol><li><a href="/ru/learn/shift-and-add/">04 Сдвиг и сложение</a></li><li><a href="/ru/learn/booths-trick/">05 Приём Бута</a></li><li><a href="/ru/learn/the-dsp48e1-slice/">06 Слайс DSP48E1</a></li></ol></li><li><h3>Модуль 3 · Фиксированная точка</h3><p>Где стоит двоичная точка, что округление делает со значением и с его средним и что даёт каждый бит квантователя.</p><ol><li><a href="/ru/learn/q-formats/">07 Форматы Q</a></li><li><a href="/ru/learn/rounding-and-saturation/">08 Округление и насыщение</a></li><li><a href="/ru/learn/quantization-noise/">09 Шум квантования</a></li></ol></li><li><h3>Модуль 4 · Функции в железе</h3><p>Синус, косинус, угол и длина из сдвигов и сложений — и когда таблица оказывается выгоднее.</p><ol><li><a href="/ru/learn/rotating-by-shifts/">10 Поворот сдвигами</a></li><li><a href="/ru/learn/angle-and-magnitude/">11 Угол и модуль</a></li><li><a href="/ru/learn/tables-and-interpolation/">12 Таблицы и интерполяция</a></li></ol></li><li><h3>Модуль 5 · Сигналы и дискретизация</h3><p>Что дискретизация делает с частотой, как генератор строится из сумматора и что измеряет бин ДПФ.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/sampling-and-aliasing/">13 Дискретизация и наложение спектров</a></li><li><a href="/ru/learn/the-nco/">14 Генератор с числовым управлением</a></li><li><a href="/ru/learn/bins-of-a-spectrum/">15 Бины спектра</a></li></ol></li><li><h3>Модуль 6 · Фильтры FIR</h3><p>Скользящее среднее, расчёт окном sinc с целыми коэффициентами и раскладка отводов по слайсам DSP.</p><ol><li><a href="/ru/learn/the-moving-average/">16 Скользящее среднее</a></li><li><a href="/ru/learn/designing-taps/">17 Расчёт коэффициентов</a></li><li><a href="/ru/learn/folding-taps-onto-slices/">18 Раскладка отводов по слайсам</a></li></ol></li><li><h3>Модуль 7 · Многоскоростная обработка</h3><p>Понижение частоты дискретизации без заворота шума: децимация, фильтр CIC и полифазная форма.</p><ol><li><a href="/ru/learn/lowering-the-rate/">19 Понижение частоты</a></li><li><a href="/ru/learn/the-cic-filter/">20 Фильтр CIC</a></li><li><a href="/ru/learn/polyphase/">21 Полифазная форма</a></li></ol></li><li><h3>Модуль 8 · БПФ</h3><p>N log N вместо N^2: бабочки, бит-реверсный порядок входа и биты, которые добавляет каждый этап.</p><ol><li><a href="/ru/learn/the-butterfly/">22 Бабочка</a></li><li><a href="/ru/learn/twiddles-and-bit-reversal/">23 Поворачивающие множители и бит-реверс</a></li><li><a href="/ru/learn/growth-through-the-stages/">24 Рост через этапы</a></li></ol></li><li><h3>Модуль 9 · На стенде</h3><p>От арифметики к плате: коррелятор из спеки модема, бюджет слайсов и тайминга и итоговый фильтр.</p><ol><li><a href="/ru/learn/a-correlator-finds-the-frame/">25 Коррелятор находит кадр</a></li><li><a href="/ru/learn/does-it-fit-and-close-timing/">26 Помещается ли и сходится ли тайминг?</a></li><li><a href="/ru/learn/capstone-a-filter-on-the-board/">27 Итог: фильтр на плате</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/scale-byte-native/index.html b/apps/website/public/ru/learn/scale-byte-native/index.html
index ce8e1ed6ae..e1a27e4353 100644
--- a/apps/website/public/ru/learn/scale-byte-native/index.html
+++ b/apps/website/public/ru/learn/scale-byte-native/index.html
@@ -336,5 +336,6 @@ <h1>Байт масштаба на настоящей машине</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Лаборатория: наши исследования</h3><p>Свой формат чисел, честная таблица результатов и таблицы модели, перемноженные на плате.</p><ol><li><a href="/ru/learn/t27-number-format/">01 T27: наш собственный формат чисел</a></li><li><a href="/ru/learn/an-honest-scoreboard/">02 Честная таблица результатов</a></li><li><a href="/ru/learn/a-model-on-the-board/">03 Таблицы модели на плате</a></li></ol></li><li><h3>Модуль 2 · ИИ-числа: блок MX</h3><p>Как ИИ-чипы хранят веса в нескольких битах: один общий масштаб на блок, сам байт масштаба и что один выброс делает с соседями.</p><ol><li><a href="/ru/learn/one-scale-per-block/">04 Один масштаб на блок весов</a></li><li class="on" aria-current="page"><a href="/ru/learn/scale-byte-native/">05 Байт масштаба на настоящей машине</a></li><li><a href="/ru/learn/one-outlier/">06 Один выброс, много нулей</a></li></ol></li><li><h3>Модуль 3 · Тернарные веса</h3><p>Веса, которые бывают только минус масштаб, ноль или плюс масштаб, пять правил, которые должен пройти тернарный алфавит, и прогон тестов, который ничего не проверил.</p><ol><li><a href="/ru/learn/phi-scaled-ternary/">07 Веса минус фи, ноль и плюс фи</a></li><li><a href="/ru/learn/a-pass-that-checked-nothing/">08 Прогон, который ничего не проверил</a></li><li><a href="/ru/learn/five-rules-for-weights/">09 Пять правил для алфавита весов</a></li></ol></li><li><h3>Модуль 4 · Тернарный формат Ternary Network Float</h3><p>Правило, которое компилятор проверяет до запуска любого теста, и 17-битное число с плавающей точкой, порядок которого — четыре сбалансированных трита.</p><ol><li><a href="/ru/learn/blocked-is-not-failed/">10 Заблокирован — не значит провален</a></li><li><a href="/ru/learn/four-trit-exponent/">11 Число, у которого порядок — четыре трита</a></li><li><a href="/ru/learn/negate-twice/">12 Дважды сменить знак — получить число обратно</a></li></ol></li><li><h3>Модуль 5 · Арифметика чисел со знаком</h3><p>Умножить два числа со знаком, сложить их, когда знаки разные, и сделать то и другое сразу в умножении с накоплением.</p><ol><li><a href="/ru/learn/a-sign-is-one-bit/">13 Знак — это один бит</a></li><li><a href="/ru/learn/opposite-signs-subtract/">14 Разные знаки — вычитание</a></li><li><a href="/ru/learn/multiply-then-add/">15 Умножить, потом сложить</a></li></ol></li><li><h3>Модуль 6 · Части нейрона</h3><p>ReLU с изломом в нуле, степень двойки для softmax и argmax, который называет ответ.</p><ol><li><a href="/ru/learn/a-bend-at-zero/">16 Излом в нуле</a></li><li><a href="/ru/learn/two-to-the-x/">17 Два в степени x</a></li><li><a href="/ru/learn/pick-the-largest/">18 Выбрать наибольшее</a></li></ol></li><li><h3>Модуль 7 · Обучение на ошибке</h3><p>Потеря, которая оценивает неверную догадку в битах, один шаг, который сдвигает вес против градиента, и скрытый слой, который нужен XOR.</p><ol><li><a href="/ru/learn/a-loss-in-bits/">19 Потеря в битах</a></li><li><a href="/ru/learn/one-step-downhill/">20 Один шаг вниз по склону</a></li><li><a href="/ru/learn/xor-needs-a-bend/">21 XOR нужен излом</a></li></ol></li><li><h3>Модуль 8 · BitNet: тернарные сети</h3><p>Порог, который сжимает сумму обратно до трёх значений, один нейрон, который становится другой функцией при смене весов, и нейрон, который читает входы по 27 тритов за раз.</p><ol><li><a href="/ru/learn/back-to-three-values/">22 Обратно в три значения</a></li><li><a href="/ru/learn/same-neuron-new-weights/">23 Тот же нейрон, другие веса</a></li><li><a href="/ru/learn/a-neuron-in-chunks/">24 Нейрон по кускам</a></li></ol></li><li><h3>Модуль 9 · Тернарный MAC как чип</h3><p>Скалярное произведение 27 тритов как провода без регистра, та же сумма, которую регистр копит на каждом такте, и небольшая целая сеть в конце курса.</p><ol><li><a href="/ru/learn/a-dot-product-in-wires/">25 Скалярное произведение из проводов</a></li><li><a href="/ru/learn/add-it-up-every-clock/">26 Складывать на каждом такте</a></li><li><a href="/ru/learn/a-whole-network/">27 Целая сеть</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/self-checking/index.html b/apps/website/public/ru/learn/self-checking/index.html
index a71b2e8573..91decc099b 100644
--- a/apps/website/public/ru/learn/self-checking/index.html
+++ b/apps/website/public/ru/learn/self-checking/index.html
@@ -209,5 +209,6 @@ <h1>Самопроверка вместо всматривания</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Зачем проверять</h3><p>Дизайны, которые компилируются и ошибаются; модель, которая выносит вердикт; план, записанный до кода.</p><ol><li><a href="/ru/learn/bugs-that-compile/">01 Баги, которые компилируются</a></li><li><a href="/ru/learn/golden-model/">02 Золотая модель</a></li><li><a href="/ru/learn/plan-before-code/">03 Сначала план</a></li></ol></li><li><h3>Модуль 2 · Тестбенчи</h3><p>Стимулы, проверки и вердикт, записанные как один spec рядом с дизайном, который они судят.</p><ol><li><a href="/ru/learn/anatomy-of-a-tb/">04 Анатомия тестбенча</a></li><li class="on" aria-current="page"><a href="/ru/learn/self-checking/">05 Самопроверка вместо всматривания</a></li><li><a href="/ru/learn/directed-vs-random/">06 Направленные против случайных</a></li></ol></li><li><h3>Модуль 3 · Временные диаграммы</h3><p>Трасса каждого сигнала, прочитанная так, как её читает инженер по железу, и два прогона, сравнённые между собой.</p><ol><li><a href="/ru/learn/vcd-format/">07 Формат VCD</a></li><li><a href="/ru/learn/debug-with-waves/">08 Отладка по волнам</a></li><li><a href="/ru/learn/compare-traces/">09 Сравнение двух трасс</a></li></ol></li><li><h3>Модуль 4 · Векторы соответствия</h3><p>Случаи с ответом, записанным рядом, там, откуда их достанет компилятор.</p><ol><li><a href="/ru/learn/what-a-vector-is/">10 Что такое вектор</a></li><li><a href="/ru/learn/golden-ruler-vectors/">11 Векторы как эталонная линейка</a></li><li><a href="/ru/learn/end-to-end-scenarios/">12 Сквозные сценарии</a></li></ol></li><li><h3>Модуль 5 · Косимуляция</h3><p>Spec, симулятор и плата сходятся в одном ответе на стенде Artix-7 XC7A200T, и что делать, когда не сходятся.</p><ol><li><a href="/ru/learn/three-way-agreement/">13 Тройное согласие</a></li><li><a href="/ru/learn/simulator-internals/">14 Внутри симулятора</a></li><li><a href="/ru/learn/when-they-disagree/">15 Когда они не сходятся</a></li></ol></li><li><h3>Модуль 6 · Покрытие</h3><p>Чего коснулись тесты: строки, переключения, состояния — и что прячет это число.</p><ol><li><a href="/ru/learn/line-and-toggle/">16 Строки и переключения</a></li><li><a href="/ru/learn/fsm-coverage/">17 Покрытие автоматов</a></li><li><a href="/ru/learn/coverage-lies/">18 Покрытие лжёт</a></li></ol></li><li><h3>Модуль 7 · Формальные методы</h3><p>Ассерты, верные каждый такт; ограниченный поиск контрпримера; и почему доказательству нужна индукция.</p><ol><li><a href="/ru/learn/assertions/">19 Ассерты</a></li><li><a href="/ru/learn/bounded-model-checking/">20 Ограниченная проверка моделей</a></li><li><a href="/ru/learn/induction/">21 Индукция</a></li></ol></li><li><h3>Модуль 8 · Мутации</h3><p>Ломайте дизайн нарочно и считайте, что заметили тесты.</p><ol><li><a href="/ru/learn/kill-the-mutant/">22 Убить мутанта</a></li><li><a href="/ru/learn/surviving-mutants/">23 Выжившие мутанты</a></li><li><a href="/ru/learn/test-strength/">24 Сила тестов</a></li></ol></li><li><h3>Модуль 9 · Приёмка</h3><p>Одна команда, все квитанции, чистый вердикт, который можно показать.</p><ol><li><a href="/ru/learn/build-verify/">25 Сборка и проверка</a></li><li><a href="/ru/learn/board-selftest/">26 Стенд проверяет себя</a></li><li><a href="/ru/learn/verify-capstone/">27 Капстоун: проверить модуль целиком</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">Этот курс сам является spec t27: прочитайте его</a></p><p>Виджеты сохраняют свои английские слова: каждый живёт в собственном spec t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/setup-hold-window/index.html b/apps/website/public/ru/learn/setup-hold-window/index.html
index 947260c837..22e003edd6 100644
--- a/apps/website/public/ru/learn/setup-hold-window/index.html
+++ b/apps/website/public/ru/learn/setup-hold-window/index.html
@@ -300,5 +300,6 @@ <h1>Окно setup-hold</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое тактовый сигнал</h3><p>Один фронт, один мир: всё, что делит фронт тактового сигнала, делит и мир; период и джиттер настоящего фронта и место, где тактовый сигнал входит в плату.</p><ol><li><a href="/ru/learn/one-edge-one-world/">01 Один фронт, один мир</a></li><li><a href="/ru/learn/period-and-jitter/">02 Период, фронт и джиттер</a></li><li><a href="/ru/learn/where-the-clock-enters/">03 Где входит тактовый сигнал</a></li></ol></li><li><h3>Модуль 2 · Деревья тактового сигнала</h3><p>Перекос и задержка прохождения, глобальная сеть буферов и ловушка стробирования клока логикой.</p><ol><li><a href="/ru/learn/skew-and-insertion/">04 Перекос и задержка прохождения</a></li><li><a href="/ru/learn/global-buffers/">05 Глобальный буфер</a></li><li><a href="/ru/learn/gated-clock-trap/">06 Ловушка стробированного клока</a></li></ol></li><li><h3>Модуль 3 · PLL и MMCM</h3><p>Умножить и поделить один тактовый сигнал в другой, сдвинуть его фазу шагами VCO и какие тактовые сигналы анализатор считает родственными.</p><ol><li><a href="/ru/learn/multiply-and-divide/">07 Умножить и поделить</a></li><li><a href="/ru/learn/phase-and-lock/">08 Фаза и захват</a></li><li><a href="/ru/learn/related-or-not/">09 Родственные или чужие</a></li></ol></li><li><h3>Модуль 4 · Сбросы</h3><p>Ассертировать асинхронно, снимать синхронно: три вида сброса, конвейер снятия и дерево, которое растит сброс.</p><ol><li><a href="/ru/learn/sync-or-async/">10 Синхронный или асинхронный</a></li><li><a href="/ru/learn/reset-release/">11 Конвейер снятия</a></li><li><a href="/ru/learn/reset-fanout/">12 Разветвление сброса</a></li></ol></li><li><h3>Модуль 5 · Метастабильность</h3><p>Окно setup-hold, среднее время между отказами в целочисленной арифметике и два триггера, которые это чинят.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/setup-hold-window/">13 Окно setup-hold</a></li><li><a href="/ru/learn/mtbf/">14 Среднее время между отказами</a></li><li><a href="/ru/learn/two-flop-sync/">15 Двухтриггерный синхронизатор</a></li></ol></li><li><h3>Модуль 6 · Пересечение многих бит</h3><p>Почему двоичная шина рвётся, почему код Грея нет, и рукопожатие, переносящее импульс между мирами.</p><ol><li><a href="/ru/learn/why-binary-fails/">16 Почему двоичный код рвётся</a></li><li><a href="/ru/learn/gray-code/">17 Код Грея</a></li><li><a href="/ru/learn/handshakes-and-pulses/">18 Рукопожатия и импульсы</a></li></ol></li><li><h3>Модуль 7 · Асинхронный FIFO</h3><p>Указатели, флаги и глубина: буфер, переносящий поток между двумя тактовыми сигналами.</p><ol><li><a href="/ru/learn/pointers-and-flags/">19 Указатели и флаги</a></li><li><a href="/ru/learn/depth-sizing/">20 Выбор глубины</a></li><li><a href="/ru/learn/fifo-testbench/">21 Тестбенч FIFO</a></li></ol></li><li><h3>Модуль 8 · Ограничения</h3><p>Строки, сообщающие анализатору, что такое тактовый сигнал, какие пути не проверять и чему должны соответствовать выводы.</p><ol><li><a href="/ru/learn/create-clock/">22 Строка create_clock</a></li><li><a href="/ru/learn/false-path-and-max-delay/">23 False path и max delay</a></li><li><a href="/ru/learn/io-timing/">24 Тайминг выводов</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Отчёт CDC, одно пересечение, захваченное на триггерах, и дифф битстрима, замыкающий курс.</p><ol><li><a href="/ru/learn/cdc-report/">25 Отчёт CDC</a></li><li><a href="/ru/learn/capture-a-crossing/">26 Захватить пересечение</a></li><li><a href="/ru/learn/capstone-diff/">27 Капстоун: поменяй и сравни</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/seven-backends-and-t27b/index.html b/apps/website/public/ru/learn/seven-backends-and-t27b/index.html
index f1276ece83..6dd5462ba0 100644
--- a/apps/website/public/ru/learn/seven-backends-and-t27b/index.html
+++ b/apps/website/public/ru/learn/seven-backends-and-t27b/index.html
@@ -159,5 +159,6 @@ <h1>Семь бэкендов и t27b</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Чип</h3><p>Что такое FPGA, какой чип у нас и как его выводы встречаются с платой.</p><ol><li><a href="/ru/learn/fpga-is-a-grid/">01 FPGA — это сетка, которую вы настраиваете</a></li><li><a href="/ru/learn/which-chip/">02 Какой это чип?</a></li><li><a href="/ru/learn/pins-and-package/">03 Выводы, шарики и корпус</a></li></ol></li><li><h3>Модуль 2 · Числа в железе</h3><p>Биты, триты и форматы чисел, и сколько стоит арифметика в логике.</p><ol><li><a href="/ru/learn/bits-and-trits/">04 Биты, триты и GF16</a></li><li><a href="/ru/learn/what-arithmetic-costs/">05 Сколько стоит арифметика</a></li><li><a href="/ru/learn/how-much-fits/">06 Сколько помещается на чип</a></li></ol></li><li><h3>Модуль 3 · Ваша программа на t27</h3><p>Напишите спеку, проверьте её и увидьте, почему «компилируется» не значит «верно».</p><ol><li><a href="/ru/learn/hello-t27/">07 Привет, t27</a></li><li><a href="/ru/learn/tests-are-the-spec/">08 Тесты — часть спеки</a></li><li><a href="/ru/learn/compiles-is-not-correct/">09 Компилируется — не значит верно</a></li></ol></li><li><h3>Модуль 4 · Внутри t27c</h3><p>Как компилятор читает спеку и что он пишет, включая нативный t27b.</p><ol><li><a href="/ru/learn/compiler-stages/">10 Стадии t27c</a></li><li><a href="/ru/learn/t27c-on-a-real-machine/">11 t27c на настоящей машине</a></li><li class="on" aria-current="page"><a href="/ru/learn/seven-backends-and-t27b/">12 Семь бэкендов и t27b</a></li></ol></li><li><h3>Модуль 5 · От спеки к железу</h3><p>Verilog, который пишет t27c, ячейки, в которые он превращается, и что делает одна LUT.</p><ol><li><a href="/ru/learn/spec-to-verilog/">13 От спеки к Verilog</a></li><li><a href="/ru/learn/count-the-cells/">14 Посчитайте ячейки</a></li><li><a href="/ru/learn/one-lut-up-close/">15 Одна LUT вблизи</a></li></ol></li><li><h3>Модуль 6 · Читаем синтез</h3><p>Что yosys сообщает о вашем дизайне и какие предупреждения важны.</p><ol><li><a href="/ru/learn/synthesis-warnings/">16 Важные предупреждения синтеза</a></li><li><a href="/ru/learn/state-machines/">17 Конечные автоматы</a></li><li><a href="/ru/learn/where-the-luts-go/">18 Куда уходят LUT</a></li></ol></li><li><h3>Модуль 7 · Размещение, трассировка, тайминг</h3><p>Где ячейки оказываются на кристалле и успевает ли тактовая частота.</p><ol><li><a href="/ru/learn/placement/">19 Размещение: где оказываются ячейки</a></li><li><a href="/ru/learn/logic-depth/">20 Глубина логики</a></li><li><a href="/ru/learn/timing-slack/">21 Тайминг и запас</a></li></ol></li><li><h3>Модуль 8 · Битстрим</h3><p>Как оттрассированный дизайн становится битами, которые загружает чип, без Vivado.</p><ol><li><a href="/ru/learn/fasm-to-frames/">22 От FASM к фреймам</a></li><li><a href="/ru/learn/a-build-receipt/">23 Сборка по пунктам</a></li><li><a href="/ru/learn/read-a-bitstream/">24 Читаем битстрим</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Спросите чип, кто он, загрузите биты и проверьте их.</p><ol><li><a href="/ru/learn/jtag-asks-the-chip/">25 JTAG спрашивает чип</a></li><li><a href="/ru/learn/same-bits-on-the-board/">26 Те же биты — на плате</a></li><li><a href="/ru/learn/capstone/">27 Итог: переугадайте yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/seven-backends/index.html b/apps/website/public/ru/learn/seven-backends/index.html
index 0ac8f27344..4fb59e8457 100644
--- a/apps/website/public/ru/learn/seven-backends/index.html
+++ b/apps/website/public/ru/learn/seven-backends/index.html
@@ -85,5 +85,6 @@ <h1>Семь бэкендов</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Файл спеки</h3><p>Что лежит в файле .t27: строка module, строки прозы и комментарии.</p><ol><li><a href="/ru/learn/what-a-spec-is/">01 Что такое спека</a></li><li><a href="/ru/learn/the-module-line/">02 Строка module</a></li><li><a href="/ru/learn/comments-and-prose/">03 Комментарии и строки прозы</a></li></ol></li><li><h3>Модуль 2 · Значения и типы</h3><p>Константы, ширина целого числа, а также true, false и текст.</p><ol><li><a href="/ru/learn/constants/">04 Константы</a></li><li><a href="/ru/learn/whole-numbers/">05 Целые числа и их ширина</a></li><li><a href="/ru/learn/true-false-and-text/">06 Истина, ложь и текст</a></li></ol></li><li><h3>Модуль 3 · Массивы, триты и выражения</h3><p>Списки значений, три значения трита и что вычисляют операторы.</p><ol><li><a href="/ru/learn/arrays/">07 Массивы</a></li><li><a href="/ru/learn/trits/">08 Триты: минус один, ноль, один</a></li><li><a href="/ru/learn/expressions/">09 Выражения</a></li></ol></li><li><h3>Модуль 4 · Тесты и функции</h3><p>Блок test, несколько маленьких тестов в одной спеке и функция с типизированными входами.</p><ol><li><a href="/ru/learn/a-test-block/">10 Блок test</a></li><li><a href="/ru/learn/many-tests/">11 Много тестов в одной спеке</a></li><li><a href="/ru/learn/functions/">12 Функции</a></li></ol></li><li><h3>Модуль 5 · Внутри функции</h3><p>Имена через let и var, выбор через if и switch и циклы.</p><ol><li><a href="/ru/learn/local-names/">13 Локальные имена: let и var</a></li><li><a href="/ru/learn/if-and-else/">14 Выбор: if, else и switch</a></li><li><a href="/ru/learn/loops/">15 Циклы: while и for</a></li></ol></li><li><h3>Модуль 6 · Видимость и свои типы</h3><p>Что pub открывает другим модулям, и структуры и перечисления, которые вы объявляете.</p><ol><li><a href="/ru/learn/pub-or-private/">16 pub или закрытое имя</a></li><li><a href="/ru/learn/structs/">17 Структуры</a></li><li><a href="/ru/learn/enums/">18 Перечисления</a></li></ol></li><li><h3>Модуль 7 · Модули, правила и gen-ts</h3><p>Чтение другого модуля, правило, которое выполняется всегда, и спека, превращённая в TypeScript.</p><ol><li><a href="/ru/learn/use-other-modules/">19 Другие модули</a></li><li><a href="/ru/learn/invariants/">20 Инварианты и замеры</a></li><li><a href="/ru/learn/gen-ts/">21 gen-ts: спека превращается в TypeScript</a></li></ol></li><li><h3>Модуль 8 · Компилятор и tri</h3><p>Бэкенды, которые пишет t27c, стадии, в которых он читает спеку, и команда tri.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/seven-backends/">22 Семь бэкендов</a></li><li><a href="/ru/learn/stages-of-t27c/">23 Стадии t27c</a></li><li><a href="/ru/learn/tri-commands/">24 tri: командная строка</a></li></ol></li><li><h3>Модуль 9 · Ошибки, программа и что дальше</h3><p>Как читать ошибку компилятора, одна маленькая законченная программа и какой курс пройти дальше.</p><ol><li><a href="/ru/learn/reading-errors/">25 Как читать ошибки компилятора</a></li><li><a href="/ru/learn/a-small-program/">26 Маленькая законченная программа</a></li><li><a href="/ru/learn/where-next/">27 Куда дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/shift-and-add/index.html b/apps/website/public/ru/learn/shift-and-add/index.html
index 39d1145376..03b444b626 100644
--- a/apps/website/public/ru/learn/shift-and-add/index.html
+++ b/apps/website/public/ru/learn/shift-and-add/index.html
@@ -372,5 +372,6 @@ <h1>Сдвиг и сложение</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Сложение</h3><p>Почему сложение медленное: перенос, который идёт вверх по слову, выделенная цепь в каждом слайсе, префиксные сети и деревья сохранения переноса.</p><ol><li><a href="/ru/learn/the-carry-chain/">01 Цепь переноса</a></li><li><a href="/ru/learn/prefix-adders/">02 Взгляд вперёд: префиксные сумматоры</a></li><li><a href="/ru/learn/carry-save/">03 Три числа сразу</a></li></ol></li><li><h3>Модуль 2 · Умножение</h3><p>Произведение — сумма сдвинутых копий; перекодирование Бута вдвое сокращает их, а слайс DSP48E1 делает остальное одним блоком.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/shift-and-add/">04 Сдвиг и сложение</a></li><li><a href="/ru/learn/booths-trick/">05 Приём Бута</a></li><li><a href="/ru/learn/the-dsp48e1-slice/">06 Слайс DSP48E1</a></li></ol></li><li><h3>Модуль 3 · Фиксированная точка</h3><p>Где стоит двоичная точка, что округление делает со значением и с его средним и что даёт каждый бит квантователя.</p><ol><li><a href="/ru/learn/q-formats/">07 Форматы Q</a></li><li><a href="/ru/learn/rounding-and-saturation/">08 Округление и насыщение</a></li><li><a href="/ru/learn/quantization-noise/">09 Шум квантования</a></li></ol></li><li><h3>Модуль 4 · Функции в железе</h3><p>Синус, косинус, угол и длина из сдвигов и сложений — и когда таблица оказывается выгоднее.</p><ol><li><a href="/ru/learn/rotating-by-shifts/">10 Поворот сдвигами</a></li><li><a href="/ru/learn/angle-and-magnitude/">11 Угол и модуль</a></li><li><a href="/ru/learn/tables-and-interpolation/">12 Таблицы и интерполяция</a></li></ol></li><li><h3>Модуль 5 · Сигналы и дискретизация</h3><p>Что дискретизация делает с частотой, как генератор строится из сумматора и что измеряет бин ДПФ.</p><ol><li><a href="/ru/learn/sampling-and-aliasing/">13 Дискретизация и наложение спектров</a></li><li><a href="/ru/learn/the-nco/">14 Генератор с числовым управлением</a></li><li><a href="/ru/learn/bins-of-a-spectrum/">15 Бины спектра</a></li></ol></li><li><h3>Модуль 6 · Фильтры FIR</h3><p>Скользящее среднее, расчёт окном sinc с целыми коэффициентами и раскладка отводов по слайсам DSP.</p><ol><li><a href="/ru/learn/the-moving-average/">16 Скользящее среднее</a></li><li><a href="/ru/learn/designing-taps/">17 Расчёт коэффициентов</a></li><li><a href="/ru/learn/folding-taps-onto-slices/">18 Раскладка отводов по слайсам</a></li></ol></li><li><h3>Модуль 7 · Многоскоростная обработка</h3><p>Понижение частоты дискретизации без заворота шума: децимация, фильтр CIC и полифазная форма.</p><ol><li><a href="/ru/learn/lowering-the-rate/">19 Понижение частоты</a></li><li><a href="/ru/learn/the-cic-filter/">20 Фильтр CIC</a></li><li><a href="/ru/learn/polyphase/">21 Полифазная форма</a></li></ol></li><li><h3>Модуль 8 · БПФ</h3><p>N log N вместо N^2: бабочки, бит-реверсный порядок входа и биты, которые добавляет каждый этап.</p><ol><li><a href="/ru/learn/the-butterfly/">22 Бабочка</a></li><li><a href="/ru/learn/twiddles-and-bit-reversal/">23 Поворачивающие множители и бит-реверс</a></li><li><a href="/ru/learn/growth-through-the-stages/">24 Рост через этапы</a></li></ol></li><li><h3>Модуль 9 · На стенде</h3><p>От арифметики к плате: коррелятор из спеки модема, бюджет слайсов и тайминга и итоговый фильтр.</p><ol><li><a href="/ru/learn/a-correlator-finds-the-frame/">25 Коррелятор находит кадр</a></li><li><a href="/ru/learn/does-it-fit-and-close-timing/">26 Помещается ли и сходится ли тайминг?</a></li><li><a href="/ru/learn/capstone-a-filter-on-the-board/">27 Итог: фильтр на плате</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/simulator-internals/index.html b/apps/website/public/ru/learn/simulator-internals/index.html
index 7894cbd8c7..7232dd3f22 100644
--- a/apps/website/public/ru/learn/simulator-internals/index.html
+++ b/apps/website/public/ru/learn/simulator-internals/index.html
@@ -384,5 +384,6 @@ <h1>Внутри симулятора</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Зачем проверять</h3><p>Дизайны, которые компилируются и ошибаются; модель, которая выносит вердикт; план, записанный до кода.</p><ol><li><a href="/ru/learn/bugs-that-compile/">01 Баги, которые компилируются</a></li><li><a href="/ru/learn/golden-model/">02 Золотая модель</a></li><li><a href="/ru/learn/plan-before-code/">03 Сначала план</a></li></ol></li><li><h3>Модуль 2 · Тестбенчи</h3><p>Стимулы, проверки и вердикт, записанные как один spec рядом с дизайном, который они судят.</p><ol><li><a href="/ru/learn/anatomy-of-a-tb/">04 Анатомия тестбенча</a></li><li><a href="/ru/learn/self-checking/">05 Самопроверка вместо всматривания</a></li><li><a href="/ru/learn/directed-vs-random/">06 Направленные против случайных</a></li></ol></li><li><h3>Модуль 3 · Временные диаграммы</h3><p>Трасса каждого сигнала, прочитанная так, как её читает инженер по железу, и два прогона, сравнённые между собой.</p><ol><li><a href="/ru/learn/vcd-format/">07 Формат VCD</a></li><li><a href="/ru/learn/debug-with-waves/">08 Отладка по волнам</a></li><li><a href="/ru/learn/compare-traces/">09 Сравнение двух трасс</a></li></ol></li><li><h3>Модуль 4 · Векторы соответствия</h3><p>Случаи с ответом, записанным рядом, там, откуда их достанет компилятор.</p><ol><li><a href="/ru/learn/what-a-vector-is/">10 Что такое вектор</a></li><li><a href="/ru/learn/golden-ruler-vectors/">11 Векторы как эталонная линейка</a></li><li><a href="/ru/learn/end-to-end-scenarios/">12 Сквозные сценарии</a></li></ol></li><li><h3>Модуль 5 · Косимуляция</h3><p>Spec, симулятор и плата сходятся в одном ответе на стенде Artix-7 XC7A200T, и что делать, когда не сходятся.</p><ol><li><a href="/ru/learn/three-way-agreement/">13 Тройное согласие</a></li><li class="on" aria-current="page"><a href="/ru/learn/simulator-internals/">14 Внутри симулятора</a></li><li><a href="/ru/learn/when-they-disagree/">15 Когда они не сходятся</a></li></ol></li><li><h3>Модуль 6 · Покрытие</h3><p>Чего коснулись тесты: строки, переключения, состояния — и что прячет это число.</p><ol><li><a href="/ru/learn/line-and-toggle/">16 Строки и переключения</a></li><li><a href="/ru/learn/fsm-coverage/">17 Покрытие автоматов</a></li><li><a href="/ru/learn/coverage-lies/">18 Покрытие лжёт</a></li></ol></li><li><h3>Модуль 7 · Формальные методы</h3><p>Ассерты, верные каждый такт; ограниченный поиск контрпримера; и почему доказательству нужна индукция.</p><ol><li><a href="/ru/learn/assertions/">19 Ассерты</a></li><li><a href="/ru/learn/bounded-model-checking/">20 Ограниченная проверка моделей</a></li><li><a href="/ru/learn/induction/">21 Индукция</a></li></ol></li><li><h3>Модуль 8 · Мутации</h3><p>Ломайте дизайн нарочно и считайте, что заметили тесты.</p><ol><li><a href="/ru/learn/kill-the-mutant/">22 Убить мутанта</a></li><li><a href="/ru/learn/surviving-mutants/">23 Выжившие мутанты</a></li><li><a href="/ru/learn/test-strength/">24 Сила тестов</a></li></ol></li><li><h3>Модуль 9 · Приёмка</h3><p>Одна команда, все квитанции, чистый вердикт, который можно показать.</p><ol><li><a href="/ru/learn/build-verify/">25 Сборка и проверка</a></li><li><a href="/ru/learn/board-selftest/">26 Стенд проверяет себя</a></li><li><a href="/ru/learn/verify-capstone/">27 Капстоун: проверить модуль целиком</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">Этот курс сам является spec t27: прочитайте его</a></p><p>Виджеты сохраняют свои английские слова: каждый живёт в собственном spec t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/skew-and-insertion/index.html b/apps/website/public/ru/learn/skew-and-insertion/index.html
index da7658fb41..a042bc2da7 100644
--- a/apps/website/public/ru/learn/skew-and-insertion/index.html
+++ b/apps/website/public/ru/learn/skew-and-insertion/index.html
@@ -294,5 +294,6 @@ <h1>Перекос и задержка прохождения</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое тактовый сигнал</h3><p>Один фронт, один мир: всё, что делит фронт тактового сигнала, делит и мир; период и джиттер настоящего фронта и место, где тактовый сигнал входит в плату.</p><ol><li><a href="/ru/learn/one-edge-one-world/">01 Один фронт, один мир</a></li><li><a href="/ru/learn/period-and-jitter/">02 Период, фронт и джиттер</a></li><li><a href="/ru/learn/where-the-clock-enters/">03 Где входит тактовый сигнал</a></li></ol></li><li><h3>Модуль 2 · Деревья тактового сигнала</h3><p>Перекос и задержка прохождения, глобальная сеть буферов и ловушка стробирования клока логикой.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/skew-and-insertion/">04 Перекос и задержка прохождения</a></li><li><a href="/ru/learn/global-buffers/">05 Глобальный буфер</a></li><li><a href="/ru/learn/gated-clock-trap/">06 Ловушка стробированного клока</a></li></ol></li><li><h3>Модуль 3 · PLL и MMCM</h3><p>Умножить и поделить один тактовый сигнал в другой, сдвинуть его фазу шагами VCO и какие тактовые сигналы анализатор считает родственными.</p><ol><li><a href="/ru/learn/multiply-and-divide/">07 Умножить и поделить</a></li><li><a href="/ru/learn/phase-and-lock/">08 Фаза и захват</a></li><li><a href="/ru/learn/related-or-not/">09 Родственные или чужие</a></li></ol></li><li><h3>Модуль 4 · Сбросы</h3><p>Ассертировать асинхронно, снимать синхронно: три вида сброса, конвейер снятия и дерево, которое растит сброс.</p><ol><li><a href="/ru/learn/sync-or-async/">10 Синхронный или асинхронный</a></li><li><a href="/ru/learn/reset-release/">11 Конвейер снятия</a></li><li><a href="/ru/learn/reset-fanout/">12 Разветвление сброса</a></li></ol></li><li><h3>Модуль 5 · Метастабильность</h3><p>Окно setup-hold, среднее время между отказами в целочисленной арифметике и два триггера, которые это чинят.</p><ol><li><a href="/ru/learn/setup-hold-window/">13 Окно setup-hold</a></li><li><a href="/ru/learn/mtbf/">14 Среднее время между отказами</a></li><li><a href="/ru/learn/two-flop-sync/">15 Двухтриггерный синхронизатор</a></li></ol></li><li><h3>Модуль 6 · Пересечение многих бит</h3><p>Почему двоичная шина рвётся, почему код Грея нет, и рукопожатие, переносящее импульс между мирами.</p><ol><li><a href="/ru/learn/why-binary-fails/">16 Почему двоичный код рвётся</a></li><li><a href="/ru/learn/gray-code/">17 Код Грея</a></li><li><a href="/ru/learn/handshakes-and-pulses/">18 Рукопожатия и импульсы</a></li></ol></li><li><h3>Модуль 7 · Асинхронный FIFO</h3><p>Указатели, флаги и глубина: буфер, переносящий поток между двумя тактовыми сигналами.</p><ol><li><a href="/ru/learn/pointers-and-flags/">19 Указатели и флаги</a></li><li><a href="/ru/learn/depth-sizing/">20 Выбор глубины</a></li><li><a href="/ru/learn/fifo-testbench/">21 Тестбенч FIFO</a></li></ol></li><li><h3>Модуль 8 · Ограничения</h3><p>Строки, сообщающие анализатору, что такое тактовый сигнал, какие пути не проверять и чему должны соответствовать выводы.</p><ol><li><a href="/ru/learn/create-clock/">22 Строка create_clock</a></li><li><a href="/ru/learn/false-path-and-max-delay/">23 False path и max delay</a></li><li><a href="/ru/learn/io-timing/">24 Тайминг выводов</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Отчёт CDC, одно пересечение, захваченное на триггерах, и дифф битстрима, замыкающий курс.</p><ol><li><a href="/ru/learn/cdc-report/">25 Отчёт CDC</a></li><li><a href="/ru/learn/capture-a-crossing/">26 Захватить пересечение</a></li><li><a href="/ru/learn/capstone-diff/">27 Капстоун: поменяй и сравни</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/sort-by-three/index.html b/apps/website/public/ru/learn/sort-by-three/index.html
index e516a3daf7..183029cecc 100644
--- a/apps/website/public/ru/learn/sort-by-three/index.html
+++ b/apps/website/public/ru/learn/sort-by-three/index.html
@@ -769,5 +769,6 @@ <h1>Сортировка трёхзначными сравнениями</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Три значения</h3><p>Почему три, как сбалансированная троичная система пишет любое число без знака и что дают переворот и отсечение тритов.</p><ol><li><a href="/ru/learn/why-three/">01 Почему три</a></li><li><a href="/ru/learn/balanced-ternary/">02 Сбалансированная троичная</a></li><li><a href="/ru/learn/flip-and-cut/">03 Переворот вместо минуса, отсечение вместо округления</a></li></ol></li><li><h3>Модуль 2 · Логика с «не знаю»</h3><p>Трёхзначные вентили Клини, два вентиля без двоичного двойника и сложение как пара таблиц.</p><ol><li><a href="/ru/learn/kleene-logic/">04 Трёхзначная логика</a></li><li><a href="/ru/learn/consensus-and-any/">05 Консенсус и «любой»</a></li><li><a href="/ru/learn/sum-and-carry/">06 Сумма и перенос</a></li></ol></li><li><h3>Модуль 3 · Складываем триты</h3><p>Полусумматор, полный сумматор и перенос, бегущий влево разряд за разрядом.</p><ol><li><a href="/ru/learn/half-adder/">07 Полусумматор</a></li><li><a href="/ru/learn/full-adder/">08 Полный сумматор</a></li><li><a href="/ru/learn/ripple-carry/">09 Сквозной перенос</a></li></ol></li><li><h3>Модуль 4 · Умножение без умножителя</h3><p>Копия, ноль или переворот: произведение на один трит, умножение столбиком и MAC, который только складывает.</p><ol><li><a href="/ru/learn/times-a-trit/">10 Умножить на один трит</a></li><li><a href="/ru/learn/long-multiplication/">11 Умножение столбиком</a></li><li><a href="/ru/learn/ternary-mac/">12 Троичный MAC</a></li></ol></li><li><h3>Модуль 5 · Триты в двоичной памяти</h3><p>Два бита на трит, пять тритов на байт и сколько бит нужно слову из 27 тритов.</p><ol><li><a href="/ru/learn/two-bits-per-trit/">13 Два бита на трит</a></li><li><a href="/ru/learn/five-trits-a-byte/">14 Пять тритов в байте</a></li><li><a href="/ru/learn/a-word-of-27/">15 Слово из 27 тритов</a></li></ol></li><li><h3>Модуль 6 · Командное слово TRI-27</h3><p>32-битное слово, которое декодирует эмулятор Trinity: его поля, 47 опкодов и 15-битное непосредственное значение.</p><ol><li><a href="/ru/learn/the-instruction-word/">16 Командное слово</a></li><li><a href="/ru/learn/the-opcode-map/">17 Карта опкодов</a></li><li><a href="/ru/learn/the-immediate/">18 Непосредственное значение</a></li></ol></li><li><h3>Модуль 7 · Троичная машина</h3><p>АЛУ из сумматоров этого курса, переход на три стороны и программа, которую можно пройти по шагам.</p><ol><li><a href="/ru/learn/a-ternary-alu/">19 Троичное АЛУ</a></li><li><a href="/ru/learn/three-way-jump/">20 Переход на три стороны</a></li><li><a href="/ru/learn/run-a-program/">21 Запустить программу</a></li></ol></li><li><h3>Модуль 8 · Три ответа</h3><p>Сравнение по старшему различному триту, поиск числа по третям и сортировка трёхзначными сравнениями.</p><ol><li><a href="/ru/learn/compare-at-the-top/">22 Сравнение сверху</a></li><li><a href="/ru/learn/questions-in-thirds/">23 Вопросы по третям</a></li><li class="on" aria-current="page"><a href="/ru/learn/sort-by-three/">24 Сортировка трёхзначными сравнениями</a></li></ol></li><li><h3>Модуль 9 · Троичные нейроны</h3><p>Веса -1, 0, +1: один нейрон, детектор и маленький слой, и нигде нет умножителя.</p><ol><li><a href="/ru/learn/one-ternary-neuron/">25 Один троичный нейрон</a></li><li><a href="/ru/learn/a-bar-detector/">26 Детектор полосы</a></li><li><a href="/ru/learn/a-tiny-layer/">27 Маленький слой</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/spec-to-verilog/index.html b/apps/website/public/ru/learn/spec-to-verilog/index.html
index 14b8d28304..2edec81097 100644
--- a/apps/website/public/ru/learn/spec-to-verilog/index.html
+++ b/apps/website/public/ru/learn/spec-to-verilog/index.html
@@ -107,5 +107,6 @@ <h1>От спеки к Verilog</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Чип</h3><p>Что такое FPGA, какой чип у нас и как его выводы встречаются с платой.</p><ol><li><a href="/ru/learn/fpga-is-a-grid/">01 FPGA — это сетка, которую вы настраиваете</a></li><li><a href="/ru/learn/which-chip/">02 Какой это чип?</a></li><li><a href="/ru/learn/pins-and-package/">03 Выводы, шарики и корпус</a></li></ol></li><li><h3>Модуль 2 · Числа в железе</h3><p>Биты, триты и форматы чисел, и сколько стоит арифметика в логике.</p><ol><li><a href="/ru/learn/bits-and-trits/">04 Биты, триты и GF16</a></li><li><a href="/ru/learn/what-arithmetic-costs/">05 Сколько стоит арифметика</a></li><li><a href="/ru/learn/how-much-fits/">06 Сколько помещается на чип</a></li></ol></li><li><h3>Модуль 3 · Ваша программа на t27</h3><p>Напишите спеку, проверьте её и увидьте, почему «компилируется» не значит «верно».</p><ol><li><a href="/ru/learn/hello-t27/">07 Привет, t27</a></li><li><a href="/ru/learn/tests-are-the-spec/">08 Тесты — часть спеки</a></li><li><a href="/ru/learn/compiles-is-not-correct/">09 Компилируется — не значит верно</a></li></ol></li><li><h3>Модуль 4 · Внутри t27c</h3><p>Как компилятор читает спеку и что он пишет, включая нативный t27b.</p><ol><li><a href="/ru/learn/compiler-stages/">10 Стадии t27c</a></li><li><a href="/ru/learn/t27c-on-a-real-machine/">11 t27c на настоящей машине</a></li><li><a href="/ru/learn/seven-backends-and-t27b/">12 Семь бэкендов и t27b</a></li></ol></li><li><h3>Модуль 5 · От спеки к железу</h3><p>Verilog, который пишет t27c, ячейки, в которые он превращается, и что делает одна LUT.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/spec-to-verilog/">13 От спеки к Verilog</a></li><li><a href="/ru/learn/count-the-cells/">14 Посчитайте ячейки</a></li><li><a href="/ru/learn/one-lut-up-close/">15 Одна LUT вблизи</a></li></ol></li><li><h3>Модуль 6 · Читаем синтез</h3><p>Что yosys сообщает о вашем дизайне и какие предупреждения важны.</p><ol><li><a href="/ru/learn/synthesis-warnings/">16 Важные предупреждения синтеза</a></li><li><a href="/ru/learn/state-machines/">17 Конечные автоматы</a></li><li><a href="/ru/learn/where-the-luts-go/">18 Куда уходят LUT</a></li></ol></li><li><h3>Модуль 7 · Размещение, трассировка, тайминг</h3><p>Где ячейки оказываются на кристалле и успевает ли тактовая частота.</p><ol><li><a href="/ru/learn/placement/">19 Размещение: где оказываются ячейки</a></li><li><a href="/ru/learn/logic-depth/">20 Глубина логики</a></li><li><a href="/ru/learn/timing-slack/">21 Тайминг и запас</a></li></ol></li><li><h3>Модуль 8 · Битстрим</h3><p>Как оттрассированный дизайн становится битами, которые загружает чип, без Vivado.</p><ol><li><a href="/ru/learn/fasm-to-frames/">22 От FASM к фреймам</a></li><li><a href="/ru/learn/a-build-receipt/">23 Сборка по пунктам</a></li><li><a href="/ru/learn/read-a-bitstream/">24 Читаем битстрим</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Спросите чип, кто он, загрузите биты и проверьте их.</p><ol><li><a href="/ru/learn/jtag-asks-the-chip/">25 JTAG спрашивает чип</a></li><li><a href="/ru/learn/same-bits-on-the-board/">26 Те же биты — на плате</a></li><li><a href="/ru/learn/capstone/">27 Итог: переугадайте yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/stages-of-t27c/index.html b/apps/website/public/ru/learn/stages-of-t27c/index.html
index 0d412d656f..128c625281 100644
--- a/apps/website/public/ru/learn/stages-of-t27c/index.html
+++ b/apps/website/public/ru/learn/stages-of-t27c/index.html
@@ -85,5 +85,6 @@ <h1>Стадии t27c</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Файл спеки</h3><p>Что лежит в файле .t27: строка module, строки прозы и комментарии.</p><ol><li><a href="/ru/learn/what-a-spec-is/">01 Что такое спека</a></li><li><a href="/ru/learn/the-module-line/">02 Строка module</a></li><li><a href="/ru/learn/comments-and-prose/">03 Комментарии и строки прозы</a></li></ol></li><li><h3>Модуль 2 · Значения и типы</h3><p>Константы, ширина целого числа, а также true, false и текст.</p><ol><li><a href="/ru/learn/constants/">04 Константы</a></li><li><a href="/ru/learn/whole-numbers/">05 Целые числа и их ширина</a></li><li><a href="/ru/learn/true-false-and-text/">06 Истина, ложь и текст</a></li></ol></li><li><h3>Модуль 3 · Массивы, триты и выражения</h3><p>Списки значений, три значения трита и что вычисляют операторы.</p><ol><li><a href="/ru/learn/arrays/">07 Массивы</a></li><li><a href="/ru/learn/trits/">08 Триты: минус один, ноль, один</a></li><li><a href="/ru/learn/expressions/">09 Выражения</a></li></ol></li><li><h3>Модуль 4 · Тесты и функции</h3><p>Блок test, несколько маленьких тестов в одной спеке и функция с типизированными входами.</p><ol><li><a href="/ru/learn/a-test-block/">10 Блок test</a></li><li><a href="/ru/learn/many-tests/">11 Много тестов в одной спеке</a></li><li><a href="/ru/learn/functions/">12 Функции</a></li></ol></li><li><h3>Модуль 5 · Внутри функции</h3><p>Имена через let и var, выбор через if и switch и циклы.</p><ol><li><a href="/ru/learn/local-names/">13 Локальные имена: let и var</a></li><li><a href="/ru/learn/if-and-else/">14 Выбор: if, else и switch</a></li><li><a href="/ru/learn/loops/">15 Циклы: while и for</a></li></ol></li><li><h3>Модуль 6 · Видимость и свои типы</h3><p>Что pub открывает другим модулям, и структуры и перечисления, которые вы объявляете.</p><ol><li><a href="/ru/learn/pub-or-private/">16 pub или закрытое имя</a></li><li><a href="/ru/learn/structs/">17 Структуры</a></li><li><a href="/ru/learn/enums/">18 Перечисления</a></li></ol></li><li><h3>Модуль 7 · Модули, правила и gen-ts</h3><p>Чтение другого модуля, правило, которое выполняется всегда, и спека, превращённая в TypeScript.</p><ol><li><a href="/ru/learn/use-other-modules/">19 Другие модули</a></li><li><a href="/ru/learn/invariants/">20 Инварианты и замеры</a></li><li><a href="/ru/learn/gen-ts/">21 gen-ts: спека превращается в TypeScript</a></li></ol></li><li><h3>Модуль 8 · Компилятор и tri</h3><p>Бэкенды, которые пишет t27c, стадии, в которых он читает спеку, и команда tri.</p><ol><li><a href="/ru/learn/seven-backends/">22 Семь бэкендов</a></li><li class="on" aria-current="page"><a href="/ru/learn/stages-of-t27c/">23 Стадии t27c</a></li><li><a href="/ru/learn/tri-commands/">24 tri: командная строка</a></li></ol></li><li><h3>Модуль 9 · Ошибки, программа и что дальше</h3><p>Как читать ошибку компилятора, одна маленькая законченная программа и какой курс пройти дальше.</p><ol><li><a href="/ru/learn/reading-errors/">25 Как читать ошибки компилятора</a></li><li><a href="/ru/learn/a-small-program/">26 Маленькая законченная программа</a></li><li><a href="/ru/learn/where-next/">27 Куда дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/state-machines/index.html b/apps/website/public/ru/learn/state-machines/index.html
index 05fdfe31e5..8359489f85 100644
--- a/apps/website/public/ru/learn/state-machines/index.html
+++ b/apps/website/public/ru/learn/state-machines/index.html
@@ -333,5 +333,6 @@ <h1>Конечные автоматы</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Чип</h3><p>Что такое FPGA, какой чип у нас и как его выводы встречаются с платой.</p><ol><li><a href="/ru/learn/fpga-is-a-grid/">01 FPGA — это сетка, которую вы настраиваете</a></li><li><a href="/ru/learn/which-chip/">02 Какой это чип?</a></li><li><a href="/ru/learn/pins-and-package/">03 Выводы, шарики и корпус</a></li></ol></li><li><h3>Модуль 2 · Числа в железе</h3><p>Биты, триты и форматы чисел, и сколько стоит арифметика в логике.</p><ol><li><a href="/ru/learn/bits-and-trits/">04 Биты, триты и GF16</a></li><li><a href="/ru/learn/what-arithmetic-costs/">05 Сколько стоит арифметика</a></li><li><a href="/ru/learn/how-much-fits/">06 Сколько помещается на чип</a></li></ol></li><li><h3>Модуль 3 · Ваша программа на t27</h3><p>Напишите спеку, проверьте её и увидьте, почему «компилируется» не значит «верно».</p><ol><li><a href="/ru/learn/hello-t27/">07 Привет, t27</a></li><li><a href="/ru/learn/tests-are-the-spec/">08 Тесты — часть спеки</a></li><li><a href="/ru/learn/compiles-is-not-correct/">09 Компилируется — не значит верно</a></li></ol></li><li><h3>Модуль 4 · Внутри t27c</h3><p>Как компилятор читает спеку и что он пишет, включая нативный t27b.</p><ol><li><a href="/ru/learn/compiler-stages/">10 Стадии t27c</a></li><li><a href="/ru/learn/t27c-on-a-real-machine/">11 t27c на настоящей машине</a></li><li><a href="/ru/learn/seven-backends-and-t27b/">12 Семь бэкендов и t27b</a></li></ol></li><li><h3>Модуль 5 · От спеки к железу</h3><p>Verilog, который пишет t27c, ячейки, в которые он превращается, и что делает одна LUT.</p><ol><li><a href="/ru/learn/spec-to-verilog/">13 От спеки к Verilog</a></li><li><a href="/ru/learn/count-the-cells/">14 Посчитайте ячейки</a></li><li><a href="/ru/learn/one-lut-up-close/">15 Одна LUT вблизи</a></li></ol></li><li><h3>Модуль 6 · Читаем синтез</h3><p>Что yosys сообщает о вашем дизайне и какие предупреждения важны.</p><ol><li><a href="/ru/learn/synthesis-warnings/">16 Важные предупреждения синтеза</a></li><li class="on" aria-current="page"><a href="/ru/learn/state-machines/">17 Конечные автоматы</a></li><li><a href="/ru/learn/where-the-luts-go/">18 Куда уходят LUT</a></li></ol></li><li><h3>Модуль 7 · Размещение, трассировка, тайминг</h3><p>Где ячейки оказываются на кристалле и успевает ли тактовая частота.</p><ol><li><a href="/ru/learn/placement/">19 Размещение: где оказываются ячейки</a></li><li><a href="/ru/learn/logic-depth/">20 Глубина логики</a></li><li><a href="/ru/learn/timing-slack/">21 Тайминг и запас</a></li></ol></li><li><h3>Модуль 8 · Битстрим</h3><p>Как оттрассированный дизайн становится битами, которые загружает чип, без Vivado.</p><ol><li><a href="/ru/learn/fasm-to-frames/">22 От FASM к фреймам</a></li><li><a href="/ru/learn/a-build-receipt/">23 Сборка по пунктам</a></li><li><a href="/ru/learn/read-a-bitstream/">24 Читаем битстрим</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Спросите чип, кто он, загрузите биты и проверьте их.</p><ol><li><a href="/ru/learn/jtag-asks-the-chip/">25 JTAG спрашивает чип</a></li><li><a href="/ru/learn/same-bits-on-the-board/">26 Те же биты — на плате</a></li><li><a href="/ru/learn/capstone/">27 Итог: переугадайте yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/status-and-fifos/index.html b/apps/website/public/ru/learn/status-and-fifos/index.html
index bbacec81a5..60676475a9 100644
--- a/apps/website/public/ru/learn/status-and-fifos/index.html
+++ b/apps/website/public/ru/learn/status-and-fifos/index.html
@@ -466,5 +466,6 @@ <h1>Статус и FIFO</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое шина</h3><p>Зачем вообще существует шина: разговор по проводам, с кадрами и адресами, и кому позволено говорить.</p><ol><li><a href="/ru/learn/wires-that-talk/">01 Провода, которые говорят</a></li><li><a href="/ru/learn/framing-a-conversation/">02 Кадрирование разговора</a></li><li><a href="/ru/learn/master-and-servant/">03 Хозяин и слуга</a></li></ol></li><li><h3>Модуль 2 · Разговор по UART</h3><p>Двухпроводная шина без такта: кадр, делитель, задающий скорость, и статус, который опрашивает драйвер.</p><ol><li><a href="/ru/learn/the-frame/">04 Кадр: старт, данные, стоп</a></li><li><a href="/ru/learn/baud-and-the-divisor/">05 Бод и делитель</a></li><li class="on" aria-current="page"><a href="/ru/learn/status-and-fifos/">06 Статус и FIFO</a></li></ol></li><li><h3>Модуль 3 · SPI под тактом</h3><p>Разговор под тактом: четыре режима, лестница предделителей и выбор кристалла на каждого подчинённого.</p><ol><li><a href="/ru/learn/modes-cpol-cpha/">07 Режимы: CPOL и CPHA</a></li><li><a href="/ru/learn/the-prescaler-ladder/">08 Лестница предделителей</a></li><li><a href="/ru/learn/chip-select-and-widths/">09 Выбор кристалла и ширины</a></li></ol></li><li><h3>Модуль 4 · Регистровая шина APB</h3><p>Шина регистров: PSEL и PENABLE, строобы и ожидания, и сколько адресных бит стоит количество периферии.</p><ol><li><a href="/ru/learn/psel-and-penable/">10 PSEL и PENABLE</a></li><li><a href="/ru/learn/strobes-and-waits/">11 Строобы и ожидания</a></li><li><a href="/ru/learn/address-decode/">12 Дешифрация адреса</a></li></ol></li><li><h3>Модуль 5 · Пять каналов AXI4</h3><p>Пять каналов: адрес, данные и отклик в обе стороны, lite или full, пакеты и идентификаторы.</p><ol><li><a href="/ru/learn/five-channels/">13 Пять каналов</a></li><li><a href="/ru/learn/lite-or-full/">14 Lite или full</a></li><li><a href="/ru/learn/bursts-and-ids/">15 Пакеты и идентификаторы</a></li></ol></li><li><h3>Модуль 6 · Память</h3><p>Что находится на дальнем конце каждой шины: карты памяти, типы портов и задержка, которую должно покрывать ожидание.</p><ol><li><a href="/ru/learn/memory-maps/">16 Карты памяти</a></li><li><a href="/ru/learn/ports-and-kinds/">17 Порты и типы</a></li><li><a href="/ru/learn/latency/">18 Задержка</a></li></ol></li><li><h3>Модуль 7 · Мосты</h3><p>Почему в дизайне появляется больше одной шины и пакетный мост, переносящий работу между ними.</p><ol><li><a href="/ru/learn/why-bridges/">19 Зачем мосты</a></li><li><a href="/ru/learn/the-packet-bridge/">20 Пакетный мост</a></li><li><a href="/ru/learn/everything-in-one-bench/">21 Всё в одном стенде</a></li></ol></li><li><h3>Модуль 8 · Ethernet: кадры и тайминг</h3><p>Кадры, контрольная сумма кадра, тайминг RGMII и заранее зарегистрированные шаги настоящего запуска.</p><ol><li><a href="/ru/learn/frames-and-the-fcs/">22 Кадры и контрольная сумма</a></li><li><a href="/ru/learn/rgmii-timing/">23 Тайминг RGMII</a></li><li><a href="/ru/learn/bring-up-on-the-bench/">24 Запуск на стенде</a></li></ol></li><li><h3>Модуль 9 · Стенд</h3><p>Дисциплина, оберегающая настоящее железо: кто держит IO, как взять и вернуть захват, и что запускается дальше.</p><ol><li><a href="/ru/learn/who-holds-the-io/">25 Кто держит IO</a></li><li><a href="/ru/learn/claim-and-release/">26 Захват и возврат</a></li><li><a href="/ru/learn/capstone-what-runs-next/">27 Капстоун: что запускается дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/strobes-and-waits/index.html b/apps/website/public/ru/learn/strobes-and-waits/index.html
index 9cd2c27ef4..5ca1a8d8b3 100644
--- a/apps/website/public/ru/learn/strobes-and-waits/index.html
+++ b/apps/website/public/ru/learn/strobes-and-waits/index.html
@@ -417,5 +417,6 @@ <h1>Строобы и ожидания</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое шина</h3><p>Зачем вообще существует шина: разговор по проводам, с кадрами и адресами, и кому позволено говорить.</p><ol><li><a href="/ru/learn/wires-that-talk/">01 Провода, которые говорят</a></li><li><a href="/ru/learn/framing-a-conversation/">02 Кадрирование разговора</a></li><li><a href="/ru/learn/master-and-servant/">03 Хозяин и слуга</a></li></ol></li><li><h3>Модуль 2 · Разговор по UART</h3><p>Двухпроводная шина без такта: кадр, делитель, задающий скорость, и статус, который опрашивает драйвер.</p><ol><li><a href="/ru/learn/the-frame/">04 Кадр: старт, данные, стоп</a></li><li><a href="/ru/learn/baud-and-the-divisor/">05 Бод и делитель</a></li><li><a href="/ru/learn/status-and-fifos/">06 Статус и FIFO</a></li></ol></li><li><h3>Модуль 3 · SPI под тактом</h3><p>Разговор под тактом: четыре режима, лестница предделителей и выбор кристалла на каждого подчинённого.</p><ol><li><a href="/ru/learn/modes-cpol-cpha/">07 Режимы: CPOL и CPHA</a></li><li><a href="/ru/learn/the-prescaler-ladder/">08 Лестница предделителей</a></li><li><a href="/ru/learn/chip-select-and-widths/">09 Выбор кристалла и ширины</a></li></ol></li><li><h3>Модуль 4 · Регистровая шина APB</h3><p>Шина регистров: PSEL и PENABLE, строобы и ожидания, и сколько адресных бит стоит количество периферии.</p><ol><li><a href="/ru/learn/psel-and-penable/">10 PSEL и PENABLE</a></li><li class="on" aria-current="page"><a href="/ru/learn/strobes-and-waits/">11 Строобы и ожидания</a></li><li><a href="/ru/learn/address-decode/">12 Дешифрация адреса</a></li></ol></li><li><h3>Модуль 5 · Пять каналов AXI4</h3><p>Пять каналов: адрес, данные и отклик в обе стороны, lite или full, пакеты и идентификаторы.</p><ol><li><a href="/ru/learn/five-channels/">13 Пять каналов</a></li><li><a href="/ru/learn/lite-or-full/">14 Lite или full</a></li><li><a href="/ru/learn/bursts-and-ids/">15 Пакеты и идентификаторы</a></li></ol></li><li><h3>Модуль 6 · Память</h3><p>Что находится на дальнем конце каждой шины: карты памяти, типы портов и задержка, которую должно покрывать ожидание.</p><ol><li><a href="/ru/learn/memory-maps/">16 Карты памяти</a></li><li><a href="/ru/learn/ports-and-kinds/">17 Порты и типы</a></li><li><a href="/ru/learn/latency/">18 Задержка</a></li></ol></li><li><h3>Модуль 7 · Мосты</h3><p>Почему в дизайне появляется больше одной шины и пакетный мост, переносящий работу между ними.</p><ol><li><a href="/ru/learn/why-bridges/">19 Зачем мосты</a></li><li><a href="/ru/learn/the-packet-bridge/">20 Пакетный мост</a></li><li><a href="/ru/learn/everything-in-one-bench/">21 Всё в одном стенде</a></li></ol></li><li><h3>Модуль 8 · Ethernet: кадры и тайминг</h3><p>Кадры, контрольная сумма кадра, тайминг RGMII и заранее зарегистрированные шаги настоящего запуска.</p><ol><li><a href="/ru/learn/frames-and-the-fcs/">22 Кадры и контрольная сумма</a></li><li><a href="/ru/learn/rgmii-timing/">23 Тайминг RGMII</a></li><li><a href="/ru/learn/bring-up-on-the-bench/">24 Запуск на стенде</a></li></ol></li><li><h3>Модуль 9 · Стенд</h3><p>Дисциплина, оберегающая настоящее железо: кто держит IO, как взять и вернуть захват, и что запускается дальше.</p><ol><li><a href="/ru/learn/who-holds-the-io/">25 Кто держит IO</a></li><li><a href="/ru/learn/claim-and-release/">26 Захват и возврат</a></li><li><a href="/ru/learn/capstone-what-runs-next/">27 Капстоун: что запускается дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/structs/index.html b/apps/website/public/ru/learn/structs/index.html
index 541e82edf4..13e9ed5a52 100644
--- a/apps/website/public/ru/learn/structs/index.html
+++ b/apps/website/public/ru/learn/structs/index.html
@@ -91,5 +91,6 @@ <h1>Структуры</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Файл спеки</h3><p>Что лежит в файле .t27: строка module, строки прозы и комментарии.</p><ol><li><a href="/ru/learn/what-a-spec-is/">01 Что такое спека</a></li><li><a href="/ru/learn/the-module-line/">02 Строка module</a></li><li><a href="/ru/learn/comments-and-prose/">03 Комментарии и строки прозы</a></li></ol></li><li><h3>Модуль 2 · Значения и типы</h3><p>Константы, ширина целого числа, а также true, false и текст.</p><ol><li><a href="/ru/learn/constants/">04 Константы</a></li><li><a href="/ru/learn/whole-numbers/">05 Целые числа и их ширина</a></li><li><a href="/ru/learn/true-false-and-text/">06 Истина, ложь и текст</a></li></ol></li><li><h3>Модуль 3 · Массивы, триты и выражения</h3><p>Списки значений, три значения трита и что вычисляют операторы.</p><ol><li><a href="/ru/learn/arrays/">07 Массивы</a></li><li><a href="/ru/learn/trits/">08 Триты: минус один, ноль, один</a></li><li><a href="/ru/learn/expressions/">09 Выражения</a></li></ol></li><li><h3>Модуль 4 · Тесты и функции</h3><p>Блок test, несколько маленьких тестов в одной спеке и функция с типизированными входами.</p><ol><li><a href="/ru/learn/a-test-block/">10 Блок test</a></li><li><a href="/ru/learn/many-tests/">11 Много тестов в одной спеке</a></li><li><a href="/ru/learn/functions/">12 Функции</a></li></ol></li><li><h3>Модуль 5 · Внутри функции</h3><p>Имена через let и var, выбор через if и switch и циклы.</p><ol><li><a href="/ru/learn/local-names/">13 Локальные имена: let и var</a></li><li><a href="/ru/learn/if-and-else/">14 Выбор: if, else и switch</a></li><li><a href="/ru/learn/loops/">15 Циклы: while и for</a></li></ol></li><li><h3>Модуль 6 · Видимость и свои типы</h3><p>Что pub открывает другим модулям, и структуры и перечисления, которые вы объявляете.</p><ol><li><a href="/ru/learn/pub-or-private/">16 pub или закрытое имя</a></li><li class="on" aria-current="page"><a href="/ru/learn/structs/">17 Структуры</a></li><li><a href="/ru/learn/enums/">18 Перечисления</a></li></ol></li><li><h3>Модуль 7 · Модули, правила и gen-ts</h3><p>Чтение другого модуля, правило, которое выполняется всегда, и спека, превращённая в TypeScript.</p><ol><li><a href="/ru/learn/use-other-modules/">19 Другие модули</a></li><li><a href="/ru/learn/invariants/">20 Инварианты и замеры</a></li><li><a href="/ru/learn/gen-ts/">21 gen-ts: спека превращается в TypeScript</a></li></ol></li><li><h3>Модуль 8 · Компилятор и tri</h3><p>Бэкенды, которые пишет t27c, стадии, в которых он читает спеку, и команда tri.</p><ol><li><a href="/ru/learn/seven-backends/">22 Семь бэкендов</a></li><li><a href="/ru/learn/stages-of-t27c/">23 Стадии t27c</a></li><li><a href="/ru/learn/tri-commands/">24 tri: командная строка</a></li></ol></li><li><h3>Модуль 9 · Ошибки, программа и что дальше</h3><p>Как читать ошибку компилятора, одна маленькая законченная программа и какой курс пройти дальше.</p><ol><li><a href="/ru/learn/reading-errors/">25 Как читать ошибки компилятора</a></li><li><a href="/ru/learn/a-small-program/">26 Маленькая законченная программа</a></li><li><a href="/ru/learn/where-next/">27 Куда дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/sum-and-carry/index.html b/apps/website/public/ru/learn/sum-and-carry/index.html
index 3710dc8e3c..a67165e0dd 100644
--- a/apps/website/public/ru/learn/sum-and-carry/index.html
+++ b/apps/website/public/ru/learn/sum-and-carry/index.html
@@ -148,5 +148,6 @@ <h1>Сумма и перенос</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Три значения</h3><p>Почему три, как сбалансированная троичная система пишет любое число без знака и что дают переворот и отсечение тритов.</p><ol><li><a href="/ru/learn/why-three/">01 Почему три</a></li><li><a href="/ru/learn/balanced-ternary/">02 Сбалансированная троичная</a></li><li><a href="/ru/learn/flip-and-cut/">03 Переворот вместо минуса, отсечение вместо округления</a></li></ol></li><li><h3>Модуль 2 · Логика с «не знаю»</h3><p>Трёхзначные вентили Клини, два вентиля без двоичного двойника и сложение как пара таблиц.</p><ol><li><a href="/ru/learn/kleene-logic/">04 Трёхзначная логика</a></li><li><a href="/ru/learn/consensus-and-any/">05 Консенсус и «любой»</a></li><li class="on" aria-current="page"><a href="/ru/learn/sum-and-carry/">06 Сумма и перенос</a></li></ol></li><li><h3>Модуль 3 · Складываем триты</h3><p>Полусумматор, полный сумматор и перенос, бегущий влево разряд за разрядом.</p><ol><li><a href="/ru/learn/half-adder/">07 Полусумматор</a></li><li><a href="/ru/learn/full-adder/">08 Полный сумматор</a></li><li><a href="/ru/learn/ripple-carry/">09 Сквозной перенос</a></li></ol></li><li><h3>Модуль 4 · Умножение без умножителя</h3><p>Копия, ноль или переворот: произведение на один трит, умножение столбиком и MAC, который только складывает.</p><ol><li><a href="/ru/learn/times-a-trit/">10 Умножить на один трит</a></li><li><a href="/ru/learn/long-multiplication/">11 Умножение столбиком</a></li><li><a href="/ru/learn/ternary-mac/">12 Троичный MAC</a></li></ol></li><li><h3>Модуль 5 · Триты в двоичной памяти</h3><p>Два бита на трит, пять тритов на байт и сколько бит нужно слову из 27 тритов.</p><ol><li><a href="/ru/learn/two-bits-per-trit/">13 Два бита на трит</a></li><li><a href="/ru/learn/five-trits-a-byte/">14 Пять тритов в байте</a></li><li><a href="/ru/learn/a-word-of-27/">15 Слово из 27 тритов</a></li></ol></li><li><h3>Модуль 6 · Командное слово TRI-27</h3><p>32-битное слово, которое декодирует эмулятор Trinity: его поля, 47 опкодов и 15-битное непосредственное значение.</p><ol><li><a href="/ru/learn/the-instruction-word/">16 Командное слово</a></li><li><a href="/ru/learn/the-opcode-map/">17 Карта опкодов</a></li><li><a href="/ru/learn/the-immediate/">18 Непосредственное значение</a></li></ol></li><li><h3>Модуль 7 · Троичная машина</h3><p>АЛУ из сумматоров этого курса, переход на три стороны и программа, которую можно пройти по шагам.</p><ol><li><a href="/ru/learn/a-ternary-alu/">19 Троичное АЛУ</a></li><li><a href="/ru/learn/three-way-jump/">20 Переход на три стороны</a></li><li><a href="/ru/learn/run-a-program/">21 Запустить программу</a></li></ol></li><li><h3>Модуль 8 · Три ответа</h3><p>Сравнение по старшему различному триту, поиск числа по третям и сортировка трёхзначными сравнениями.</p><ol><li><a href="/ru/learn/compare-at-the-top/">22 Сравнение сверху</a></li><li><a href="/ru/learn/questions-in-thirds/">23 Вопросы по третям</a></li><li><a href="/ru/learn/sort-by-three/">24 Сортировка трёхзначными сравнениями</a></li></ol></li><li><h3>Модуль 9 · Троичные нейроны</h3><p>Веса -1, 0, +1: один нейрон, детектор и маленький слой, и нигде нет умножителя.</p><ol><li><a href="/ru/learn/one-ternary-neuron/">25 Один троичный нейрон</a></li><li><a href="/ru/learn/a-bar-detector/">26 Детектор полосы</a></li><li><a href="/ru/learn/a-tiny-layer/">27 Маленький слой</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/surviving-mutants/index.html b/apps/website/public/ru/learn/surviving-mutants/index.html
index 9a5c5b9ac7..3ad8dac54c 100644
--- a/apps/website/public/ru/learn/surviving-mutants/index.html
+++ b/apps/website/public/ru/learn/surviving-mutants/index.html
@@ -110,5 +110,6 @@ <h1>Выжившие мутанты</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Зачем проверять</h3><p>Дизайны, которые компилируются и ошибаются; модель, которая выносит вердикт; план, записанный до кода.</p><ol><li><a href="/ru/learn/bugs-that-compile/">01 Баги, которые компилируются</a></li><li><a href="/ru/learn/golden-model/">02 Золотая модель</a></li><li><a href="/ru/learn/plan-before-code/">03 Сначала план</a></li></ol></li><li><h3>Модуль 2 · Тестбенчи</h3><p>Стимулы, проверки и вердикт, записанные как один spec рядом с дизайном, который они судят.</p><ol><li><a href="/ru/learn/anatomy-of-a-tb/">04 Анатомия тестбенча</a></li><li><a href="/ru/learn/self-checking/">05 Самопроверка вместо всматривания</a></li><li><a href="/ru/learn/directed-vs-random/">06 Направленные против случайных</a></li></ol></li><li><h3>Модуль 3 · Временные диаграммы</h3><p>Трасса каждого сигнала, прочитанная так, как её читает инженер по железу, и два прогона, сравнённые между собой.</p><ol><li><a href="/ru/learn/vcd-format/">07 Формат VCD</a></li><li><a href="/ru/learn/debug-with-waves/">08 Отладка по волнам</a></li><li><a href="/ru/learn/compare-traces/">09 Сравнение двух трасс</a></li></ol></li><li><h3>Модуль 4 · Векторы соответствия</h3><p>Случаи с ответом, записанным рядом, там, откуда их достанет компилятор.</p><ol><li><a href="/ru/learn/what-a-vector-is/">10 Что такое вектор</a></li><li><a href="/ru/learn/golden-ruler-vectors/">11 Векторы как эталонная линейка</a></li><li><a href="/ru/learn/end-to-end-scenarios/">12 Сквозные сценарии</a></li></ol></li><li><h3>Модуль 5 · Косимуляция</h3><p>Spec, симулятор и плата сходятся в одном ответе на стенде Artix-7 XC7A200T, и что делать, когда не сходятся.</p><ol><li><a href="/ru/learn/three-way-agreement/">13 Тройное согласие</a></li><li><a href="/ru/learn/simulator-internals/">14 Внутри симулятора</a></li><li><a href="/ru/learn/when-they-disagree/">15 Когда они не сходятся</a></li></ol></li><li><h3>Модуль 6 · Покрытие</h3><p>Чего коснулись тесты: строки, переключения, состояния — и что прячет это число.</p><ol><li><a href="/ru/learn/line-and-toggle/">16 Строки и переключения</a></li><li><a href="/ru/learn/fsm-coverage/">17 Покрытие автоматов</a></li><li><a href="/ru/learn/coverage-lies/">18 Покрытие лжёт</a></li></ol></li><li><h3>Модуль 7 · Формальные методы</h3><p>Ассерты, верные каждый такт; ограниченный поиск контрпримера; и почему доказательству нужна индукция.</p><ol><li><a href="/ru/learn/assertions/">19 Ассерты</a></li><li><a href="/ru/learn/bounded-model-checking/">20 Ограниченная проверка моделей</a></li><li><a href="/ru/learn/induction/">21 Индукция</a></li></ol></li><li><h3>Модуль 8 · Мутации</h3><p>Ломайте дизайн нарочно и считайте, что заметили тесты.</p><ol><li><a href="/ru/learn/kill-the-mutant/">22 Убить мутанта</a></li><li class="on" aria-current="page"><a href="/ru/learn/surviving-mutants/">23 Выжившие мутанты</a></li><li><a href="/ru/learn/test-strength/">24 Сила тестов</a></li></ol></li><li><h3>Модуль 9 · Приёмка</h3><p>Одна команда, все квитанции, чистый вердикт, который можно показать.</p><ol><li><a href="/ru/learn/build-verify/">25 Сборка и проверка</a></li><li><a href="/ru/learn/board-selftest/">26 Стенд проверяет себя</a></li><li><a href="/ru/learn/verify-capstone/">27 Капстоун: проверить модуль целиком</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">Этот курс сам является spec t27: прочитайте его</a></p><p>Виджеты сохраняют свои английские слова: каждый живёт в собственном spec t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/sync-or-async/index.html b/apps/website/public/ru/learn/sync-or-async/index.html
index e735f6fc9a..ffbfbccc4e 100644
--- a/apps/website/public/ru/learn/sync-or-async/index.html
+++ b/apps/website/public/ru/learn/sync-or-async/index.html
@@ -329,5 +329,6 @@ <h1>Синхронный или асинхронный</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое тактовый сигнал</h3><p>Один фронт, один мир: всё, что делит фронт тактового сигнала, делит и мир; период и джиттер настоящего фронта и место, где тактовый сигнал входит в плату.</p><ol><li><a href="/ru/learn/one-edge-one-world/">01 Один фронт, один мир</a></li><li><a href="/ru/learn/period-and-jitter/">02 Период, фронт и джиттер</a></li><li><a href="/ru/learn/where-the-clock-enters/">03 Где входит тактовый сигнал</a></li></ol></li><li><h3>Модуль 2 · Деревья тактового сигнала</h3><p>Перекос и задержка прохождения, глобальная сеть буферов и ловушка стробирования клока логикой.</p><ol><li><a href="/ru/learn/skew-and-insertion/">04 Перекос и задержка прохождения</a></li><li><a href="/ru/learn/global-buffers/">05 Глобальный буфер</a></li><li><a href="/ru/learn/gated-clock-trap/">06 Ловушка стробированного клока</a></li></ol></li><li><h3>Модуль 3 · PLL и MMCM</h3><p>Умножить и поделить один тактовый сигнал в другой, сдвинуть его фазу шагами VCO и какие тактовые сигналы анализатор считает родственными.</p><ol><li><a href="/ru/learn/multiply-and-divide/">07 Умножить и поделить</a></li><li><a href="/ru/learn/phase-and-lock/">08 Фаза и захват</a></li><li><a href="/ru/learn/related-or-not/">09 Родственные или чужие</a></li></ol></li><li><h3>Модуль 4 · Сбросы</h3><p>Ассертировать асинхронно, снимать синхронно: три вида сброса, конвейер снятия и дерево, которое растит сброс.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/sync-or-async/">10 Синхронный или асинхронный</a></li><li><a href="/ru/learn/reset-release/">11 Конвейер снятия</a></li><li><a href="/ru/learn/reset-fanout/">12 Разветвление сброса</a></li></ol></li><li><h3>Модуль 5 · Метастабильность</h3><p>Окно setup-hold, среднее время между отказами в целочисленной арифметике и два триггера, которые это чинят.</p><ol><li><a href="/ru/learn/setup-hold-window/">13 Окно setup-hold</a></li><li><a href="/ru/learn/mtbf/">14 Среднее время между отказами</a></li><li><a href="/ru/learn/two-flop-sync/">15 Двухтриггерный синхронизатор</a></li></ol></li><li><h3>Модуль 6 · Пересечение многих бит</h3><p>Почему двоичная шина рвётся, почему код Грея нет, и рукопожатие, переносящее импульс между мирами.</p><ol><li><a href="/ru/learn/why-binary-fails/">16 Почему двоичный код рвётся</a></li><li><a href="/ru/learn/gray-code/">17 Код Грея</a></li><li><a href="/ru/learn/handshakes-and-pulses/">18 Рукопожатия и импульсы</a></li></ol></li><li><h3>Модуль 7 · Асинхронный FIFO</h3><p>Указатели, флаги и глубина: буфер, переносящий поток между двумя тактовыми сигналами.</p><ol><li><a href="/ru/learn/pointers-and-flags/">19 Указатели и флаги</a></li><li><a href="/ru/learn/depth-sizing/">20 Выбор глубины</a></li><li><a href="/ru/learn/fifo-testbench/">21 Тестбенч FIFO</a></li></ol></li><li><h3>Модуль 8 · Ограничения</h3><p>Строки, сообщающие анализатору, что такое тактовый сигнал, какие пути не проверять и чему должны соответствовать выводы.</p><ol><li><a href="/ru/learn/create-clock/">22 Строка create_clock</a></li><li><a href="/ru/learn/false-path-and-max-delay/">23 False path и max delay</a></li><li><a href="/ru/learn/io-timing/">24 Тайминг выводов</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Отчёт CDC, одно пересечение, захваченное на триггерах, и дифф битстрима, замыкающий курс.</p><ol><li><a href="/ru/learn/cdc-report/">25 Отчёт CDC</a></li><li><a href="/ru/learn/capture-a-crossing/">26 Захватить пересечение</a></li><li><a href="/ru/learn/capstone-diff/">27 Капстоун: поменяй и сравни</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/synthesis-warnings/index.html b/apps/website/public/ru/learn/synthesis-warnings/index.html
index 63b0dd4778..7b74910e81 100644
--- a/apps/website/public/ru/learn/synthesis-warnings/index.html
+++ b/apps/website/public/ru/learn/synthesis-warnings/index.html
@@ -195,5 +195,6 @@ <h1>Важные предупреждения синтеза</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Чип</h3><p>Что такое FPGA, какой чип у нас и как его выводы встречаются с платой.</p><ol><li><a href="/ru/learn/fpga-is-a-grid/">01 FPGA — это сетка, которую вы настраиваете</a></li><li><a href="/ru/learn/which-chip/">02 Какой это чип?</a></li><li><a href="/ru/learn/pins-and-package/">03 Выводы, шарики и корпус</a></li></ol></li><li><h3>Модуль 2 · Числа в железе</h3><p>Биты, триты и форматы чисел, и сколько стоит арифметика в логике.</p><ol><li><a href="/ru/learn/bits-and-trits/">04 Биты, триты и GF16</a></li><li><a href="/ru/learn/what-arithmetic-costs/">05 Сколько стоит арифметика</a></li><li><a href="/ru/learn/how-much-fits/">06 Сколько помещается на чип</a></li></ol></li><li><h3>Модуль 3 · Ваша программа на t27</h3><p>Напишите спеку, проверьте её и увидьте, почему «компилируется» не значит «верно».</p><ol><li><a href="/ru/learn/hello-t27/">07 Привет, t27</a></li><li><a href="/ru/learn/tests-are-the-spec/">08 Тесты — часть спеки</a></li><li><a href="/ru/learn/compiles-is-not-correct/">09 Компилируется — не значит верно</a></li></ol></li><li><h3>Модуль 4 · Внутри t27c</h3><p>Как компилятор читает спеку и что он пишет, включая нативный t27b.</p><ol><li><a href="/ru/learn/compiler-stages/">10 Стадии t27c</a></li><li><a href="/ru/learn/t27c-on-a-real-machine/">11 t27c на настоящей машине</a></li><li><a href="/ru/learn/seven-backends-and-t27b/">12 Семь бэкендов и t27b</a></li></ol></li><li><h3>Модуль 5 · От спеки к железу</h3><p>Verilog, который пишет t27c, ячейки, в которые он превращается, и что делает одна LUT.</p><ol><li><a href="/ru/learn/spec-to-verilog/">13 От спеки к Verilog</a></li><li><a href="/ru/learn/count-the-cells/">14 Посчитайте ячейки</a></li><li><a href="/ru/learn/one-lut-up-close/">15 Одна LUT вблизи</a></li></ol></li><li><h3>Модуль 6 · Читаем синтез</h3><p>Что yosys сообщает о вашем дизайне и какие предупреждения важны.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/synthesis-warnings/">16 Важные предупреждения синтеза</a></li><li><a href="/ru/learn/state-machines/">17 Конечные автоматы</a></li><li><a href="/ru/learn/where-the-luts-go/">18 Куда уходят LUT</a></li></ol></li><li><h3>Модуль 7 · Размещение, трассировка, тайминг</h3><p>Где ячейки оказываются на кристалле и успевает ли тактовая частота.</p><ol><li><a href="/ru/learn/placement/">19 Размещение: где оказываются ячейки</a></li><li><a href="/ru/learn/logic-depth/">20 Глубина логики</a></li><li><a href="/ru/learn/timing-slack/">21 Тайминг и запас</a></li></ol></li><li><h3>Модуль 8 · Битстрим</h3><p>Как оттрассированный дизайн становится битами, которые загружает чип, без Vivado.</p><ol><li><a href="/ru/learn/fasm-to-frames/">22 От FASM к фреймам</a></li><li><a href="/ru/learn/a-build-receipt/">23 Сборка по пунктам</a></li><li><a href="/ru/learn/read-a-bitstream/">24 Читаем битстрим</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Спросите чип, кто он, загрузите биты и проверьте их.</p><ol><li><a href="/ru/learn/jtag-asks-the-chip/">25 JTAG спрашивает чип</a></li><li><a href="/ru/learn/same-bits-on-the-board/">26 Те же биты — на плате</a></li><li><a href="/ru/learn/capstone/">27 Итог: переугадайте yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/t27-number-format/index.html b/apps/website/public/ru/learn/t27-number-format/index.html
index 964df67c06..842ad09f37 100644
--- a/apps/website/public/ru/learn/t27-number-format/index.html
+++ b/apps/website/public/ru/learn/t27-number-format/index.html
@@ -112,5 +112,6 @@ <h1>T27: наш собственный формат чисел</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Лаборатория: наши исследования</h3><p>Свой формат чисел, честная таблица результатов и таблицы модели, перемноженные на плате.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/t27-number-format/">01 T27: наш собственный формат чисел</a></li><li><a href="/ru/learn/an-honest-scoreboard/">02 Честная таблица результатов</a></li><li><a href="/ru/learn/a-model-on-the-board/">03 Таблицы модели на плате</a></li></ol></li><li><h3>Модуль 2 · ИИ-числа: блок MX</h3><p>Как ИИ-чипы хранят веса в нескольких битах: один общий масштаб на блок, сам байт масштаба и что один выброс делает с соседями.</p><ol><li><a href="/ru/learn/one-scale-per-block/">04 Один масштаб на блок весов</a></li><li><a href="/ru/learn/scale-byte-native/">05 Байт масштаба на настоящей машине</a></li><li><a href="/ru/learn/one-outlier/">06 Один выброс, много нулей</a></li></ol></li><li><h3>Модуль 3 · Тернарные веса</h3><p>Веса, которые бывают только минус масштаб, ноль или плюс масштаб, пять правил, которые должен пройти тернарный алфавит, и прогон тестов, который ничего не проверил.</p><ol><li><a href="/ru/learn/phi-scaled-ternary/">07 Веса минус фи, ноль и плюс фи</a></li><li><a href="/ru/learn/a-pass-that-checked-nothing/">08 Прогон, который ничего не проверил</a></li><li><a href="/ru/learn/five-rules-for-weights/">09 Пять правил для алфавита весов</a></li></ol></li><li><h3>Модуль 4 · Тернарный формат Ternary Network Float</h3><p>Правило, которое компилятор проверяет до запуска любого теста, и 17-битное число с плавающей точкой, порядок которого — четыре сбалансированных трита.</p><ol><li><a href="/ru/learn/blocked-is-not-failed/">10 Заблокирован — не значит провален</a></li><li><a href="/ru/learn/four-trit-exponent/">11 Число, у которого порядок — четыре трита</a></li><li><a href="/ru/learn/negate-twice/">12 Дважды сменить знак — получить число обратно</a></li></ol></li><li><h3>Модуль 5 · Арифметика чисел со знаком</h3><p>Умножить два числа со знаком, сложить их, когда знаки разные, и сделать то и другое сразу в умножении с накоплением.</p><ol><li><a href="/ru/learn/a-sign-is-one-bit/">13 Знак — это один бит</a></li><li><a href="/ru/learn/opposite-signs-subtract/">14 Разные знаки — вычитание</a></li><li><a href="/ru/learn/multiply-then-add/">15 Умножить, потом сложить</a></li></ol></li><li><h3>Модуль 6 · Части нейрона</h3><p>ReLU с изломом в нуле, степень двойки для softmax и argmax, который называет ответ.</p><ol><li><a href="/ru/learn/a-bend-at-zero/">16 Излом в нуле</a></li><li><a href="/ru/learn/two-to-the-x/">17 Два в степени x</a></li><li><a href="/ru/learn/pick-the-largest/">18 Выбрать наибольшее</a></li></ol></li><li><h3>Модуль 7 · Обучение на ошибке</h3><p>Потеря, которая оценивает неверную догадку в битах, один шаг, который сдвигает вес против градиента, и скрытый слой, который нужен XOR.</p><ol><li><a href="/ru/learn/a-loss-in-bits/">19 Потеря в битах</a></li><li><a href="/ru/learn/one-step-downhill/">20 Один шаг вниз по склону</a></li><li><a href="/ru/learn/xor-needs-a-bend/">21 XOR нужен излом</a></li></ol></li><li><h3>Модуль 8 · BitNet: тернарные сети</h3><p>Порог, который сжимает сумму обратно до трёх значений, один нейрон, который становится другой функцией при смене весов, и нейрон, который читает входы по 27 тритов за раз.</p><ol><li><a href="/ru/learn/back-to-three-values/">22 Обратно в три значения</a></li><li><a href="/ru/learn/same-neuron-new-weights/">23 Тот же нейрон, другие веса</a></li><li><a href="/ru/learn/a-neuron-in-chunks/">24 Нейрон по кускам</a></li></ol></li><li><h3>Модуль 9 · Тернарный MAC как чип</h3><p>Скалярное произведение 27 тритов как провода без регистра, та же сумма, которую регистр копит на каждом такте, и небольшая целая сеть в конце курса.</p><ol><li><a href="/ru/learn/a-dot-product-in-wires/">25 Скалярное произведение из проводов</a></li><li><a href="/ru/learn/add-it-up-every-clock/">26 Складывать на каждом такте</a></li><li><a href="/ru/learn/a-whole-network/">27 Целая сеть</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/t27c-on-a-real-machine/index.html b/apps/website/public/ru/learn/t27c-on-a-real-machine/index.html
index 07bd9119e3..fa7a73feec 100644
--- a/apps/website/public/ru/learn/t27c-on-a-real-machine/index.html
+++ b/apps/website/public/ru/learn/t27c-on-a-real-machine/index.html
@@ -171,5 +171,6 @@ <h1>t27c на настоящей машине</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Чип</h3><p>Что такое FPGA, какой чип у нас и как его выводы встречаются с платой.</p><ol><li><a href="/ru/learn/fpga-is-a-grid/">01 FPGA — это сетка, которую вы настраиваете</a></li><li><a href="/ru/learn/which-chip/">02 Какой это чип?</a></li><li><a href="/ru/learn/pins-and-package/">03 Выводы, шарики и корпус</a></li></ol></li><li><h3>Модуль 2 · Числа в железе</h3><p>Биты, триты и форматы чисел, и сколько стоит арифметика в логике.</p><ol><li><a href="/ru/learn/bits-and-trits/">04 Биты, триты и GF16</a></li><li><a href="/ru/learn/what-arithmetic-costs/">05 Сколько стоит арифметика</a></li><li><a href="/ru/learn/how-much-fits/">06 Сколько помещается на чип</a></li></ol></li><li><h3>Модуль 3 · Ваша программа на t27</h3><p>Напишите спеку, проверьте её и увидьте, почему «компилируется» не значит «верно».</p><ol><li><a href="/ru/learn/hello-t27/">07 Привет, t27</a></li><li><a href="/ru/learn/tests-are-the-spec/">08 Тесты — часть спеки</a></li><li><a href="/ru/learn/compiles-is-not-correct/">09 Компилируется — не значит верно</a></li></ol></li><li><h3>Модуль 4 · Внутри t27c</h3><p>Как компилятор читает спеку и что он пишет, включая нативный t27b.</p><ol><li><a href="/ru/learn/compiler-stages/">10 Стадии t27c</a></li><li class="on" aria-current="page"><a href="/ru/learn/t27c-on-a-real-machine/">11 t27c на настоящей машине</a></li><li><a href="/ru/learn/seven-backends-and-t27b/">12 Семь бэкендов и t27b</a></li></ol></li><li><h3>Модуль 5 · От спеки к железу</h3><p>Verilog, который пишет t27c, ячейки, в которые он превращается, и что делает одна LUT.</p><ol><li><a href="/ru/learn/spec-to-verilog/">13 От спеки к Verilog</a></li><li><a href="/ru/learn/count-the-cells/">14 Посчитайте ячейки</a></li><li><a href="/ru/learn/one-lut-up-close/">15 Одна LUT вблизи</a></li></ol></li><li><h3>Модуль 6 · Читаем синтез</h3><p>Что yosys сообщает о вашем дизайне и какие предупреждения важны.</p><ol><li><a href="/ru/learn/synthesis-warnings/">16 Важные предупреждения синтеза</a></li><li><a href="/ru/learn/state-machines/">17 Конечные автоматы</a></li><li><a href="/ru/learn/where-the-luts-go/">18 Куда уходят LUT</a></li></ol></li><li><h3>Модуль 7 · Размещение, трассировка, тайминг</h3><p>Где ячейки оказываются на кристалле и успевает ли тактовая частота.</p><ol><li><a href="/ru/learn/placement/">19 Размещение: где оказываются ячейки</a></li><li><a href="/ru/learn/logic-depth/">20 Глубина логики</a></li><li><a href="/ru/learn/timing-slack/">21 Тайминг и запас</a></li></ol></li><li><h3>Модуль 8 · Битстрим</h3><p>Как оттрассированный дизайн становится битами, которые загружает чип, без Vivado.</p><ol><li><a href="/ru/learn/fasm-to-frames/">22 От FASM к фреймам</a></li><li><a href="/ru/learn/a-build-receipt/">23 Сборка по пунктам</a></li><li><a href="/ru/learn/read-a-bitstream/">24 Читаем битстрим</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Спросите чип, кто он, загрузите биты и проверьте их.</p><ol><li><a href="/ru/learn/jtag-asks-the-chip/">25 JTAG спрашивает чип</a></li><li><a href="/ru/learn/same-bits-on-the-board/">26 Те же биты — на плате</a></li><li><a href="/ru/learn/capstone/">27 Итог: переугадайте yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/tables-and-interpolation/index.html b/apps/website/public/ru/learn/tables-and-interpolation/index.html
index c1c8d8fa51..06ba56d713 100644
--- a/apps/website/public/ru/learn/tables-and-interpolation/index.html
+++ b/apps/website/public/ru/learn/tables-and-interpolation/index.html
@@ -439,5 +439,6 @@ <h1>Таблицы и интерполяция</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Сложение</h3><p>Почему сложение медленное: перенос, который идёт вверх по слову, выделенная цепь в каждом слайсе, префиксные сети и деревья сохранения переноса.</p><ol><li><a href="/ru/learn/the-carry-chain/">01 Цепь переноса</a></li><li><a href="/ru/learn/prefix-adders/">02 Взгляд вперёд: префиксные сумматоры</a></li><li><a href="/ru/learn/carry-save/">03 Три числа сразу</a></li></ol></li><li><h3>Модуль 2 · Умножение</h3><p>Произведение — сумма сдвинутых копий; перекодирование Бута вдвое сокращает их, а слайс DSP48E1 делает остальное одним блоком.</p><ol><li><a href="/ru/learn/shift-and-add/">04 Сдвиг и сложение</a></li><li><a href="/ru/learn/booths-trick/">05 Приём Бута</a></li><li><a href="/ru/learn/the-dsp48e1-slice/">06 Слайс DSP48E1</a></li></ol></li><li><h3>Модуль 3 · Фиксированная точка</h3><p>Где стоит двоичная точка, что округление делает со значением и с его средним и что даёт каждый бит квантователя.</p><ol><li><a href="/ru/learn/q-formats/">07 Форматы Q</a></li><li><a href="/ru/learn/rounding-and-saturation/">08 Округление и насыщение</a></li><li><a href="/ru/learn/quantization-noise/">09 Шум квантования</a></li></ol></li><li><h3>Модуль 4 · Функции в железе</h3><p>Синус, косинус, угол и длина из сдвигов и сложений — и когда таблица оказывается выгоднее.</p><ol><li><a href="/ru/learn/rotating-by-shifts/">10 Поворот сдвигами</a></li><li><a href="/ru/learn/angle-and-magnitude/">11 Угол и модуль</a></li><li class="on" aria-current="page"><a href="/ru/learn/tables-and-interpolation/">12 Таблицы и интерполяция</a></li></ol></li><li><h3>Модуль 5 · Сигналы и дискретизация</h3><p>Что дискретизация делает с частотой, как генератор строится из сумматора и что измеряет бин ДПФ.</p><ol><li><a href="/ru/learn/sampling-and-aliasing/">13 Дискретизация и наложение спектров</a></li><li><a href="/ru/learn/the-nco/">14 Генератор с числовым управлением</a></li><li><a href="/ru/learn/bins-of-a-spectrum/">15 Бины спектра</a></li></ol></li><li><h3>Модуль 6 · Фильтры FIR</h3><p>Скользящее среднее, расчёт окном sinc с целыми коэффициентами и раскладка отводов по слайсам DSP.</p><ol><li><a href="/ru/learn/the-moving-average/">16 Скользящее среднее</a></li><li><a href="/ru/learn/designing-taps/">17 Расчёт коэффициентов</a></li><li><a href="/ru/learn/folding-taps-onto-slices/">18 Раскладка отводов по слайсам</a></li></ol></li><li><h3>Модуль 7 · Многоскоростная обработка</h3><p>Понижение частоты дискретизации без заворота шума: децимация, фильтр CIC и полифазная форма.</p><ol><li><a href="/ru/learn/lowering-the-rate/">19 Понижение частоты</a></li><li><a href="/ru/learn/the-cic-filter/">20 Фильтр CIC</a></li><li><a href="/ru/learn/polyphase/">21 Полифазная форма</a></li></ol></li><li><h3>Модуль 8 · БПФ</h3><p>N log N вместо N^2: бабочки, бит-реверсный порядок входа и биты, которые добавляет каждый этап.</p><ol><li><a href="/ru/learn/the-butterfly/">22 Бабочка</a></li><li><a href="/ru/learn/twiddles-and-bit-reversal/">23 Поворачивающие множители и бит-реверс</a></li><li><a href="/ru/learn/growth-through-the-stages/">24 Рост через этапы</a></li></ol></li><li><h3>Модуль 9 · На стенде</h3><p>От арифметики к плате: коррелятор из спеки модема, бюджет слайсов и тайминга и итоговый фильтр.</p><ol><li><a href="/ru/learn/a-correlator-finds-the-frame/">25 Коррелятор находит кадр</a></li><li><a href="/ru/learn/does-it-fit-and-close-timing/">26 Помещается ли и сходится ли тайминг?</a></li><li><a href="/ru/learn/capstone-a-filter-on-the-board/">27 Итог: фильтр на плате</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/ternary-computing/index.html b/apps/website/public/ru/learn/ternary-computing/index.html
index 7649be1eb2..f0d7a75aca 100644
--- a/apps/website/public/ru/learn/ternary-computing/index.html
+++ b/apps/website/public/ru/learn/ternary-computing/index.html
@@ -67,5 +67,6 @@ <h1>Троичная машина: от трита до работающей п
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Три значения</h3><p>Почему три, как сбалансированная троичная система пишет любое число без знака и что дают переворот и отсечение тритов.</p><ol><li><a href="/ru/learn/why-three/">01 Почему три</a></li><li><a href="/ru/learn/balanced-ternary/">02 Сбалансированная троичная</a></li><li><a href="/ru/learn/flip-and-cut/">03 Переворот вместо минуса, отсечение вместо округления</a></li></ol></li><li><h3>Модуль 2 · Логика с «не знаю»</h3><p>Трёхзначные вентили Клини, два вентиля без двоичного двойника и сложение как пара таблиц.</p><ol><li><a href="/ru/learn/kleene-logic/">04 Трёхзначная логика</a></li><li><a href="/ru/learn/consensus-and-any/">05 Консенсус и «любой»</a></li><li><a href="/ru/learn/sum-and-carry/">06 Сумма и перенос</a></li></ol></li><li><h3>Модуль 3 · Складываем триты</h3><p>Полусумматор, полный сумматор и перенос, бегущий влево разряд за разрядом.</p><ol><li><a href="/ru/learn/half-adder/">07 Полусумматор</a></li><li><a href="/ru/learn/full-adder/">08 Полный сумматор</a></li><li><a href="/ru/learn/ripple-carry/">09 Сквозной перенос</a></li></ol></li><li><h3>Модуль 4 · Умножение без умножителя</h3><p>Копия, ноль или переворот: произведение на один трит, умножение столбиком и MAC, который только складывает.</p><ol><li><a href="/ru/learn/times-a-trit/">10 Умножить на один трит</a></li><li><a href="/ru/learn/long-multiplication/">11 Умножение столбиком</a></li><li><a href="/ru/learn/ternary-mac/">12 Троичный MAC</a></li></ol></li><li><h3>Модуль 5 · Триты в двоичной памяти</h3><p>Два бита на трит, пять тритов на байт и сколько бит нужно слову из 27 тритов.</p><ol><li><a href="/ru/learn/two-bits-per-trit/">13 Два бита на трит</a></li><li><a href="/ru/learn/five-trits-a-byte/">14 Пять тритов в байте</a></li><li><a href="/ru/learn/a-word-of-27/">15 Слово из 27 тритов</a></li></ol></li><li><h3>Модуль 6 · Командное слово TRI-27</h3><p>32-битное слово, которое декодирует эмулятор Trinity: его поля, 47 опкодов и 15-битное непосредственное значение.</p><ol><li><a href="/ru/learn/the-instruction-word/">16 Командное слово</a></li><li><a href="/ru/learn/the-opcode-map/">17 Карта опкодов</a></li><li><a href="/ru/learn/the-immediate/">18 Непосредственное значение</a></li></ol></li><li><h3>Модуль 7 · Троичная машина</h3><p>АЛУ из сумматоров этого курса, переход на три стороны и программа, которую можно пройти по шагам.</p><ol><li><a href="/ru/learn/a-ternary-alu/">19 Троичное АЛУ</a></li><li><a href="/ru/learn/three-way-jump/">20 Переход на три стороны</a></li><li><a href="/ru/learn/run-a-program/">21 Запустить программу</a></li></ol></li><li><h3>Модуль 8 · Три ответа</h3><p>Сравнение по старшему различному триту, поиск числа по третям и сортировка трёхзначными сравнениями.</p><ol><li><a href="/ru/learn/compare-at-the-top/">22 Сравнение сверху</a></li><li><a href="/ru/learn/questions-in-thirds/">23 Вопросы по третям</a></li><li><a href="/ru/learn/sort-by-three/">24 Сортировка трёхзначными сравнениями</a></li></ol></li><li><h3>Модуль 9 · Троичные нейроны</h3><p>Веса -1, 0, +1: один нейрон, детектор и маленький слой, и нигде нет умножителя.</p><ol><li><a href="/ru/learn/one-ternary-neuron/">25 Один троичный нейрон</a></li><li><a href="/ru/learn/a-bar-detector/">26 Детектор полосы</a></li><li><a href="/ru/learn/a-tiny-layer/">27 Маленький слой</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/ternary-mac/index.html b/apps/website/public/ru/learn/ternary-mac/index.html
index cdea5fb826..14bbff49cc 100644
--- a/apps/website/public/ru/learn/ternary-mac/index.html
+++ b/apps/website/public/ru/learn/ternary-mac/index.html
@@ -115,5 +115,6 @@ <h1>Троичный MAC</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Три значения</h3><p>Почему три, как сбалансированная троичная система пишет любое число без знака и что дают переворот и отсечение тритов.</p><ol><li><a href="/ru/learn/why-three/">01 Почему три</a></li><li><a href="/ru/learn/balanced-ternary/">02 Сбалансированная троичная</a></li><li><a href="/ru/learn/flip-and-cut/">03 Переворот вместо минуса, отсечение вместо округления</a></li></ol></li><li><h3>Модуль 2 · Логика с «не знаю»</h3><p>Трёхзначные вентили Клини, два вентиля без двоичного двойника и сложение как пара таблиц.</p><ol><li><a href="/ru/learn/kleene-logic/">04 Трёхзначная логика</a></li><li><a href="/ru/learn/consensus-and-any/">05 Консенсус и «любой»</a></li><li><a href="/ru/learn/sum-and-carry/">06 Сумма и перенос</a></li></ol></li><li><h3>Модуль 3 · Складываем триты</h3><p>Полусумматор, полный сумматор и перенос, бегущий влево разряд за разрядом.</p><ol><li><a href="/ru/learn/half-adder/">07 Полусумматор</a></li><li><a href="/ru/learn/full-adder/">08 Полный сумматор</a></li><li><a href="/ru/learn/ripple-carry/">09 Сквозной перенос</a></li></ol></li><li><h3>Модуль 4 · Умножение без умножителя</h3><p>Копия, ноль или переворот: произведение на один трит, умножение столбиком и MAC, который только складывает.</p><ol><li><a href="/ru/learn/times-a-trit/">10 Умножить на один трит</a></li><li><a href="/ru/learn/long-multiplication/">11 Умножение столбиком</a></li><li class="on" aria-current="page"><a href="/ru/learn/ternary-mac/">12 Троичный MAC</a></li></ol></li><li><h3>Модуль 5 · Триты в двоичной памяти</h3><p>Два бита на трит, пять тритов на байт и сколько бит нужно слову из 27 тритов.</p><ol><li><a href="/ru/learn/two-bits-per-trit/">13 Два бита на трит</a></li><li><a href="/ru/learn/five-trits-a-byte/">14 Пять тритов в байте</a></li><li><a href="/ru/learn/a-word-of-27/">15 Слово из 27 тритов</a></li></ol></li><li><h3>Модуль 6 · Командное слово TRI-27</h3><p>32-битное слово, которое декодирует эмулятор Trinity: его поля, 47 опкодов и 15-битное непосредственное значение.</p><ol><li><a href="/ru/learn/the-instruction-word/">16 Командное слово</a></li><li><a href="/ru/learn/the-opcode-map/">17 Карта опкодов</a></li><li><a href="/ru/learn/the-immediate/">18 Непосредственное значение</a></li></ol></li><li><h3>Модуль 7 · Троичная машина</h3><p>АЛУ из сумматоров этого курса, переход на три стороны и программа, которую можно пройти по шагам.</p><ol><li><a href="/ru/learn/a-ternary-alu/">19 Троичное АЛУ</a></li><li><a href="/ru/learn/three-way-jump/">20 Переход на три стороны</a></li><li><a href="/ru/learn/run-a-program/">21 Запустить программу</a></li></ol></li><li><h3>Модуль 8 · Три ответа</h3><p>Сравнение по старшему различному триту, поиск числа по третям и сортировка трёхзначными сравнениями.</p><ol><li><a href="/ru/learn/compare-at-the-top/">22 Сравнение сверху</a></li><li><a href="/ru/learn/questions-in-thirds/">23 Вопросы по третям</a></li><li><a href="/ru/learn/sort-by-three/">24 Сортировка трёхзначными сравнениями</a></li></ol></li><li><h3>Модуль 9 · Троичные нейроны</h3><p>Веса -1, 0, +1: один нейрон, детектор и маленький слой, и нигде нет умножителя.</p><ol><li><a href="/ru/learn/one-ternary-neuron/">25 Один троичный нейрон</a></li><li><a href="/ru/learn/a-bar-detector/">26 Детектор полосы</a></li><li><a href="/ru/learn/a-tiny-layer/">27 Маленький слой</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/test-strength/index.html b/apps/website/public/ru/learn/test-strength/index.html
index acd6b4c60c..c171573cc4 100644
--- a/apps/website/public/ru/learn/test-strength/index.html
+++ b/apps/website/public/ru/learn/test-strength/index.html
@@ -110,5 +110,6 @@ <h1>Сила тестов</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Зачем проверять</h3><p>Дизайны, которые компилируются и ошибаются; модель, которая выносит вердикт; план, записанный до кода.</p><ol><li><a href="/ru/learn/bugs-that-compile/">01 Баги, которые компилируются</a></li><li><a href="/ru/learn/golden-model/">02 Золотая модель</a></li><li><a href="/ru/learn/plan-before-code/">03 Сначала план</a></li></ol></li><li><h3>Модуль 2 · Тестбенчи</h3><p>Стимулы, проверки и вердикт, записанные как один spec рядом с дизайном, который они судят.</p><ol><li><a href="/ru/learn/anatomy-of-a-tb/">04 Анатомия тестбенча</a></li><li><a href="/ru/learn/self-checking/">05 Самопроверка вместо всматривания</a></li><li><a href="/ru/learn/directed-vs-random/">06 Направленные против случайных</a></li></ol></li><li><h3>Модуль 3 · Временные диаграммы</h3><p>Трасса каждого сигнала, прочитанная так, как её читает инженер по железу, и два прогона, сравнённые между собой.</p><ol><li><a href="/ru/learn/vcd-format/">07 Формат VCD</a></li><li><a href="/ru/learn/debug-with-waves/">08 Отладка по волнам</a></li><li><a href="/ru/learn/compare-traces/">09 Сравнение двух трасс</a></li></ol></li><li><h3>Модуль 4 · Векторы соответствия</h3><p>Случаи с ответом, записанным рядом, там, откуда их достанет компилятор.</p><ol><li><a href="/ru/learn/what-a-vector-is/">10 Что такое вектор</a></li><li><a href="/ru/learn/golden-ruler-vectors/">11 Векторы как эталонная линейка</a></li><li><a href="/ru/learn/end-to-end-scenarios/">12 Сквозные сценарии</a></li></ol></li><li><h3>Модуль 5 · Косимуляция</h3><p>Spec, симулятор и плата сходятся в одном ответе на стенде Artix-7 XC7A200T, и что делать, когда не сходятся.</p><ol><li><a href="/ru/learn/three-way-agreement/">13 Тройное согласие</a></li><li><a href="/ru/learn/simulator-internals/">14 Внутри симулятора</a></li><li><a href="/ru/learn/when-they-disagree/">15 Когда они не сходятся</a></li></ol></li><li><h3>Модуль 6 · Покрытие</h3><p>Чего коснулись тесты: строки, переключения, состояния — и что прячет это число.</p><ol><li><a href="/ru/learn/line-and-toggle/">16 Строки и переключения</a></li><li><a href="/ru/learn/fsm-coverage/">17 Покрытие автоматов</a></li><li><a href="/ru/learn/coverage-lies/">18 Покрытие лжёт</a></li></ol></li><li><h3>Модуль 7 · Формальные методы</h3><p>Ассерты, верные каждый такт; ограниченный поиск контрпримера; и почему доказательству нужна индукция.</p><ol><li><a href="/ru/learn/assertions/">19 Ассерты</a></li><li><a href="/ru/learn/bounded-model-checking/">20 Ограниченная проверка моделей</a></li><li><a href="/ru/learn/induction/">21 Индукция</a></li></ol></li><li><h3>Модуль 8 · Мутации</h3><p>Ломайте дизайн нарочно и считайте, что заметили тесты.</p><ol><li><a href="/ru/learn/kill-the-mutant/">22 Убить мутанта</a></li><li><a href="/ru/learn/surviving-mutants/">23 Выжившие мутанты</a></li><li class="on" aria-current="page"><a href="/ru/learn/test-strength/">24 Сила тестов</a></li></ol></li><li><h3>Модуль 9 · Приёмка</h3><p>Одна команда, все квитанции, чистый вердикт, который можно показать.</p><ol><li><a href="/ru/learn/build-verify/">25 Сборка и проверка</a></li><li><a href="/ru/learn/board-selftest/">26 Стенд проверяет себя</a></li><li><a href="/ru/learn/verify-capstone/">27 Капстоун: проверить модуль целиком</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">Этот курс сам является spec t27: прочитайте его</a></p><p>Виджеты сохраняют свои английские слова: каждый живёт в собственном spec t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/tests-are-the-spec/index.html b/apps/website/public/ru/learn/tests-are-the-spec/index.html
index 06c1e62b57..7aee829642 100644
--- a/apps/website/public/ru/learn/tests-are-the-spec/index.html
+++ b/apps/website/public/ru/learn/tests-are-the-spec/index.html
@@ -156,5 +156,6 @@ <h1>Тесты — часть спеки</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Чип</h3><p>Что такое FPGA, какой чип у нас и как его выводы встречаются с платой.</p><ol><li><a href="/ru/learn/fpga-is-a-grid/">01 FPGA — это сетка, которую вы настраиваете</a></li><li><a href="/ru/learn/which-chip/">02 Какой это чип?</a></li><li><a href="/ru/learn/pins-and-package/">03 Выводы, шарики и корпус</a></li></ol></li><li><h3>Модуль 2 · Числа в железе</h3><p>Биты, триты и форматы чисел, и сколько стоит арифметика в логике.</p><ol><li><a href="/ru/learn/bits-and-trits/">04 Биты, триты и GF16</a></li><li><a href="/ru/learn/what-arithmetic-costs/">05 Сколько стоит арифметика</a></li><li><a href="/ru/learn/how-much-fits/">06 Сколько помещается на чип</a></li></ol></li><li><h3>Модуль 3 · Ваша программа на t27</h3><p>Напишите спеку, проверьте её и увидьте, почему «компилируется» не значит «верно».</p><ol><li><a href="/ru/learn/hello-t27/">07 Привет, t27</a></li><li class="on" aria-current="page"><a href="/ru/learn/tests-are-the-spec/">08 Тесты — часть спеки</a></li><li><a href="/ru/learn/compiles-is-not-correct/">09 Компилируется — не значит верно</a></li></ol></li><li><h3>Модуль 4 · Внутри t27c</h3><p>Как компилятор читает спеку и что он пишет, включая нативный t27b.</p><ol><li><a href="/ru/learn/compiler-stages/">10 Стадии t27c</a></li><li><a href="/ru/learn/t27c-on-a-real-machine/">11 t27c на настоящей машине</a></li><li><a href="/ru/learn/seven-backends-and-t27b/">12 Семь бэкендов и t27b</a></li></ol></li><li><h3>Модуль 5 · От спеки к железу</h3><p>Verilog, который пишет t27c, ячейки, в которые он превращается, и что делает одна LUT.</p><ol><li><a href="/ru/learn/spec-to-verilog/">13 От спеки к Verilog</a></li><li><a href="/ru/learn/count-the-cells/">14 Посчитайте ячейки</a></li><li><a href="/ru/learn/one-lut-up-close/">15 Одна LUT вблизи</a></li></ol></li><li><h3>Модуль 6 · Читаем синтез</h3><p>Что yosys сообщает о вашем дизайне и какие предупреждения важны.</p><ol><li><a href="/ru/learn/synthesis-warnings/">16 Важные предупреждения синтеза</a></li><li><a href="/ru/learn/state-machines/">17 Конечные автоматы</a></li><li><a href="/ru/learn/where-the-luts-go/">18 Куда уходят LUT</a></li></ol></li><li><h3>Модуль 7 · Размещение, трассировка, тайминг</h3><p>Где ячейки оказываются на кристалле и успевает ли тактовая частота.</p><ol><li><a href="/ru/learn/placement/">19 Размещение: где оказываются ячейки</a></li><li><a href="/ru/learn/logic-depth/">20 Глубина логики</a></li><li><a href="/ru/learn/timing-slack/">21 Тайминг и запас</a></li></ol></li><li><h3>Модуль 8 · Битстрим</h3><p>Как оттрассированный дизайн становится битами, которые загружает чип, без Vivado.</p><ol><li><a href="/ru/learn/fasm-to-frames/">22 От FASM к фреймам</a></li><li><a href="/ru/learn/a-build-receipt/">23 Сборка по пунктам</a></li><li><a href="/ru/learn/read-a-bitstream/">24 Читаем битстрим</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Спросите чип, кто он, загрузите биты и проверьте их.</p><ol><li><a href="/ru/learn/jtag-asks-the-chip/">25 JTAG спрашивает чип</a></li><li><a href="/ru/learn/same-bits-on-the-board/">26 Те же биты — на плате</a></li><li><a href="/ru/learn/capstone/">27 Итог: переугадайте yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/the-butterfly/index.html b/apps/website/public/ru/learn/the-butterfly/index.html
index a49b8b8d88..ac3bd0c469 100644
--- a/apps/website/public/ru/learn/the-butterfly/index.html
+++ b/apps/website/public/ru/learn/the-butterfly/index.html
@@ -1122,5 +1122,6 @@ <h1>Бабочка</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Сложение</h3><p>Почему сложение медленное: перенос, который идёт вверх по слову, выделенная цепь в каждом слайсе, префиксные сети и деревья сохранения переноса.</p><ol><li><a href="/ru/learn/the-carry-chain/">01 Цепь переноса</a></li><li><a href="/ru/learn/prefix-adders/">02 Взгляд вперёд: префиксные сумматоры</a></li><li><a href="/ru/learn/carry-save/">03 Три числа сразу</a></li></ol></li><li><h3>Модуль 2 · Умножение</h3><p>Произведение — сумма сдвинутых копий; перекодирование Бута вдвое сокращает их, а слайс DSP48E1 делает остальное одним блоком.</p><ol><li><a href="/ru/learn/shift-and-add/">04 Сдвиг и сложение</a></li><li><a href="/ru/learn/booths-trick/">05 Приём Бута</a></li><li><a href="/ru/learn/the-dsp48e1-slice/">06 Слайс DSP48E1</a></li></ol></li><li><h3>Модуль 3 · Фиксированная точка</h3><p>Где стоит двоичная точка, что округление делает со значением и с его средним и что даёт каждый бит квантователя.</p><ol><li><a href="/ru/learn/q-formats/">07 Форматы Q</a></li><li><a href="/ru/learn/rounding-and-saturation/">08 Округление и насыщение</a></li><li><a href="/ru/learn/quantization-noise/">09 Шум квантования</a></li></ol></li><li><h3>Модуль 4 · Функции в железе</h3><p>Синус, косинус, угол и длина из сдвигов и сложений — и когда таблица оказывается выгоднее.</p><ol><li><a href="/ru/learn/rotating-by-shifts/">10 Поворот сдвигами</a></li><li><a href="/ru/learn/angle-and-magnitude/">11 Угол и модуль</a></li><li><a href="/ru/learn/tables-and-interpolation/">12 Таблицы и интерполяция</a></li></ol></li><li><h3>Модуль 5 · Сигналы и дискретизация</h3><p>Что дискретизация делает с частотой, как генератор строится из сумматора и что измеряет бин ДПФ.</p><ol><li><a href="/ru/learn/sampling-and-aliasing/">13 Дискретизация и наложение спектров</a></li><li><a href="/ru/learn/the-nco/">14 Генератор с числовым управлением</a></li><li><a href="/ru/learn/bins-of-a-spectrum/">15 Бины спектра</a></li></ol></li><li><h3>Модуль 6 · Фильтры FIR</h3><p>Скользящее среднее, расчёт окном sinc с целыми коэффициентами и раскладка отводов по слайсам DSP.</p><ol><li><a href="/ru/learn/the-moving-average/">16 Скользящее среднее</a></li><li><a href="/ru/learn/designing-taps/">17 Расчёт коэффициентов</a></li><li><a href="/ru/learn/folding-taps-onto-slices/">18 Раскладка отводов по слайсам</a></li></ol></li><li><h3>Модуль 7 · Многоскоростная обработка</h3><p>Понижение частоты дискретизации без заворота шума: децимация, фильтр CIC и полифазная форма.</p><ol><li><a href="/ru/learn/lowering-the-rate/">19 Понижение частоты</a></li><li><a href="/ru/learn/the-cic-filter/">20 Фильтр CIC</a></li><li><a href="/ru/learn/polyphase/">21 Полифазная форма</a></li></ol></li><li><h3>Модуль 8 · БПФ</h3><p>N log N вместо N^2: бабочки, бит-реверсный порядок входа и биты, которые добавляет каждый этап.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/the-butterfly/">22 Бабочка</a></li><li><a href="/ru/learn/twiddles-and-bit-reversal/">23 Поворачивающие множители и бит-реверс</a></li><li><a href="/ru/learn/growth-through-the-stages/">24 Рост через этапы</a></li></ol></li><li><h3>Модуль 9 · На стенде</h3><p>От арифметики к плате: коррелятор из спеки модема, бюджет слайсов и тайминга и итоговый фильтр.</p><ol><li><a href="/ru/learn/a-correlator-finds-the-frame/">25 Коррелятор находит кадр</a></li><li><a href="/ru/learn/does-it-fit-and-close-timing/">26 Помещается ли и сходится ли тайминг?</a></li><li><a href="/ru/learn/capstone-a-filter-on-the-board/">27 Итог: фильтр на плате</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/the-carry-chain/index.html b/apps/website/public/ru/learn/the-carry-chain/index.html
index 78d5ebeca1..28216ec9fc 100644
--- a/apps/website/public/ru/learn/the-carry-chain/index.html
+++ b/apps/website/public/ru/learn/the-carry-chain/index.html
@@ -554,5 +554,6 @@ <h1>Цепь переноса</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Сложение</h3><p>Почему сложение медленное: перенос, который идёт вверх по слову, выделенная цепь в каждом слайсе, префиксные сети и деревья сохранения переноса.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/the-carry-chain/">01 Цепь переноса</a></li><li><a href="/ru/learn/prefix-adders/">02 Взгляд вперёд: префиксные сумматоры</a></li><li><a href="/ru/learn/carry-save/">03 Три числа сразу</a></li></ol></li><li><h3>Модуль 2 · Умножение</h3><p>Произведение — сумма сдвинутых копий; перекодирование Бута вдвое сокращает их, а слайс DSP48E1 делает остальное одним блоком.</p><ol><li><a href="/ru/learn/shift-and-add/">04 Сдвиг и сложение</a></li><li><a href="/ru/learn/booths-trick/">05 Приём Бута</a></li><li><a href="/ru/learn/the-dsp48e1-slice/">06 Слайс DSP48E1</a></li></ol></li><li><h3>Модуль 3 · Фиксированная точка</h3><p>Где стоит двоичная точка, что округление делает со значением и с его средним и что даёт каждый бит квантователя.</p><ol><li><a href="/ru/learn/q-formats/">07 Форматы Q</a></li><li><a href="/ru/learn/rounding-and-saturation/">08 Округление и насыщение</a></li><li><a href="/ru/learn/quantization-noise/">09 Шум квантования</a></li></ol></li><li><h3>Модуль 4 · Функции в железе</h3><p>Синус, косинус, угол и длина из сдвигов и сложений — и когда таблица оказывается выгоднее.</p><ol><li><a href="/ru/learn/rotating-by-shifts/">10 Поворот сдвигами</a></li><li><a href="/ru/learn/angle-and-magnitude/">11 Угол и модуль</a></li><li><a href="/ru/learn/tables-and-interpolation/">12 Таблицы и интерполяция</a></li></ol></li><li><h3>Модуль 5 · Сигналы и дискретизация</h3><p>Что дискретизация делает с частотой, как генератор строится из сумматора и что измеряет бин ДПФ.</p><ol><li><a href="/ru/learn/sampling-and-aliasing/">13 Дискретизация и наложение спектров</a></li><li><a href="/ru/learn/the-nco/">14 Генератор с числовым управлением</a></li><li><a href="/ru/learn/bins-of-a-spectrum/">15 Бины спектра</a></li></ol></li><li><h3>Модуль 6 · Фильтры FIR</h3><p>Скользящее среднее, расчёт окном sinc с целыми коэффициентами и раскладка отводов по слайсам DSP.</p><ol><li><a href="/ru/learn/the-moving-average/">16 Скользящее среднее</a></li><li><a href="/ru/learn/designing-taps/">17 Расчёт коэффициентов</a></li><li><a href="/ru/learn/folding-taps-onto-slices/">18 Раскладка отводов по слайсам</a></li></ol></li><li><h3>Модуль 7 · Многоскоростная обработка</h3><p>Понижение частоты дискретизации без заворота шума: децимация, фильтр CIC и полифазная форма.</p><ol><li><a href="/ru/learn/lowering-the-rate/">19 Понижение частоты</a></li><li><a href="/ru/learn/the-cic-filter/">20 Фильтр CIC</a></li><li><a href="/ru/learn/polyphase/">21 Полифазная форма</a></li></ol></li><li><h3>Модуль 8 · БПФ</h3><p>N log N вместо N^2: бабочки, бит-реверсный порядок входа и биты, которые добавляет каждый этап.</p><ol><li><a href="/ru/learn/the-butterfly/">22 Бабочка</a></li><li><a href="/ru/learn/twiddles-and-bit-reversal/">23 Поворачивающие множители и бит-реверс</a></li><li><a href="/ru/learn/growth-through-the-stages/">24 Рост через этапы</a></li></ol></li><li><h3>Модуль 9 · На стенде</h3><p>От арифметики к плате: коррелятор из спеки модема, бюджет слайсов и тайминга и итоговый фильтр.</p><ol><li><a href="/ru/learn/a-correlator-finds-the-frame/">25 Коррелятор находит кадр</a></li><li><a href="/ru/learn/does-it-fit-and-close-timing/">26 Помещается ли и сходится ли тайминг?</a></li><li><a href="/ru/learn/capstone-a-filter-on-the-board/">27 Итог: фильтр на плате</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/the-cic-filter/index.html b/apps/website/public/ru/learn/the-cic-filter/index.html
index 5889938a46..ced926931f 100644
--- a/apps/website/public/ru/learn/the-cic-filter/index.html
+++ b/apps/website/public/ru/learn/the-cic-filter/index.html
@@ -1122,5 +1122,6 @@ <h1>Фильтр CIC</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Сложение</h3><p>Почему сложение медленное: перенос, который идёт вверх по слову, выделенная цепь в каждом слайсе, префиксные сети и деревья сохранения переноса.</p><ol><li><a href="/ru/learn/the-carry-chain/">01 Цепь переноса</a></li><li><a href="/ru/learn/prefix-adders/">02 Взгляд вперёд: префиксные сумматоры</a></li><li><a href="/ru/learn/carry-save/">03 Три числа сразу</a></li></ol></li><li><h3>Модуль 2 · Умножение</h3><p>Произведение — сумма сдвинутых копий; перекодирование Бута вдвое сокращает их, а слайс DSP48E1 делает остальное одним блоком.</p><ol><li><a href="/ru/learn/shift-and-add/">04 Сдвиг и сложение</a></li><li><a href="/ru/learn/booths-trick/">05 Приём Бута</a></li><li><a href="/ru/learn/the-dsp48e1-slice/">06 Слайс DSP48E1</a></li></ol></li><li><h3>Модуль 3 · Фиксированная точка</h3><p>Где стоит двоичная точка, что округление делает со значением и с его средним и что даёт каждый бит квантователя.</p><ol><li><a href="/ru/learn/q-formats/">07 Форматы Q</a></li><li><a href="/ru/learn/rounding-and-saturation/">08 Округление и насыщение</a></li><li><a href="/ru/learn/quantization-noise/">09 Шум квантования</a></li></ol></li><li><h3>Модуль 4 · Функции в железе</h3><p>Синус, косинус, угол и длина из сдвигов и сложений — и когда таблица оказывается выгоднее.</p><ol><li><a href="/ru/learn/rotating-by-shifts/">10 Поворот сдвигами</a></li><li><a href="/ru/learn/angle-and-magnitude/">11 Угол и модуль</a></li><li><a href="/ru/learn/tables-and-interpolation/">12 Таблицы и интерполяция</a></li></ol></li><li><h3>Модуль 5 · Сигналы и дискретизация</h3><p>Что дискретизация делает с частотой, как генератор строится из сумматора и что измеряет бин ДПФ.</p><ol><li><a href="/ru/learn/sampling-and-aliasing/">13 Дискретизация и наложение спектров</a></li><li><a href="/ru/learn/the-nco/">14 Генератор с числовым управлением</a></li><li><a href="/ru/learn/bins-of-a-spectrum/">15 Бины спектра</a></li></ol></li><li><h3>Модуль 6 · Фильтры FIR</h3><p>Скользящее среднее, расчёт окном sinc с целыми коэффициентами и раскладка отводов по слайсам DSP.</p><ol><li><a href="/ru/learn/the-moving-average/">16 Скользящее среднее</a></li><li><a href="/ru/learn/designing-taps/">17 Расчёт коэффициентов</a></li><li><a href="/ru/learn/folding-taps-onto-slices/">18 Раскладка отводов по слайсам</a></li></ol></li><li><h3>Модуль 7 · Многоскоростная обработка</h3><p>Понижение частоты дискретизации без заворота шума: децимация, фильтр CIC и полифазная форма.</p><ol><li><a href="/ru/learn/lowering-the-rate/">19 Понижение частоты</a></li><li class="on" aria-current="page"><a href="/ru/learn/the-cic-filter/">20 Фильтр CIC</a></li><li><a href="/ru/learn/polyphase/">21 Полифазная форма</a></li></ol></li><li><h3>Модуль 8 · БПФ</h3><p>N log N вместо N^2: бабочки, бит-реверсный порядок входа и биты, которые добавляет каждый этап.</p><ol><li><a href="/ru/learn/the-butterfly/">22 Бабочка</a></li><li><a href="/ru/learn/twiddles-and-bit-reversal/">23 Поворачивающие множители и бит-реверс</a></li><li><a href="/ru/learn/growth-through-the-stages/">24 Рост через этапы</a></li></ol></li><li><h3>Модуль 9 · На стенде</h3><p>От арифметики к плате: коррелятор из спеки модема, бюджет слайсов и тайминга и итоговый фильтр.</p><ol><li><a href="/ru/learn/a-correlator-finds-the-frame/">25 Коррелятор находит кадр</a></li><li><a href="/ru/learn/does-it-fit-and-close-timing/">26 Помещается ли и сходится ли тайминг?</a></li><li><a href="/ru/learn/capstone-a-filter-on-the-board/">27 Итог: фильтр на плате</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/the-dsp48e1-slice/index.html b/apps/website/public/ru/learn/the-dsp48e1-slice/index.html
index 6d3acb4663..e441b0c0ff 100644
--- a/apps/website/public/ru/learn/the-dsp48e1-slice/index.html
+++ b/apps/website/public/ru/learn/the-dsp48e1-slice/index.html
@@ -372,5 +372,6 @@ <h1>Слайс DSP48E1</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Сложение</h3><p>Почему сложение медленное: перенос, который идёт вверх по слову, выделенная цепь в каждом слайсе, префиксные сети и деревья сохранения переноса.</p><ol><li><a href="/ru/learn/the-carry-chain/">01 Цепь переноса</a></li><li><a href="/ru/learn/prefix-adders/">02 Взгляд вперёд: префиксные сумматоры</a></li><li><a href="/ru/learn/carry-save/">03 Три числа сразу</a></li></ol></li><li><h3>Модуль 2 · Умножение</h3><p>Произведение — сумма сдвинутых копий; перекодирование Бута вдвое сокращает их, а слайс DSP48E1 делает остальное одним блоком.</p><ol><li><a href="/ru/learn/shift-and-add/">04 Сдвиг и сложение</a></li><li><a href="/ru/learn/booths-trick/">05 Приём Бута</a></li><li class="on" aria-current="page"><a href="/ru/learn/the-dsp48e1-slice/">06 Слайс DSP48E1</a></li></ol></li><li><h3>Модуль 3 · Фиксированная точка</h3><p>Где стоит двоичная точка, что округление делает со значением и с его средним и что даёт каждый бит квантователя.</p><ol><li><a href="/ru/learn/q-formats/">07 Форматы Q</a></li><li><a href="/ru/learn/rounding-and-saturation/">08 Округление и насыщение</a></li><li><a href="/ru/learn/quantization-noise/">09 Шум квантования</a></li></ol></li><li><h3>Модуль 4 · Функции в железе</h3><p>Синус, косинус, угол и длина из сдвигов и сложений — и когда таблица оказывается выгоднее.</p><ol><li><a href="/ru/learn/rotating-by-shifts/">10 Поворот сдвигами</a></li><li><a href="/ru/learn/angle-and-magnitude/">11 Угол и модуль</a></li><li><a href="/ru/learn/tables-and-interpolation/">12 Таблицы и интерполяция</a></li></ol></li><li><h3>Модуль 5 · Сигналы и дискретизация</h3><p>Что дискретизация делает с частотой, как генератор строится из сумматора и что измеряет бин ДПФ.</p><ol><li><a href="/ru/learn/sampling-and-aliasing/">13 Дискретизация и наложение спектров</a></li><li><a href="/ru/learn/the-nco/">14 Генератор с числовым управлением</a></li><li><a href="/ru/learn/bins-of-a-spectrum/">15 Бины спектра</a></li></ol></li><li><h3>Модуль 6 · Фильтры FIR</h3><p>Скользящее среднее, расчёт окном sinc с целыми коэффициентами и раскладка отводов по слайсам DSP.</p><ol><li><a href="/ru/learn/the-moving-average/">16 Скользящее среднее</a></li><li><a href="/ru/learn/designing-taps/">17 Расчёт коэффициентов</a></li><li><a href="/ru/learn/folding-taps-onto-slices/">18 Раскладка отводов по слайсам</a></li></ol></li><li><h3>Модуль 7 · Многоскоростная обработка</h3><p>Понижение частоты дискретизации без заворота шума: децимация, фильтр CIC и полифазная форма.</p><ol><li><a href="/ru/learn/lowering-the-rate/">19 Понижение частоты</a></li><li><a href="/ru/learn/the-cic-filter/">20 Фильтр CIC</a></li><li><a href="/ru/learn/polyphase/">21 Полифазная форма</a></li></ol></li><li><h3>Модуль 8 · БПФ</h3><p>N log N вместо N^2: бабочки, бит-реверсный порядок входа и биты, которые добавляет каждый этап.</p><ol><li><a href="/ru/learn/the-butterfly/">22 Бабочка</a></li><li><a href="/ru/learn/twiddles-and-bit-reversal/">23 Поворачивающие множители и бит-реверс</a></li><li><a href="/ru/learn/growth-through-the-stages/">24 Рост через этапы</a></li></ol></li><li><h3>Модуль 9 · На стенде</h3><p>От арифметики к плате: коррелятор из спеки модема, бюджет слайсов и тайминга и итоговый фильтр.</p><ol><li><a href="/ru/learn/a-correlator-finds-the-frame/">25 Коррелятор находит кадр</a></li><li><a href="/ru/learn/does-it-fit-and-close-timing/">26 Помещается ли и сходится ли тайминг?</a></li><li><a href="/ru/learn/capstone-a-filter-on-the-board/">27 Итог: фильтр на плате</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/the-frame/index.html b/apps/website/public/ru/learn/the-frame/index.html
index 3acb45824c..66f291210a 100644
--- a/apps/website/public/ru/learn/the-frame/index.html
+++ b/apps/website/public/ru/learn/the-frame/index.html
@@ -299,5 +299,6 @@ <h1>Кадр: старт, данные, стоп</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое шина</h3><p>Зачем вообще существует шина: разговор по проводам, с кадрами и адресами, и кому позволено говорить.</p><ol><li><a href="/ru/learn/wires-that-talk/">01 Провода, которые говорят</a></li><li><a href="/ru/learn/framing-a-conversation/">02 Кадрирование разговора</a></li><li><a href="/ru/learn/master-and-servant/">03 Хозяин и слуга</a></li></ol></li><li><h3>Модуль 2 · Разговор по UART</h3><p>Двухпроводная шина без такта: кадр, делитель, задающий скорость, и статус, который опрашивает драйвер.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/the-frame/">04 Кадр: старт, данные, стоп</a></li><li><a href="/ru/learn/baud-and-the-divisor/">05 Бод и делитель</a></li><li><a href="/ru/learn/status-and-fifos/">06 Статус и FIFO</a></li></ol></li><li><h3>Модуль 3 · SPI под тактом</h3><p>Разговор под тактом: четыре режима, лестница предделителей и выбор кристалла на каждого подчинённого.</p><ol><li><a href="/ru/learn/modes-cpol-cpha/">07 Режимы: CPOL и CPHA</a></li><li><a href="/ru/learn/the-prescaler-ladder/">08 Лестница предделителей</a></li><li><a href="/ru/learn/chip-select-and-widths/">09 Выбор кристалла и ширины</a></li></ol></li><li><h3>Модуль 4 · Регистровая шина APB</h3><p>Шина регистров: PSEL и PENABLE, строобы и ожидания, и сколько адресных бит стоит количество периферии.</p><ol><li><a href="/ru/learn/psel-and-penable/">10 PSEL и PENABLE</a></li><li><a href="/ru/learn/strobes-and-waits/">11 Строобы и ожидания</a></li><li><a href="/ru/learn/address-decode/">12 Дешифрация адреса</a></li></ol></li><li><h3>Модуль 5 · Пять каналов AXI4</h3><p>Пять каналов: адрес, данные и отклик в обе стороны, lite или full, пакеты и идентификаторы.</p><ol><li><a href="/ru/learn/five-channels/">13 Пять каналов</a></li><li><a href="/ru/learn/lite-or-full/">14 Lite или full</a></li><li><a href="/ru/learn/bursts-and-ids/">15 Пакеты и идентификаторы</a></li></ol></li><li><h3>Модуль 6 · Память</h3><p>Что находится на дальнем конце каждой шины: карты памяти, типы портов и задержка, которую должно покрывать ожидание.</p><ol><li><a href="/ru/learn/memory-maps/">16 Карты памяти</a></li><li><a href="/ru/learn/ports-and-kinds/">17 Порты и типы</a></li><li><a href="/ru/learn/latency/">18 Задержка</a></li></ol></li><li><h3>Модуль 7 · Мосты</h3><p>Почему в дизайне появляется больше одной шины и пакетный мост, переносящий работу между ними.</p><ol><li><a href="/ru/learn/why-bridges/">19 Зачем мосты</a></li><li><a href="/ru/learn/the-packet-bridge/">20 Пакетный мост</a></li><li><a href="/ru/learn/everything-in-one-bench/">21 Всё в одном стенде</a></li></ol></li><li><h3>Модуль 8 · Ethernet: кадры и тайминг</h3><p>Кадры, контрольная сумма кадра, тайминг RGMII и заранее зарегистрированные шаги настоящего запуска.</p><ol><li><a href="/ru/learn/frames-and-the-fcs/">22 Кадры и контрольная сумма</a></li><li><a href="/ru/learn/rgmii-timing/">23 Тайминг RGMII</a></li><li><a href="/ru/learn/bring-up-on-the-bench/">24 Запуск на стенде</a></li></ol></li><li><h3>Модуль 9 · Стенд</h3><p>Дисциплина, оберегающая настоящее железо: кто держит IO, как взять и вернуть захват, и что запускается дальше.</p><ol><li><a href="/ru/learn/who-holds-the-io/">25 Кто держит IO</a></li><li><a href="/ru/learn/claim-and-release/">26 Захват и возврат</a></li><li><a href="/ru/learn/capstone-what-runs-next/">27 Капстоун: что запускается дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/the-immediate/index.html b/apps/website/public/ru/learn/the-immediate/index.html
index 7e9a39f23a..9c4fb66235 100644
--- a/apps/website/public/ru/learn/the-immediate/index.html
+++ b/apps/website/public/ru/learn/the-immediate/index.html
@@ -764,5 +764,6 @@ <h1>Непосредственное значение</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Три значения</h3><p>Почему три, как сбалансированная троичная система пишет любое число без знака и что дают переворот и отсечение тритов.</p><ol><li><a href="/ru/learn/why-three/">01 Почему три</a></li><li><a href="/ru/learn/balanced-ternary/">02 Сбалансированная троичная</a></li><li><a href="/ru/learn/flip-and-cut/">03 Переворот вместо минуса, отсечение вместо округления</a></li></ol></li><li><h3>Модуль 2 · Логика с «не знаю»</h3><p>Трёхзначные вентили Клини, два вентиля без двоичного двойника и сложение как пара таблиц.</p><ol><li><a href="/ru/learn/kleene-logic/">04 Трёхзначная логика</a></li><li><a href="/ru/learn/consensus-and-any/">05 Консенсус и «любой»</a></li><li><a href="/ru/learn/sum-and-carry/">06 Сумма и перенос</a></li></ol></li><li><h3>Модуль 3 · Складываем триты</h3><p>Полусумматор, полный сумматор и перенос, бегущий влево разряд за разрядом.</p><ol><li><a href="/ru/learn/half-adder/">07 Полусумматор</a></li><li><a href="/ru/learn/full-adder/">08 Полный сумматор</a></li><li><a href="/ru/learn/ripple-carry/">09 Сквозной перенос</a></li></ol></li><li><h3>Модуль 4 · Умножение без умножителя</h3><p>Копия, ноль или переворот: произведение на один трит, умножение столбиком и MAC, который только складывает.</p><ol><li><a href="/ru/learn/times-a-trit/">10 Умножить на один трит</a></li><li><a href="/ru/learn/long-multiplication/">11 Умножение столбиком</a></li><li><a href="/ru/learn/ternary-mac/">12 Троичный MAC</a></li></ol></li><li><h3>Модуль 5 · Триты в двоичной памяти</h3><p>Два бита на трит, пять тритов на байт и сколько бит нужно слову из 27 тритов.</p><ol><li><a href="/ru/learn/two-bits-per-trit/">13 Два бита на трит</a></li><li><a href="/ru/learn/five-trits-a-byte/">14 Пять тритов в байте</a></li><li><a href="/ru/learn/a-word-of-27/">15 Слово из 27 тритов</a></li></ol></li><li><h3>Модуль 6 · Командное слово TRI-27</h3><p>32-битное слово, которое декодирует эмулятор Trinity: его поля, 47 опкодов и 15-битное непосредственное значение.</p><ol><li><a href="/ru/learn/the-instruction-word/">16 Командное слово</a></li><li><a href="/ru/learn/the-opcode-map/">17 Карта опкодов</a></li><li class="on" aria-current="page"><a href="/ru/learn/the-immediate/">18 Непосредственное значение</a></li></ol></li><li><h3>Модуль 7 · Троичная машина</h3><p>АЛУ из сумматоров этого курса, переход на три стороны и программа, которую можно пройти по шагам.</p><ol><li><a href="/ru/learn/a-ternary-alu/">19 Троичное АЛУ</a></li><li><a href="/ru/learn/three-way-jump/">20 Переход на три стороны</a></li><li><a href="/ru/learn/run-a-program/">21 Запустить программу</a></li></ol></li><li><h3>Модуль 8 · Три ответа</h3><p>Сравнение по старшему различному триту, поиск числа по третям и сортировка трёхзначными сравнениями.</p><ol><li><a href="/ru/learn/compare-at-the-top/">22 Сравнение сверху</a></li><li><a href="/ru/learn/questions-in-thirds/">23 Вопросы по третям</a></li><li><a href="/ru/learn/sort-by-three/">24 Сортировка трёхзначными сравнениями</a></li></ol></li><li><h3>Модуль 9 · Троичные нейроны</h3><p>Веса -1, 0, +1: один нейрон, детектор и маленький слой, и нигде нет умножителя.</p><ol><li><a href="/ru/learn/one-ternary-neuron/">25 Один троичный нейрон</a></li><li><a href="/ru/learn/a-bar-detector/">26 Детектор полосы</a></li><li><a href="/ru/learn/a-tiny-layer/">27 Маленький слой</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/the-instruction-word/index.html b/apps/website/public/ru/learn/the-instruction-word/index.html
index 5df83316c9..2cc9f9f6a6 100644
--- a/apps/website/public/ru/learn/the-instruction-word/index.html
+++ b/apps/website/public/ru/learn/the-instruction-word/index.html
@@ -548,5 +548,6 @@ <h1>Командное слово</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Три значения</h3><p>Почему три, как сбалансированная троичная система пишет любое число без знака и что дают переворот и отсечение тритов.</p><ol><li><a href="/ru/learn/why-three/">01 Почему три</a></li><li><a href="/ru/learn/balanced-ternary/">02 Сбалансированная троичная</a></li><li><a href="/ru/learn/flip-and-cut/">03 Переворот вместо минуса, отсечение вместо округления</a></li></ol></li><li><h3>Модуль 2 · Логика с «не знаю»</h3><p>Трёхзначные вентили Клини, два вентиля без двоичного двойника и сложение как пара таблиц.</p><ol><li><a href="/ru/learn/kleene-logic/">04 Трёхзначная логика</a></li><li><a href="/ru/learn/consensus-and-any/">05 Консенсус и «любой»</a></li><li><a href="/ru/learn/sum-and-carry/">06 Сумма и перенос</a></li></ol></li><li><h3>Модуль 3 · Складываем триты</h3><p>Полусумматор, полный сумматор и перенос, бегущий влево разряд за разрядом.</p><ol><li><a href="/ru/learn/half-adder/">07 Полусумматор</a></li><li><a href="/ru/learn/full-adder/">08 Полный сумматор</a></li><li><a href="/ru/learn/ripple-carry/">09 Сквозной перенос</a></li></ol></li><li><h3>Модуль 4 · Умножение без умножителя</h3><p>Копия, ноль или переворот: произведение на один трит, умножение столбиком и MAC, который только складывает.</p><ol><li><a href="/ru/learn/times-a-trit/">10 Умножить на один трит</a></li><li><a href="/ru/learn/long-multiplication/">11 Умножение столбиком</a></li><li><a href="/ru/learn/ternary-mac/">12 Троичный MAC</a></li></ol></li><li><h3>Модуль 5 · Триты в двоичной памяти</h3><p>Два бита на трит, пять тритов на байт и сколько бит нужно слову из 27 тритов.</p><ol><li><a href="/ru/learn/two-bits-per-trit/">13 Два бита на трит</a></li><li><a href="/ru/learn/five-trits-a-byte/">14 Пять тритов в байте</a></li><li><a href="/ru/learn/a-word-of-27/">15 Слово из 27 тритов</a></li></ol></li><li><h3>Модуль 6 · Командное слово TRI-27</h3><p>32-битное слово, которое декодирует эмулятор Trinity: его поля, 47 опкодов и 15-битное непосредственное значение.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/the-instruction-word/">16 Командное слово</a></li><li><a href="/ru/learn/the-opcode-map/">17 Карта опкодов</a></li><li><a href="/ru/learn/the-immediate/">18 Непосредственное значение</a></li></ol></li><li><h3>Модуль 7 · Троичная машина</h3><p>АЛУ из сумматоров этого курса, переход на три стороны и программа, которую можно пройти по шагам.</p><ol><li><a href="/ru/learn/a-ternary-alu/">19 Троичное АЛУ</a></li><li><a href="/ru/learn/three-way-jump/">20 Переход на три стороны</a></li><li><a href="/ru/learn/run-a-program/">21 Запустить программу</a></li></ol></li><li><h3>Модуль 8 · Три ответа</h3><p>Сравнение по старшему различному триту, поиск числа по третям и сортировка трёхзначными сравнениями.</p><ol><li><a href="/ru/learn/compare-at-the-top/">22 Сравнение сверху</a></li><li><a href="/ru/learn/questions-in-thirds/">23 Вопросы по третям</a></li><li><a href="/ru/learn/sort-by-three/">24 Сортировка трёхзначными сравнениями</a></li></ol></li><li><h3>Модуль 9 · Троичные нейроны</h3><p>Веса -1, 0, +1: один нейрон, детектор и маленький слой, и нигде нет умножителя.</p><ol><li><a href="/ru/learn/one-ternary-neuron/">25 Один троичный нейрон</a></li><li><a href="/ru/learn/a-bar-detector/">26 Детектор полосы</a></li><li><a href="/ru/learn/a-tiny-layer/">27 Маленький слой</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/the-module-line/index.html b/apps/website/public/ru/learn/the-module-line/index.html
index 64ed4fcd00..0ba3ac033c 100644
--- a/apps/website/public/ru/learn/the-module-line/index.html
+++ b/apps/website/public/ru/learn/the-module-line/index.html
@@ -81,5 +81,6 @@ <h1>Строка module</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Файл спеки</h3><p>Что лежит в файле .t27: строка module, строки прозы и комментарии.</p><ol><li><a href="/ru/learn/what-a-spec-is/">01 Что такое спека</a></li><li class="on" aria-current="page"><a href="/ru/learn/the-module-line/">02 Строка module</a></li><li><a href="/ru/learn/comments-and-prose/">03 Комментарии и строки прозы</a></li></ol></li><li><h3>Модуль 2 · Значения и типы</h3><p>Константы, ширина целого числа, а также true, false и текст.</p><ol><li><a href="/ru/learn/constants/">04 Константы</a></li><li><a href="/ru/learn/whole-numbers/">05 Целые числа и их ширина</a></li><li><a href="/ru/learn/true-false-and-text/">06 Истина, ложь и текст</a></li></ol></li><li><h3>Модуль 3 · Массивы, триты и выражения</h3><p>Списки значений, три значения трита и что вычисляют операторы.</p><ol><li><a href="/ru/learn/arrays/">07 Массивы</a></li><li><a href="/ru/learn/trits/">08 Триты: минус один, ноль, один</a></li><li><a href="/ru/learn/expressions/">09 Выражения</a></li></ol></li><li><h3>Модуль 4 · Тесты и функции</h3><p>Блок test, несколько маленьких тестов в одной спеке и функция с типизированными входами.</p><ol><li><a href="/ru/learn/a-test-block/">10 Блок test</a></li><li><a href="/ru/learn/many-tests/">11 Много тестов в одной спеке</a></li><li><a href="/ru/learn/functions/">12 Функции</a></li></ol></li><li><h3>Модуль 5 · Внутри функции</h3><p>Имена через let и var, выбор через if и switch и циклы.</p><ol><li><a href="/ru/learn/local-names/">13 Локальные имена: let и var</a></li><li><a href="/ru/learn/if-and-else/">14 Выбор: if, else и switch</a></li><li><a href="/ru/learn/loops/">15 Циклы: while и for</a></li></ol></li><li><h3>Модуль 6 · Видимость и свои типы</h3><p>Что pub открывает другим модулям, и структуры и перечисления, которые вы объявляете.</p><ol><li><a href="/ru/learn/pub-or-private/">16 pub или закрытое имя</a></li><li><a href="/ru/learn/structs/">17 Структуры</a></li><li><a href="/ru/learn/enums/">18 Перечисления</a></li></ol></li><li><h3>Модуль 7 · Модули, правила и gen-ts</h3><p>Чтение другого модуля, правило, которое выполняется всегда, и спека, превращённая в TypeScript.</p><ol><li><a href="/ru/learn/use-other-modules/">19 Другие модули</a></li><li><a href="/ru/learn/invariants/">20 Инварианты и замеры</a></li><li><a href="/ru/learn/gen-ts/">21 gen-ts: спека превращается в TypeScript</a></li></ol></li><li><h3>Модуль 8 · Компилятор и tri</h3><p>Бэкенды, которые пишет t27c, стадии, в которых он читает спеку, и команда tri.</p><ol><li><a href="/ru/learn/seven-backends/">22 Семь бэкендов</a></li><li><a href="/ru/learn/stages-of-t27c/">23 Стадии t27c</a></li><li><a href="/ru/learn/tri-commands/">24 tri: командная строка</a></li></ol></li><li><h3>Модуль 9 · Ошибки, программа и что дальше</h3><p>Как читать ошибку компилятора, одна маленькая законченная программа и какой курс пройти дальше.</p><ol><li><a href="/ru/learn/reading-errors/">25 Как читать ошибки компилятора</a></li><li><a href="/ru/learn/a-small-program/">26 Маленькая законченная программа</a></li><li><a href="/ru/learn/where-next/">27 Куда дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/the-moving-average/index.html b/apps/website/public/ru/learn/the-moving-average/index.html
index 3a6a90157d..3443f40637 100644
--- a/apps/website/public/ru/learn/the-moving-average/index.html
+++ b/apps/website/public/ru/learn/the-moving-average/index.html
@@ -1122,5 +1122,6 @@ <h1>Скользящее среднее</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Сложение</h3><p>Почему сложение медленное: перенос, который идёт вверх по слову, выделенная цепь в каждом слайсе, префиксные сети и деревья сохранения переноса.</p><ol><li><a href="/ru/learn/the-carry-chain/">01 Цепь переноса</a></li><li><a href="/ru/learn/prefix-adders/">02 Взгляд вперёд: префиксные сумматоры</a></li><li><a href="/ru/learn/carry-save/">03 Три числа сразу</a></li></ol></li><li><h3>Модуль 2 · Умножение</h3><p>Произведение — сумма сдвинутых копий; перекодирование Бута вдвое сокращает их, а слайс DSP48E1 делает остальное одним блоком.</p><ol><li><a href="/ru/learn/shift-and-add/">04 Сдвиг и сложение</a></li><li><a href="/ru/learn/booths-trick/">05 Приём Бута</a></li><li><a href="/ru/learn/the-dsp48e1-slice/">06 Слайс DSP48E1</a></li></ol></li><li><h3>Модуль 3 · Фиксированная точка</h3><p>Где стоит двоичная точка, что округление делает со значением и с его средним и что даёт каждый бит квантователя.</p><ol><li><a href="/ru/learn/q-formats/">07 Форматы Q</a></li><li><a href="/ru/learn/rounding-and-saturation/">08 Округление и насыщение</a></li><li><a href="/ru/learn/quantization-noise/">09 Шум квантования</a></li></ol></li><li><h3>Модуль 4 · Функции в железе</h3><p>Синус, косинус, угол и длина из сдвигов и сложений — и когда таблица оказывается выгоднее.</p><ol><li><a href="/ru/learn/rotating-by-shifts/">10 Поворот сдвигами</a></li><li><a href="/ru/learn/angle-and-magnitude/">11 Угол и модуль</a></li><li><a href="/ru/learn/tables-and-interpolation/">12 Таблицы и интерполяция</a></li></ol></li><li><h3>Модуль 5 · Сигналы и дискретизация</h3><p>Что дискретизация делает с частотой, как генератор строится из сумматора и что измеряет бин ДПФ.</p><ol><li><a href="/ru/learn/sampling-and-aliasing/">13 Дискретизация и наложение спектров</a></li><li><a href="/ru/learn/the-nco/">14 Генератор с числовым управлением</a></li><li><a href="/ru/learn/bins-of-a-spectrum/">15 Бины спектра</a></li></ol></li><li><h3>Модуль 6 · Фильтры FIR</h3><p>Скользящее среднее, расчёт окном sinc с целыми коэффициентами и раскладка отводов по слайсам DSP.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/the-moving-average/">16 Скользящее среднее</a></li><li><a href="/ru/learn/designing-taps/">17 Расчёт коэффициентов</a></li><li><a href="/ru/learn/folding-taps-onto-slices/">18 Раскладка отводов по слайсам</a></li></ol></li><li><h3>Модуль 7 · Многоскоростная обработка</h3><p>Понижение частоты дискретизации без заворота шума: децимация, фильтр CIC и полифазная форма.</p><ol><li><a href="/ru/learn/lowering-the-rate/">19 Понижение частоты</a></li><li><a href="/ru/learn/the-cic-filter/">20 Фильтр CIC</a></li><li><a href="/ru/learn/polyphase/">21 Полифазная форма</a></li></ol></li><li><h3>Модуль 8 · БПФ</h3><p>N log N вместо N^2: бабочки, бит-реверсный порядок входа и биты, которые добавляет каждый этап.</p><ol><li><a href="/ru/learn/the-butterfly/">22 Бабочка</a></li><li><a href="/ru/learn/twiddles-and-bit-reversal/">23 Поворачивающие множители и бит-реверс</a></li><li><a href="/ru/learn/growth-through-the-stages/">24 Рост через этапы</a></li></ol></li><li><h3>Модуль 9 · На стенде</h3><p>От арифметики к плате: коррелятор из спеки модема, бюджет слайсов и тайминга и итоговый фильтр.</p><ol><li><a href="/ru/learn/a-correlator-finds-the-frame/">25 Коррелятор находит кадр</a></li><li><a href="/ru/learn/does-it-fit-and-close-timing/">26 Помещается ли и сходится ли тайминг?</a></li><li><a href="/ru/learn/capstone-a-filter-on-the-board/">27 Итог: фильтр на плате</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/the-nco/index.html b/apps/website/public/ru/learn/the-nco/index.html
index 749432d208..d09b9dffda 100644
--- a/apps/website/public/ru/learn/the-nco/index.html
+++ b/apps/website/public/ru/learn/the-nco/index.html
@@ -1122,5 +1122,6 @@ <h1>Генератор с числовым управлением</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Сложение</h3><p>Почему сложение медленное: перенос, который идёт вверх по слову, выделенная цепь в каждом слайсе, префиксные сети и деревья сохранения переноса.</p><ol><li><a href="/ru/learn/the-carry-chain/">01 Цепь переноса</a></li><li><a href="/ru/learn/prefix-adders/">02 Взгляд вперёд: префиксные сумматоры</a></li><li><a href="/ru/learn/carry-save/">03 Три числа сразу</a></li></ol></li><li><h3>Модуль 2 · Умножение</h3><p>Произведение — сумма сдвинутых копий; перекодирование Бута вдвое сокращает их, а слайс DSP48E1 делает остальное одним блоком.</p><ol><li><a href="/ru/learn/shift-and-add/">04 Сдвиг и сложение</a></li><li><a href="/ru/learn/booths-trick/">05 Приём Бута</a></li><li><a href="/ru/learn/the-dsp48e1-slice/">06 Слайс DSP48E1</a></li></ol></li><li><h3>Модуль 3 · Фиксированная точка</h3><p>Где стоит двоичная точка, что округление делает со значением и с его средним и что даёт каждый бит квантователя.</p><ol><li><a href="/ru/learn/q-formats/">07 Форматы Q</a></li><li><a href="/ru/learn/rounding-and-saturation/">08 Округление и насыщение</a></li><li><a href="/ru/learn/quantization-noise/">09 Шум квантования</a></li></ol></li><li><h3>Модуль 4 · Функции в железе</h3><p>Синус, косинус, угол и длина из сдвигов и сложений — и когда таблица оказывается выгоднее.</p><ol><li><a href="/ru/learn/rotating-by-shifts/">10 Поворот сдвигами</a></li><li><a href="/ru/learn/angle-and-magnitude/">11 Угол и модуль</a></li><li><a href="/ru/learn/tables-and-interpolation/">12 Таблицы и интерполяция</a></li></ol></li><li><h3>Модуль 5 · Сигналы и дискретизация</h3><p>Что дискретизация делает с частотой, как генератор строится из сумматора и что измеряет бин ДПФ.</p><ol><li><a href="/ru/learn/sampling-and-aliasing/">13 Дискретизация и наложение спектров</a></li><li class="on" aria-current="page"><a href="/ru/learn/the-nco/">14 Генератор с числовым управлением</a></li><li><a href="/ru/learn/bins-of-a-spectrum/">15 Бины спектра</a></li></ol></li><li><h3>Модуль 6 · Фильтры FIR</h3><p>Скользящее среднее, расчёт окном sinc с целыми коэффициентами и раскладка отводов по слайсам DSP.</p><ol><li><a href="/ru/learn/the-moving-average/">16 Скользящее среднее</a></li><li><a href="/ru/learn/designing-taps/">17 Расчёт коэффициентов</a></li><li><a href="/ru/learn/folding-taps-onto-slices/">18 Раскладка отводов по слайсам</a></li></ol></li><li><h3>Модуль 7 · Многоскоростная обработка</h3><p>Понижение частоты дискретизации без заворота шума: децимация, фильтр CIC и полифазная форма.</p><ol><li><a href="/ru/learn/lowering-the-rate/">19 Понижение частоты</a></li><li><a href="/ru/learn/the-cic-filter/">20 Фильтр CIC</a></li><li><a href="/ru/learn/polyphase/">21 Полифазная форма</a></li></ol></li><li><h3>Модуль 8 · БПФ</h3><p>N log N вместо N^2: бабочки, бит-реверсный порядок входа и биты, которые добавляет каждый этап.</p><ol><li><a href="/ru/learn/the-butterfly/">22 Бабочка</a></li><li><a href="/ru/learn/twiddles-and-bit-reversal/">23 Поворачивающие множители и бит-реверс</a></li><li><a href="/ru/learn/growth-through-the-stages/">24 Рост через этапы</a></li></ol></li><li><h3>Модуль 9 · На стенде</h3><p>От арифметики к плате: коррелятор из спеки модема, бюджет слайсов и тайминга и итоговый фильтр.</p><ol><li><a href="/ru/learn/a-correlator-finds-the-frame/">25 Коррелятор находит кадр</a></li><li><a href="/ru/learn/does-it-fit-and-close-timing/">26 Помещается ли и сходится ли тайминг?</a></li><li><a href="/ru/learn/capstone-a-filter-on-the-board/">27 Итог: фильтр на плате</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/the-opcode-map/index.html b/apps/website/public/ru/learn/the-opcode-map/index.html
index 4d087e5292..4c4e8d56d1 100644
--- a/apps/website/public/ru/learn/the-opcode-map/index.html
+++ b/apps/website/public/ru/learn/the-opcode-map/index.html
@@ -304,5 +304,6 @@ <h1>Карта опкодов</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Три значения</h3><p>Почему три, как сбалансированная троичная система пишет любое число без знака и что дают переворот и отсечение тритов.</p><ol><li><a href="/ru/learn/why-three/">01 Почему три</a></li><li><a href="/ru/learn/balanced-ternary/">02 Сбалансированная троичная</a></li><li><a href="/ru/learn/flip-and-cut/">03 Переворот вместо минуса, отсечение вместо округления</a></li></ol></li><li><h3>Модуль 2 · Логика с «не знаю»</h3><p>Трёхзначные вентили Клини, два вентиля без двоичного двойника и сложение как пара таблиц.</p><ol><li><a href="/ru/learn/kleene-logic/">04 Трёхзначная логика</a></li><li><a href="/ru/learn/consensus-and-any/">05 Консенсус и «любой»</a></li><li><a href="/ru/learn/sum-and-carry/">06 Сумма и перенос</a></li></ol></li><li><h3>Модуль 3 · Складываем триты</h3><p>Полусумматор, полный сумматор и перенос, бегущий влево разряд за разрядом.</p><ol><li><a href="/ru/learn/half-adder/">07 Полусумматор</a></li><li><a href="/ru/learn/full-adder/">08 Полный сумматор</a></li><li><a href="/ru/learn/ripple-carry/">09 Сквозной перенос</a></li></ol></li><li><h3>Модуль 4 · Умножение без умножителя</h3><p>Копия, ноль или переворот: произведение на один трит, умножение столбиком и MAC, который только складывает.</p><ol><li><a href="/ru/learn/times-a-trit/">10 Умножить на один трит</a></li><li><a href="/ru/learn/long-multiplication/">11 Умножение столбиком</a></li><li><a href="/ru/learn/ternary-mac/">12 Троичный MAC</a></li></ol></li><li><h3>Модуль 5 · Триты в двоичной памяти</h3><p>Два бита на трит, пять тритов на байт и сколько бит нужно слову из 27 тритов.</p><ol><li><a href="/ru/learn/two-bits-per-trit/">13 Два бита на трит</a></li><li><a href="/ru/learn/five-trits-a-byte/">14 Пять тритов в байте</a></li><li><a href="/ru/learn/a-word-of-27/">15 Слово из 27 тритов</a></li></ol></li><li><h3>Модуль 6 · Командное слово TRI-27</h3><p>32-битное слово, которое декодирует эмулятор Trinity: его поля, 47 опкодов и 15-битное непосредственное значение.</p><ol><li><a href="/ru/learn/the-instruction-word/">16 Командное слово</a></li><li class="on" aria-current="page"><a href="/ru/learn/the-opcode-map/">17 Карта опкодов</a></li><li><a href="/ru/learn/the-immediate/">18 Непосредственное значение</a></li></ol></li><li><h3>Модуль 7 · Троичная машина</h3><p>АЛУ из сумматоров этого курса, переход на три стороны и программа, которую можно пройти по шагам.</p><ol><li><a href="/ru/learn/a-ternary-alu/">19 Троичное АЛУ</a></li><li><a href="/ru/learn/three-way-jump/">20 Переход на три стороны</a></li><li><a href="/ru/learn/run-a-program/">21 Запустить программу</a></li></ol></li><li><h3>Модуль 8 · Три ответа</h3><p>Сравнение по старшему различному триту, поиск числа по третям и сортировка трёхзначными сравнениями.</p><ol><li><a href="/ru/learn/compare-at-the-top/">22 Сравнение сверху</a></li><li><a href="/ru/learn/questions-in-thirds/">23 Вопросы по третям</a></li><li><a href="/ru/learn/sort-by-three/">24 Сортировка трёхзначными сравнениями</a></li></ol></li><li><h3>Модуль 9 · Троичные нейроны</h3><p>Веса -1, 0, +1: один нейрон, детектор и маленький слой, и нигде нет умножителя.</p><ol><li><a href="/ru/learn/one-ternary-neuron/">25 Один троичный нейрон</a></li><li><a href="/ru/learn/a-bar-detector/">26 Детектор полосы</a></li><li><a href="/ru/learn/a-tiny-layer/">27 Маленький слой</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/the-packet-bridge/index.html b/apps/website/public/ru/learn/the-packet-bridge/index.html
index 1b90af3e98..4dc9a4b4a5 100644
--- a/apps/website/public/ru/learn/the-packet-bridge/index.html
+++ b/apps/website/public/ru/learn/the-packet-bridge/index.html
@@ -593,5 +593,6 @@ <h1>Пакетный мост</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое шина</h3><p>Зачем вообще существует шина: разговор по проводам, с кадрами и адресами, и кому позволено говорить.</p><ol><li><a href="/ru/learn/wires-that-talk/">01 Провода, которые говорят</a></li><li><a href="/ru/learn/framing-a-conversation/">02 Кадрирование разговора</a></li><li><a href="/ru/learn/master-and-servant/">03 Хозяин и слуга</a></li></ol></li><li><h3>Модуль 2 · Разговор по UART</h3><p>Двухпроводная шина без такта: кадр, делитель, задающий скорость, и статус, который опрашивает драйвер.</p><ol><li><a href="/ru/learn/the-frame/">04 Кадр: старт, данные, стоп</a></li><li><a href="/ru/learn/baud-and-the-divisor/">05 Бод и делитель</a></li><li><a href="/ru/learn/status-and-fifos/">06 Статус и FIFO</a></li></ol></li><li><h3>Модуль 3 · SPI под тактом</h3><p>Разговор под тактом: четыре режима, лестница предделителей и выбор кристалла на каждого подчинённого.</p><ol><li><a href="/ru/learn/modes-cpol-cpha/">07 Режимы: CPOL и CPHA</a></li><li><a href="/ru/learn/the-prescaler-ladder/">08 Лестница предделителей</a></li><li><a href="/ru/learn/chip-select-and-widths/">09 Выбор кристалла и ширины</a></li></ol></li><li><h3>Модуль 4 · Регистровая шина APB</h3><p>Шина регистров: PSEL и PENABLE, строобы и ожидания, и сколько адресных бит стоит количество периферии.</p><ol><li><a href="/ru/learn/psel-and-penable/">10 PSEL и PENABLE</a></li><li><a href="/ru/learn/strobes-and-waits/">11 Строобы и ожидания</a></li><li><a href="/ru/learn/address-decode/">12 Дешифрация адреса</a></li></ol></li><li><h3>Модуль 5 · Пять каналов AXI4</h3><p>Пять каналов: адрес, данные и отклик в обе стороны, lite или full, пакеты и идентификаторы.</p><ol><li><a href="/ru/learn/five-channels/">13 Пять каналов</a></li><li><a href="/ru/learn/lite-or-full/">14 Lite или full</a></li><li><a href="/ru/learn/bursts-and-ids/">15 Пакеты и идентификаторы</a></li></ol></li><li><h3>Модуль 6 · Память</h3><p>Что находится на дальнем конце каждой шины: карты памяти, типы портов и задержка, которую должно покрывать ожидание.</p><ol><li><a href="/ru/learn/memory-maps/">16 Карты памяти</a></li><li><a href="/ru/learn/ports-and-kinds/">17 Порты и типы</a></li><li><a href="/ru/learn/latency/">18 Задержка</a></li></ol></li><li><h3>Модуль 7 · Мосты</h3><p>Почему в дизайне появляется больше одной шины и пакетный мост, переносящий работу между ними.</p><ol><li><a href="/ru/learn/why-bridges/">19 Зачем мосты</a></li><li class="on" aria-current="page"><a href="/ru/learn/the-packet-bridge/">20 Пакетный мост</a></li><li><a href="/ru/learn/everything-in-one-bench/">21 Всё в одном стенде</a></li></ol></li><li><h3>Модуль 8 · Ethernet: кадры и тайминг</h3><p>Кадры, контрольная сумма кадра, тайминг RGMII и заранее зарегистрированные шаги настоящего запуска.</p><ol><li><a href="/ru/learn/frames-and-the-fcs/">22 Кадры и контрольная сумма</a></li><li><a href="/ru/learn/rgmii-timing/">23 Тайминг RGMII</a></li><li><a href="/ru/learn/bring-up-on-the-bench/">24 Запуск на стенде</a></li></ol></li><li><h3>Модуль 9 · Стенд</h3><p>Дисциплина, оберегающая настоящее железо: кто держит IO, как взять и вернуть захват, и что запускается дальше.</p><ol><li><a href="/ru/learn/who-holds-the-io/">25 Кто держит IO</a></li><li><a href="/ru/learn/claim-and-release/">26 Захват и возврат</a></li><li><a href="/ru/learn/capstone-what-runs-next/">27 Капстоун: что запускается дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/the-prescaler-ladder/index.html b/apps/website/public/ru/learn/the-prescaler-ladder/index.html
index 59e2dd72e8..e0c831eb42 100644
--- a/apps/website/public/ru/learn/the-prescaler-ladder/index.html
+++ b/apps/website/public/ru/learn/the-prescaler-ladder/index.html
@@ -209,5 +209,6 @@ <h1>Лестница предделителей</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое шина</h3><p>Зачем вообще существует шина: разговор по проводам, с кадрами и адресами, и кому позволено говорить.</p><ol><li><a href="/ru/learn/wires-that-talk/">01 Провода, которые говорят</a></li><li><a href="/ru/learn/framing-a-conversation/">02 Кадрирование разговора</a></li><li><a href="/ru/learn/master-and-servant/">03 Хозяин и слуга</a></li></ol></li><li><h3>Модуль 2 · Разговор по UART</h3><p>Двухпроводная шина без такта: кадр, делитель, задающий скорость, и статус, который опрашивает драйвер.</p><ol><li><a href="/ru/learn/the-frame/">04 Кадр: старт, данные, стоп</a></li><li><a href="/ru/learn/baud-and-the-divisor/">05 Бод и делитель</a></li><li><a href="/ru/learn/status-and-fifos/">06 Статус и FIFO</a></li></ol></li><li><h3>Модуль 3 · SPI под тактом</h3><p>Разговор под тактом: четыре режима, лестница предделителей и выбор кристалла на каждого подчинённого.</p><ol><li><a href="/ru/learn/modes-cpol-cpha/">07 Режимы: CPOL и CPHA</a></li><li class="on" aria-current="page"><a href="/ru/learn/the-prescaler-ladder/">08 Лестница предделителей</a></li><li><a href="/ru/learn/chip-select-and-widths/">09 Выбор кристалла и ширины</a></li></ol></li><li><h3>Модуль 4 · Регистровая шина APB</h3><p>Шина регистров: PSEL и PENABLE, строобы и ожидания, и сколько адресных бит стоит количество периферии.</p><ol><li><a href="/ru/learn/psel-and-penable/">10 PSEL и PENABLE</a></li><li><a href="/ru/learn/strobes-and-waits/">11 Строобы и ожидания</a></li><li><a href="/ru/learn/address-decode/">12 Дешифрация адреса</a></li></ol></li><li><h3>Модуль 5 · Пять каналов AXI4</h3><p>Пять каналов: адрес, данные и отклик в обе стороны, lite или full, пакеты и идентификаторы.</p><ol><li><a href="/ru/learn/five-channels/">13 Пять каналов</a></li><li><a href="/ru/learn/lite-or-full/">14 Lite или full</a></li><li><a href="/ru/learn/bursts-and-ids/">15 Пакеты и идентификаторы</a></li></ol></li><li><h3>Модуль 6 · Память</h3><p>Что находится на дальнем конце каждой шины: карты памяти, типы портов и задержка, которую должно покрывать ожидание.</p><ol><li><a href="/ru/learn/memory-maps/">16 Карты памяти</a></li><li><a href="/ru/learn/ports-and-kinds/">17 Порты и типы</a></li><li><a href="/ru/learn/latency/">18 Задержка</a></li></ol></li><li><h3>Модуль 7 · Мосты</h3><p>Почему в дизайне появляется больше одной шины и пакетный мост, переносящий работу между ними.</p><ol><li><a href="/ru/learn/why-bridges/">19 Зачем мосты</a></li><li><a href="/ru/learn/the-packet-bridge/">20 Пакетный мост</a></li><li><a href="/ru/learn/everything-in-one-bench/">21 Всё в одном стенде</a></li></ol></li><li><h3>Модуль 8 · Ethernet: кадры и тайминг</h3><p>Кадры, контрольная сумма кадра, тайминг RGMII и заранее зарегистрированные шаги настоящего запуска.</p><ol><li><a href="/ru/learn/frames-and-the-fcs/">22 Кадры и контрольная сумма</a></li><li><a href="/ru/learn/rgmii-timing/">23 Тайминг RGMII</a></li><li><a href="/ru/learn/bring-up-on-the-bench/">24 Запуск на стенде</a></li></ol></li><li><h3>Модуль 9 · Стенд</h3><p>Дисциплина, оберегающая настоящее железо: кто держит IO, как взять и вернуть захват, и что запускается дальше.</p><ol><li><a href="/ru/learn/who-holds-the-io/">25 Кто держит IO</a></li><li><a href="/ru/learn/claim-and-release/">26 Захват и возврат</a></li><li><a href="/ru/learn/capstone-what-runs-next/">27 Капстоун: что запускается дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/three-way-agreement/index.html b/apps/website/public/ru/learn/three-way-agreement/index.html
index 4004868189..6895e26f94 100644
--- a/apps/website/public/ru/learn/three-way-agreement/index.html
+++ b/apps/website/public/ru/learn/three-way-agreement/index.html
@@ -107,5 +107,6 @@ <h1>Тройное согласие</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Зачем проверять</h3><p>Дизайны, которые компилируются и ошибаются; модель, которая выносит вердикт; план, записанный до кода.</p><ol><li><a href="/ru/learn/bugs-that-compile/">01 Баги, которые компилируются</a></li><li><a href="/ru/learn/golden-model/">02 Золотая модель</a></li><li><a href="/ru/learn/plan-before-code/">03 Сначала план</a></li></ol></li><li><h3>Модуль 2 · Тестбенчи</h3><p>Стимулы, проверки и вердикт, записанные как один spec рядом с дизайном, который они судят.</p><ol><li><a href="/ru/learn/anatomy-of-a-tb/">04 Анатомия тестбенча</a></li><li><a href="/ru/learn/self-checking/">05 Самопроверка вместо всматривания</a></li><li><a href="/ru/learn/directed-vs-random/">06 Направленные против случайных</a></li></ol></li><li><h3>Модуль 3 · Временные диаграммы</h3><p>Трасса каждого сигнала, прочитанная так, как её читает инженер по железу, и два прогона, сравнённые между собой.</p><ol><li><a href="/ru/learn/vcd-format/">07 Формат VCD</a></li><li><a href="/ru/learn/debug-with-waves/">08 Отладка по волнам</a></li><li><a href="/ru/learn/compare-traces/">09 Сравнение двух трасс</a></li></ol></li><li><h3>Модуль 4 · Векторы соответствия</h3><p>Случаи с ответом, записанным рядом, там, откуда их достанет компилятор.</p><ol><li><a href="/ru/learn/what-a-vector-is/">10 Что такое вектор</a></li><li><a href="/ru/learn/golden-ruler-vectors/">11 Векторы как эталонная линейка</a></li><li><a href="/ru/learn/end-to-end-scenarios/">12 Сквозные сценарии</a></li></ol></li><li><h3>Модуль 5 · Косимуляция</h3><p>Spec, симулятор и плата сходятся в одном ответе на стенде Artix-7 XC7A200T, и что делать, когда не сходятся.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/three-way-agreement/">13 Тройное согласие</a></li><li><a href="/ru/learn/simulator-internals/">14 Внутри симулятора</a></li><li><a href="/ru/learn/when-they-disagree/">15 Когда они не сходятся</a></li></ol></li><li><h3>Модуль 6 · Покрытие</h3><p>Чего коснулись тесты: строки, переключения, состояния — и что прячет это число.</p><ol><li><a href="/ru/learn/line-and-toggle/">16 Строки и переключения</a></li><li><a href="/ru/learn/fsm-coverage/">17 Покрытие автоматов</a></li><li><a href="/ru/learn/coverage-lies/">18 Покрытие лжёт</a></li></ol></li><li><h3>Модуль 7 · Формальные методы</h3><p>Ассерты, верные каждый такт; ограниченный поиск контрпримера; и почему доказательству нужна индукция.</p><ol><li><a href="/ru/learn/assertions/">19 Ассерты</a></li><li><a href="/ru/learn/bounded-model-checking/">20 Ограниченная проверка моделей</a></li><li><a href="/ru/learn/induction/">21 Индукция</a></li></ol></li><li><h3>Модуль 8 · Мутации</h3><p>Ломайте дизайн нарочно и считайте, что заметили тесты.</p><ol><li><a href="/ru/learn/kill-the-mutant/">22 Убить мутанта</a></li><li><a href="/ru/learn/surviving-mutants/">23 Выжившие мутанты</a></li><li><a href="/ru/learn/test-strength/">24 Сила тестов</a></li></ol></li><li><h3>Модуль 9 · Приёмка</h3><p>Одна команда, все квитанции, чистый вердикт, который можно показать.</p><ol><li><a href="/ru/learn/build-verify/">25 Сборка и проверка</a></li><li><a href="/ru/learn/board-selftest/">26 Стенд проверяет себя</a></li><li><a href="/ru/learn/verify-capstone/">27 Капстоун: проверить модуль целиком</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">Этот курс сам является spec t27: прочитайте его</a></p><p>Виджеты сохраняют свои английские слова: каждый живёт в собственном spec t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/three-way-jump/index.html b/apps/website/public/ru/learn/three-way-jump/index.html
index ff34ec33a1..1ee313de0e 100644
--- a/apps/website/public/ru/learn/three-way-jump/index.html
+++ b/apps/website/public/ru/learn/three-way-jump/index.html
@@ -769,5 +769,6 @@ <h1>Переход на три стороны</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Три значения</h3><p>Почему три, как сбалансированная троичная система пишет любое число без знака и что дают переворот и отсечение тритов.</p><ol><li><a href="/ru/learn/why-three/">01 Почему три</a></li><li><a href="/ru/learn/balanced-ternary/">02 Сбалансированная троичная</a></li><li><a href="/ru/learn/flip-and-cut/">03 Переворот вместо минуса, отсечение вместо округления</a></li></ol></li><li><h3>Модуль 2 · Логика с «не знаю»</h3><p>Трёхзначные вентили Клини, два вентиля без двоичного двойника и сложение как пара таблиц.</p><ol><li><a href="/ru/learn/kleene-logic/">04 Трёхзначная логика</a></li><li><a href="/ru/learn/consensus-and-any/">05 Консенсус и «любой»</a></li><li><a href="/ru/learn/sum-and-carry/">06 Сумма и перенос</a></li></ol></li><li><h3>Модуль 3 · Складываем триты</h3><p>Полусумматор, полный сумматор и перенос, бегущий влево разряд за разрядом.</p><ol><li><a href="/ru/learn/half-adder/">07 Полусумматор</a></li><li><a href="/ru/learn/full-adder/">08 Полный сумматор</a></li><li><a href="/ru/learn/ripple-carry/">09 Сквозной перенос</a></li></ol></li><li><h3>Модуль 4 · Умножение без умножителя</h3><p>Копия, ноль или переворот: произведение на один трит, умножение столбиком и MAC, который только складывает.</p><ol><li><a href="/ru/learn/times-a-trit/">10 Умножить на один трит</a></li><li><a href="/ru/learn/long-multiplication/">11 Умножение столбиком</a></li><li><a href="/ru/learn/ternary-mac/">12 Троичный MAC</a></li></ol></li><li><h3>Модуль 5 · Триты в двоичной памяти</h3><p>Два бита на трит, пять тритов на байт и сколько бит нужно слову из 27 тритов.</p><ol><li><a href="/ru/learn/two-bits-per-trit/">13 Два бита на трит</a></li><li><a href="/ru/learn/five-trits-a-byte/">14 Пять тритов в байте</a></li><li><a href="/ru/learn/a-word-of-27/">15 Слово из 27 тритов</a></li></ol></li><li><h3>Модуль 6 · Командное слово TRI-27</h3><p>32-битное слово, которое декодирует эмулятор Trinity: его поля, 47 опкодов и 15-битное непосредственное значение.</p><ol><li><a href="/ru/learn/the-instruction-word/">16 Командное слово</a></li><li><a href="/ru/learn/the-opcode-map/">17 Карта опкодов</a></li><li><a href="/ru/learn/the-immediate/">18 Непосредственное значение</a></li></ol></li><li><h3>Модуль 7 · Троичная машина</h3><p>АЛУ из сумматоров этого курса, переход на три стороны и программа, которую можно пройти по шагам.</p><ol><li><a href="/ru/learn/a-ternary-alu/">19 Троичное АЛУ</a></li><li class="on" aria-current="page"><a href="/ru/learn/three-way-jump/">20 Переход на три стороны</a></li><li><a href="/ru/learn/run-a-program/">21 Запустить программу</a></li></ol></li><li><h3>Модуль 8 · Три ответа</h3><p>Сравнение по старшему различному триту, поиск числа по третям и сортировка трёхзначными сравнениями.</p><ol><li><a href="/ru/learn/compare-at-the-top/">22 Сравнение сверху</a></li><li><a href="/ru/learn/questions-in-thirds/">23 Вопросы по третям</a></li><li><a href="/ru/learn/sort-by-three/">24 Сортировка трёхзначными сравнениями</a></li></ol></li><li><h3>Модуль 9 · Троичные нейроны</h3><p>Веса -1, 0, +1: один нейрон, детектор и маленький слой, и нигде нет умножителя.</p><ol><li><a href="/ru/learn/one-ternary-neuron/">25 Один троичный нейрон</a></li><li><a href="/ru/learn/a-bar-detector/">26 Детектор полосы</a></li><li><a href="/ru/learn/a-tiny-layer/">27 Маленький слой</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/times-a-trit/index.html b/apps/website/public/ru/learn/times-a-trit/index.html
index 0ef0091230..26c3ac7899 100644
--- a/apps/website/public/ru/learn/times-a-trit/index.html
+++ b/apps/website/public/ru/learn/times-a-trit/index.html
@@ -769,5 +769,6 @@ <h1>Умножить на один трит</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Три значения</h3><p>Почему три, как сбалансированная троичная система пишет любое число без знака и что дают переворот и отсечение тритов.</p><ol><li><a href="/ru/learn/why-three/">01 Почему три</a></li><li><a href="/ru/learn/balanced-ternary/">02 Сбалансированная троичная</a></li><li><a href="/ru/learn/flip-and-cut/">03 Переворот вместо минуса, отсечение вместо округления</a></li></ol></li><li><h3>Модуль 2 · Логика с «не знаю»</h3><p>Трёхзначные вентили Клини, два вентиля без двоичного двойника и сложение как пара таблиц.</p><ol><li><a href="/ru/learn/kleene-logic/">04 Трёхзначная логика</a></li><li><a href="/ru/learn/consensus-and-any/">05 Консенсус и «любой»</a></li><li><a href="/ru/learn/sum-and-carry/">06 Сумма и перенос</a></li></ol></li><li><h3>Модуль 3 · Складываем триты</h3><p>Полусумматор, полный сумматор и перенос, бегущий влево разряд за разрядом.</p><ol><li><a href="/ru/learn/half-adder/">07 Полусумматор</a></li><li><a href="/ru/learn/full-adder/">08 Полный сумматор</a></li><li><a href="/ru/learn/ripple-carry/">09 Сквозной перенос</a></li></ol></li><li><h3>Модуль 4 · Умножение без умножителя</h3><p>Копия, ноль или переворот: произведение на один трит, умножение столбиком и MAC, который только складывает.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/times-a-trit/">10 Умножить на один трит</a></li><li><a href="/ru/learn/long-multiplication/">11 Умножение столбиком</a></li><li><a href="/ru/learn/ternary-mac/">12 Троичный MAC</a></li></ol></li><li><h3>Модуль 5 · Триты в двоичной памяти</h3><p>Два бита на трит, пять тритов на байт и сколько бит нужно слову из 27 тритов.</p><ol><li><a href="/ru/learn/two-bits-per-trit/">13 Два бита на трит</a></li><li><a href="/ru/learn/five-trits-a-byte/">14 Пять тритов в байте</a></li><li><a href="/ru/learn/a-word-of-27/">15 Слово из 27 тритов</a></li></ol></li><li><h3>Модуль 6 · Командное слово TRI-27</h3><p>32-битное слово, которое декодирует эмулятор Trinity: его поля, 47 опкодов и 15-битное непосредственное значение.</p><ol><li><a href="/ru/learn/the-instruction-word/">16 Командное слово</a></li><li><a href="/ru/learn/the-opcode-map/">17 Карта опкодов</a></li><li><a href="/ru/learn/the-immediate/">18 Непосредственное значение</a></li></ol></li><li><h3>Модуль 7 · Троичная машина</h3><p>АЛУ из сумматоров этого курса, переход на три стороны и программа, которую можно пройти по шагам.</p><ol><li><a href="/ru/learn/a-ternary-alu/">19 Троичное АЛУ</a></li><li><a href="/ru/learn/three-way-jump/">20 Переход на три стороны</a></li><li><a href="/ru/learn/run-a-program/">21 Запустить программу</a></li></ol></li><li><h3>Модуль 8 · Три ответа</h3><p>Сравнение по старшему различному триту, поиск числа по третям и сортировка трёхзначными сравнениями.</p><ol><li><a href="/ru/learn/compare-at-the-top/">22 Сравнение сверху</a></li><li><a href="/ru/learn/questions-in-thirds/">23 Вопросы по третям</a></li><li><a href="/ru/learn/sort-by-three/">24 Сортировка трёхзначными сравнениями</a></li></ol></li><li><h3>Модуль 9 · Троичные нейроны</h3><p>Веса -1, 0, +1: один нейрон, детектор и маленький слой, и нигде нет умножителя.</p><ol><li><a href="/ru/learn/one-ternary-neuron/">25 Один троичный нейрон</a></li><li><a href="/ru/learn/a-bar-detector/">26 Детектор полосы</a></li><li><a href="/ru/learn/a-tiny-layer/">27 Маленький слой</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/timing-slack/index.html b/apps/website/public/ru/learn/timing-slack/index.html
index a1eed080cc..88fddced12 100644
--- a/apps/website/public/ru/learn/timing-slack/index.html
+++ b/apps/website/public/ru/learn/timing-slack/index.html
@@ -294,5 +294,6 @@ <h1>Тайминг и запас</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Чип</h3><p>Что такое FPGA, какой чип у нас и как его выводы встречаются с платой.</p><ol><li><a href="/ru/learn/fpga-is-a-grid/">01 FPGA — это сетка, которую вы настраиваете</a></li><li><a href="/ru/learn/which-chip/">02 Какой это чип?</a></li><li><a href="/ru/learn/pins-and-package/">03 Выводы, шарики и корпус</a></li></ol></li><li><h3>Модуль 2 · Числа в железе</h3><p>Биты, триты и форматы чисел, и сколько стоит арифметика в логике.</p><ol><li><a href="/ru/learn/bits-and-trits/">04 Биты, триты и GF16</a></li><li><a href="/ru/learn/what-arithmetic-costs/">05 Сколько стоит арифметика</a></li><li><a href="/ru/learn/how-much-fits/">06 Сколько помещается на чип</a></li></ol></li><li><h3>Модуль 3 · Ваша программа на t27</h3><p>Напишите спеку, проверьте её и увидьте, почему «компилируется» не значит «верно».</p><ol><li><a href="/ru/learn/hello-t27/">07 Привет, t27</a></li><li><a href="/ru/learn/tests-are-the-spec/">08 Тесты — часть спеки</a></li><li><a href="/ru/learn/compiles-is-not-correct/">09 Компилируется — не значит верно</a></li></ol></li><li><h3>Модуль 4 · Внутри t27c</h3><p>Как компилятор читает спеку и что он пишет, включая нативный t27b.</p><ol><li><a href="/ru/learn/compiler-stages/">10 Стадии t27c</a></li><li><a href="/ru/learn/t27c-on-a-real-machine/">11 t27c на настоящей машине</a></li><li><a href="/ru/learn/seven-backends-and-t27b/">12 Семь бэкендов и t27b</a></li></ol></li><li><h3>Модуль 5 · От спеки к железу</h3><p>Verilog, который пишет t27c, ячейки, в которые он превращается, и что делает одна LUT.</p><ol><li><a href="/ru/learn/spec-to-verilog/">13 От спеки к Verilog</a></li><li><a href="/ru/learn/count-the-cells/">14 Посчитайте ячейки</a></li><li><a href="/ru/learn/one-lut-up-close/">15 Одна LUT вблизи</a></li></ol></li><li><h3>Модуль 6 · Читаем синтез</h3><p>Что yosys сообщает о вашем дизайне и какие предупреждения важны.</p><ol><li><a href="/ru/learn/synthesis-warnings/">16 Важные предупреждения синтеза</a></li><li><a href="/ru/learn/state-machines/">17 Конечные автоматы</a></li><li><a href="/ru/learn/where-the-luts-go/">18 Куда уходят LUT</a></li></ol></li><li><h3>Модуль 7 · Размещение, трассировка, тайминг</h3><p>Где ячейки оказываются на кристалле и успевает ли тактовая частота.</p><ol><li><a href="/ru/learn/placement/">19 Размещение: где оказываются ячейки</a></li><li><a href="/ru/learn/logic-depth/">20 Глубина логики</a></li><li class="on" aria-current="page"><a href="/ru/learn/timing-slack/">21 Тайминг и запас</a></li></ol></li><li><h3>Модуль 8 · Битстрим</h3><p>Как оттрассированный дизайн становится битами, которые загружает чип, без Vivado.</p><ol><li><a href="/ru/learn/fasm-to-frames/">22 От FASM к фреймам</a></li><li><a href="/ru/learn/a-build-receipt/">23 Сборка по пунктам</a></li><li><a href="/ru/learn/read-a-bitstream/">24 Читаем битстрим</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Спросите чип, кто он, загрузите биты и проверьте их.</p><ol><li><a href="/ru/learn/jtag-asks-the-chip/">25 JTAG спрашивает чип</a></li><li><a href="/ru/learn/same-bits-on-the-board/">26 Те же биты — на плате</a></li><li><a href="/ru/learn/capstone/">27 Итог: переугадайте yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/tri-commands/index.html b/apps/website/public/ru/learn/tri-commands/index.html
index f831abb524..f6742708a7 100644
--- a/apps/website/public/ru/learn/tri-commands/index.html
+++ b/apps/website/public/ru/learn/tri-commands/index.html
@@ -85,5 +85,6 @@ <h1>tri: командная строка</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Файл спеки</h3><p>Что лежит в файле .t27: строка module, строки прозы и комментарии.</p><ol><li><a href="/ru/learn/what-a-spec-is/">01 Что такое спека</a></li><li><a href="/ru/learn/the-module-line/">02 Строка module</a></li><li><a href="/ru/learn/comments-and-prose/">03 Комментарии и строки прозы</a></li></ol></li><li><h3>Модуль 2 · Значения и типы</h3><p>Константы, ширина целого числа, а также true, false и текст.</p><ol><li><a href="/ru/learn/constants/">04 Константы</a></li><li><a href="/ru/learn/whole-numbers/">05 Целые числа и их ширина</a></li><li><a href="/ru/learn/true-false-and-text/">06 Истина, ложь и текст</a></li></ol></li><li><h3>Модуль 3 · Массивы, триты и выражения</h3><p>Списки значений, три значения трита и что вычисляют операторы.</p><ol><li><a href="/ru/learn/arrays/">07 Массивы</a></li><li><a href="/ru/learn/trits/">08 Триты: минус один, ноль, один</a></li><li><a href="/ru/learn/expressions/">09 Выражения</a></li></ol></li><li><h3>Модуль 4 · Тесты и функции</h3><p>Блок test, несколько маленьких тестов в одной спеке и функция с типизированными входами.</p><ol><li><a href="/ru/learn/a-test-block/">10 Блок test</a></li><li><a href="/ru/learn/many-tests/">11 Много тестов в одной спеке</a></li><li><a href="/ru/learn/functions/">12 Функции</a></li></ol></li><li><h3>Модуль 5 · Внутри функции</h3><p>Имена через let и var, выбор через if и switch и циклы.</p><ol><li><a href="/ru/learn/local-names/">13 Локальные имена: let и var</a></li><li><a href="/ru/learn/if-and-else/">14 Выбор: if, else и switch</a></li><li><a href="/ru/learn/loops/">15 Циклы: while и for</a></li></ol></li><li><h3>Модуль 6 · Видимость и свои типы</h3><p>Что pub открывает другим модулям, и структуры и перечисления, которые вы объявляете.</p><ol><li><a href="/ru/learn/pub-or-private/">16 pub или закрытое имя</a></li><li><a href="/ru/learn/structs/">17 Структуры</a></li><li><a href="/ru/learn/enums/">18 Перечисления</a></li></ol></li><li><h3>Модуль 7 · Модули, правила и gen-ts</h3><p>Чтение другого модуля, правило, которое выполняется всегда, и спека, превращённая в TypeScript.</p><ol><li><a href="/ru/learn/use-other-modules/">19 Другие модули</a></li><li><a href="/ru/learn/invariants/">20 Инварианты и замеры</a></li><li><a href="/ru/learn/gen-ts/">21 gen-ts: спека превращается в TypeScript</a></li></ol></li><li><h3>Модуль 8 · Компилятор и tri</h3><p>Бэкенды, которые пишет t27c, стадии, в которых он читает спеку, и команда tri.</p><ol><li><a href="/ru/learn/seven-backends/">22 Семь бэкендов</a></li><li><a href="/ru/learn/stages-of-t27c/">23 Стадии t27c</a></li><li class="on" aria-current="page"><a href="/ru/learn/tri-commands/">24 tri: командная строка</a></li></ol></li><li><h3>Модуль 9 · Ошибки, программа и что дальше</h3><p>Как читать ошибку компилятора, одна маленькая законченная программа и какой курс пройти дальше.</p><ol><li><a href="/ru/learn/reading-errors/">25 Как читать ошибки компилятора</a></li><li><a href="/ru/learn/a-small-program/">26 Маленькая законченная программа</a></li><li><a href="/ru/learn/where-next/">27 Куда дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/trits/index.html b/apps/website/public/ru/learn/trits/index.html
index 26c95ac0e4..3d05ba6b2b 100644
--- a/apps/website/public/ru/learn/trits/index.html
+++ b/apps/website/public/ru/learn/trits/index.html
@@ -89,5 +89,6 @@ <h1>Триты: минус один, ноль, один</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Файл спеки</h3><p>Что лежит в файле .t27: строка module, строки прозы и комментарии.</p><ol><li><a href="/ru/learn/what-a-spec-is/">01 Что такое спека</a></li><li><a href="/ru/learn/the-module-line/">02 Строка module</a></li><li><a href="/ru/learn/comments-and-prose/">03 Комментарии и строки прозы</a></li></ol></li><li><h3>Модуль 2 · Значения и типы</h3><p>Константы, ширина целого числа, а также true, false и текст.</p><ol><li><a href="/ru/learn/constants/">04 Константы</a></li><li><a href="/ru/learn/whole-numbers/">05 Целые числа и их ширина</a></li><li><a href="/ru/learn/true-false-and-text/">06 Истина, ложь и текст</a></li></ol></li><li><h3>Модуль 3 · Массивы, триты и выражения</h3><p>Списки значений, три значения трита и что вычисляют операторы.</p><ol><li><a href="/ru/learn/arrays/">07 Массивы</a></li><li class="on" aria-current="page"><a href="/ru/learn/trits/">08 Триты: минус один, ноль, один</a></li><li><a href="/ru/learn/expressions/">09 Выражения</a></li></ol></li><li><h3>Модуль 4 · Тесты и функции</h3><p>Блок test, несколько маленьких тестов в одной спеке и функция с типизированными входами.</p><ol><li><a href="/ru/learn/a-test-block/">10 Блок test</a></li><li><a href="/ru/learn/many-tests/">11 Много тестов в одной спеке</a></li><li><a href="/ru/learn/functions/">12 Функции</a></li></ol></li><li><h3>Модуль 5 · Внутри функции</h3><p>Имена через let и var, выбор через if и switch и циклы.</p><ol><li><a href="/ru/learn/local-names/">13 Локальные имена: let и var</a></li><li><a href="/ru/learn/if-and-else/">14 Выбор: if, else и switch</a></li><li><a href="/ru/learn/loops/">15 Циклы: while и for</a></li></ol></li><li><h3>Модуль 6 · Видимость и свои типы</h3><p>Что pub открывает другим модулям, и структуры и перечисления, которые вы объявляете.</p><ol><li><a href="/ru/learn/pub-or-private/">16 pub или закрытое имя</a></li><li><a href="/ru/learn/structs/">17 Структуры</a></li><li><a href="/ru/learn/enums/">18 Перечисления</a></li></ol></li><li><h3>Модуль 7 · Модули, правила и gen-ts</h3><p>Чтение другого модуля, правило, которое выполняется всегда, и спека, превращённая в TypeScript.</p><ol><li><a href="/ru/learn/use-other-modules/">19 Другие модули</a></li><li><a href="/ru/learn/invariants/">20 Инварианты и замеры</a></li><li><a href="/ru/learn/gen-ts/">21 gen-ts: спека превращается в TypeScript</a></li></ol></li><li><h3>Модуль 8 · Компилятор и tri</h3><p>Бэкенды, которые пишет t27c, стадии, в которых он читает спеку, и команда tri.</p><ol><li><a href="/ru/learn/seven-backends/">22 Семь бэкендов</a></li><li><a href="/ru/learn/stages-of-t27c/">23 Стадии t27c</a></li><li><a href="/ru/learn/tri-commands/">24 tri: командная строка</a></li></ol></li><li><h3>Модуль 9 · Ошибки, программа и что дальше</h3><p>Как читать ошибку компилятора, одна маленькая законченная программа и какой курс пройти дальше.</p><ol><li><a href="/ru/learn/reading-errors/">25 Как читать ошибки компилятора</a></li><li><a href="/ru/learn/a-small-program/">26 Маленькая законченная программа</a></li><li><a href="/ru/learn/where-next/">27 Куда дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/true-false-and-text/index.html b/apps/website/public/ru/learn/true-false-and-text/index.html
index ae99df61c2..ff76e81226 100644
--- a/apps/website/public/ru/learn/true-false-and-text/index.html
+++ b/apps/website/public/ru/learn/true-false-and-text/index.html
@@ -86,5 +86,6 @@ <h1>Истина, ложь и текст</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Файл спеки</h3><p>Что лежит в файле .t27: строка module, строки прозы и комментарии.</p><ol><li><a href="/ru/learn/what-a-spec-is/">01 Что такое спека</a></li><li><a href="/ru/learn/the-module-line/">02 Строка module</a></li><li><a href="/ru/learn/comments-and-prose/">03 Комментарии и строки прозы</a></li></ol></li><li><h3>Модуль 2 · Значения и типы</h3><p>Константы, ширина целого числа, а также true, false и текст.</p><ol><li><a href="/ru/learn/constants/">04 Константы</a></li><li><a href="/ru/learn/whole-numbers/">05 Целые числа и их ширина</a></li><li class="on" aria-current="page"><a href="/ru/learn/true-false-and-text/">06 Истина, ложь и текст</a></li></ol></li><li><h3>Модуль 3 · Массивы, триты и выражения</h3><p>Списки значений, три значения трита и что вычисляют операторы.</p><ol><li><a href="/ru/learn/arrays/">07 Массивы</a></li><li><a href="/ru/learn/trits/">08 Триты: минус один, ноль, один</a></li><li><a href="/ru/learn/expressions/">09 Выражения</a></li></ol></li><li><h3>Модуль 4 · Тесты и функции</h3><p>Блок test, несколько маленьких тестов в одной спеке и функция с типизированными входами.</p><ol><li><a href="/ru/learn/a-test-block/">10 Блок test</a></li><li><a href="/ru/learn/many-tests/">11 Много тестов в одной спеке</a></li><li><a href="/ru/learn/functions/">12 Функции</a></li></ol></li><li><h3>Модуль 5 · Внутри функции</h3><p>Имена через let и var, выбор через if и switch и циклы.</p><ol><li><a href="/ru/learn/local-names/">13 Локальные имена: let и var</a></li><li><a href="/ru/learn/if-and-else/">14 Выбор: if, else и switch</a></li><li><a href="/ru/learn/loops/">15 Циклы: while и for</a></li></ol></li><li><h3>Модуль 6 · Видимость и свои типы</h3><p>Что pub открывает другим модулям, и структуры и перечисления, которые вы объявляете.</p><ol><li><a href="/ru/learn/pub-or-private/">16 pub или закрытое имя</a></li><li><a href="/ru/learn/structs/">17 Структуры</a></li><li><a href="/ru/learn/enums/">18 Перечисления</a></li></ol></li><li><h3>Модуль 7 · Модули, правила и gen-ts</h3><p>Чтение другого модуля, правило, которое выполняется всегда, и спека, превращённая в TypeScript.</p><ol><li><a href="/ru/learn/use-other-modules/">19 Другие модули</a></li><li><a href="/ru/learn/invariants/">20 Инварианты и замеры</a></li><li><a href="/ru/learn/gen-ts/">21 gen-ts: спека превращается в TypeScript</a></li></ol></li><li><h3>Модуль 8 · Компилятор и tri</h3><p>Бэкенды, которые пишет t27c, стадии, в которых он читает спеку, и команда tri.</p><ol><li><a href="/ru/learn/seven-backends/">22 Семь бэкендов</a></li><li><a href="/ru/learn/stages-of-t27c/">23 Стадии t27c</a></li><li><a href="/ru/learn/tri-commands/">24 tri: командная строка</a></li></ol></li><li><h3>Модуль 9 · Ошибки, программа и что дальше</h3><p>Как читать ошибку компилятора, одна маленькая законченная программа и какой курс пройти дальше.</p><ol><li><a href="/ru/learn/reading-errors/">25 Как читать ошибки компилятора</a></li><li><a href="/ru/learn/a-small-program/">26 Маленькая законченная программа</a></li><li><a href="/ru/learn/where-next/">27 Куда дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/twiddles-and-bit-reversal/index.html b/apps/website/public/ru/learn/twiddles-and-bit-reversal/index.html
index 85d17b9a89..bce67fefe7 100644
--- a/apps/website/public/ru/learn/twiddles-and-bit-reversal/index.html
+++ b/apps/website/public/ru/learn/twiddles-and-bit-reversal/index.html
@@ -1122,5 +1122,6 @@ <h1>Поворачивающие множители и бит-реверс</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Сложение</h3><p>Почему сложение медленное: перенос, который идёт вверх по слову, выделенная цепь в каждом слайсе, префиксные сети и деревья сохранения переноса.</p><ol><li><a href="/ru/learn/the-carry-chain/">01 Цепь переноса</a></li><li><a href="/ru/learn/prefix-adders/">02 Взгляд вперёд: префиксные сумматоры</a></li><li><a href="/ru/learn/carry-save/">03 Три числа сразу</a></li></ol></li><li><h3>Модуль 2 · Умножение</h3><p>Произведение — сумма сдвинутых копий; перекодирование Бута вдвое сокращает их, а слайс DSP48E1 делает остальное одним блоком.</p><ol><li><a href="/ru/learn/shift-and-add/">04 Сдвиг и сложение</a></li><li><a href="/ru/learn/booths-trick/">05 Приём Бута</a></li><li><a href="/ru/learn/the-dsp48e1-slice/">06 Слайс DSP48E1</a></li></ol></li><li><h3>Модуль 3 · Фиксированная точка</h3><p>Где стоит двоичная точка, что округление делает со значением и с его средним и что даёт каждый бит квантователя.</p><ol><li><a href="/ru/learn/q-formats/">07 Форматы Q</a></li><li><a href="/ru/learn/rounding-and-saturation/">08 Округление и насыщение</a></li><li><a href="/ru/learn/quantization-noise/">09 Шум квантования</a></li></ol></li><li><h3>Модуль 4 · Функции в железе</h3><p>Синус, косинус, угол и длина из сдвигов и сложений — и когда таблица оказывается выгоднее.</p><ol><li><a href="/ru/learn/rotating-by-shifts/">10 Поворот сдвигами</a></li><li><a href="/ru/learn/angle-and-magnitude/">11 Угол и модуль</a></li><li><a href="/ru/learn/tables-and-interpolation/">12 Таблицы и интерполяция</a></li></ol></li><li><h3>Модуль 5 · Сигналы и дискретизация</h3><p>Что дискретизация делает с частотой, как генератор строится из сумматора и что измеряет бин ДПФ.</p><ol><li><a href="/ru/learn/sampling-and-aliasing/">13 Дискретизация и наложение спектров</a></li><li><a href="/ru/learn/the-nco/">14 Генератор с числовым управлением</a></li><li><a href="/ru/learn/bins-of-a-spectrum/">15 Бины спектра</a></li></ol></li><li><h3>Модуль 6 · Фильтры FIR</h3><p>Скользящее среднее, расчёт окном sinc с целыми коэффициентами и раскладка отводов по слайсам DSP.</p><ol><li><a href="/ru/learn/the-moving-average/">16 Скользящее среднее</a></li><li><a href="/ru/learn/designing-taps/">17 Расчёт коэффициентов</a></li><li><a href="/ru/learn/folding-taps-onto-slices/">18 Раскладка отводов по слайсам</a></li></ol></li><li><h3>Модуль 7 · Многоскоростная обработка</h3><p>Понижение частоты дискретизации без заворота шума: децимация, фильтр CIC и полифазная форма.</p><ol><li><a href="/ru/learn/lowering-the-rate/">19 Понижение частоты</a></li><li><a href="/ru/learn/the-cic-filter/">20 Фильтр CIC</a></li><li><a href="/ru/learn/polyphase/">21 Полифазная форма</a></li></ol></li><li><h3>Модуль 8 · БПФ</h3><p>N log N вместо N^2: бабочки, бит-реверсный порядок входа и биты, которые добавляет каждый этап.</p><ol><li><a href="/ru/learn/the-butterfly/">22 Бабочка</a></li><li class="on" aria-current="page"><a href="/ru/learn/twiddles-and-bit-reversal/">23 Поворачивающие множители и бит-реверс</a></li><li><a href="/ru/learn/growth-through-the-stages/">24 Рост через этапы</a></li></ol></li><li><h3>Модуль 9 · На стенде</h3><p>От арифметики к плате: коррелятор из спеки модема, бюджет слайсов и тайминга и итоговый фильтр.</p><ol><li><a href="/ru/learn/a-correlator-finds-the-frame/">25 Коррелятор находит кадр</a></li><li><a href="/ru/learn/does-it-fit-and-close-timing/">26 Помещается ли и сходится ли тайминг?</a></li><li><a href="/ru/learn/capstone-a-filter-on-the-board/">27 Итог: фильтр на плате</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/arithmetic-and-dsp.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/two-bits-per-trit/index.html b/apps/website/public/ru/learn/two-bits-per-trit/index.html
index ac5bc620c4..f52895b0fe 100644
--- a/apps/website/public/ru/learn/two-bits-per-trit/index.html
+++ b/apps/website/public/ru/learn/two-bits-per-trit/index.html
@@ -769,5 +769,6 @@ <h1>Два бита на трит</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Три значения</h3><p>Почему три, как сбалансированная троичная система пишет любое число без знака и что дают переворот и отсечение тритов.</p><ol><li><a href="/ru/learn/why-three/">01 Почему три</a></li><li><a href="/ru/learn/balanced-ternary/">02 Сбалансированная троичная</a></li><li><a href="/ru/learn/flip-and-cut/">03 Переворот вместо минуса, отсечение вместо округления</a></li></ol></li><li><h3>Модуль 2 · Логика с «не знаю»</h3><p>Трёхзначные вентили Клини, два вентиля без двоичного двойника и сложение как пара таблиц.</p><ol><li><a href="/ru/learn/kleene-logic/">04 Трёхзначная логика</a></li><li><a href="/ru/learn/consensus-and-any/">05 Консенсус и «любой»</a></li><li><a href="/ru/learn/sum-and-carry/">06 Сумма и перенос</a></li></ol></li><li><h3>Модуль 3 · Складываем триты</h3><p>Полусумматор, полный сумматор и перенос, бегущий влево разряд за разрядом.</p><ol><li><a href="/ru/learn/half-adder/">07 Полусумматор</a></li><li><a href="/ru/learn/full-adder/">08 Полный сумматор</a></li><li><a href="/ru/learn/ripple-carry/">09 Сквозной перенос</a></li></ol></li><li><h3>Модуль 4 · Умножение без умножителя</h3><p>Копия, ноль или переворот: произведение на один трит, умножение столбиком и MAC, который только складывает.</p><ol><li><a href="/ru/learn/times-a-trit/">10 Умножить на один трит</a></li><li><a href="/ru/learn/long-multiplication/">11 Умножение столбиком</a></li><li><a href="/ru/learn/ternary-mac/">12 Троичный MAC</a></li></ol></li><li><h3>Модуль 5 · Триты в двоичной памяти</h3><p>Два бита на трит, пять тритов на байт и сколько бит нужно слову из 27 тритов.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/two-bits-per-trit/">13 Два бита на трит</a></li><li><a href="/ru/learn/five-trits-a-byte/">14 Пять тритов в байте</a></li><li><a href="/ru/learn/a-word-of-27/">15 Слово из 27 тритов</a></li></ol></li><li><h3>Модуль 6 · Командное слово TRI-27</h3><p>32-битное слово, которое декодирует эмулятор Trinity: его поля, 47 опкодов и 15-битное непосредственное значение.</p><ol><li><a href="/ru/learn/the-instruction-word/">16 Командное слово</a></li><li><a href="/ru/learn/the-opcode-map/">17 Карта опкодов</a></li><li><a href="/ru/learn/the-immediate/">18 Непосредственное значение</a></li></ol></li><li><h3>Модуль 7 · Троичная машина</h3><p>АЛУ из сумматоров этого курса, переход на три стороны и программа, которую можно пройти по шагам.</p><ol><li><a href="/ru/learn/a-ternary-alu/">19 Троичное АЛУ</a></li><li><a href="/ru/learn/three-way-jump/">20 Переход на три стороны</a></li><li><a href="/ru/learn/run-a-program/">21 Запустить программу</a></li></ol></li><li><h3>Модуль 8 · Три ответа</h3><p>Сравнение по старшему различному триту, поиск числа по третям и сортировка трёхзначными сравнениями.</p><ol><li><a href="/ru/learn/compare-at-the-top/">22 Сравнение сверху</a></li><li><a href="/ru/learn/questions-in-thirds/">23 Вопросы по третям</a></li><li><a href="/ru/learn/sort-by-three/">24 Сортировка трёхзначными сравнениями</a></li></ol></li><li><h3>Модуль 9 · Троичные нейроны</h3><p>Веса -1, 0, +1: один нейрон, детектор и маленький слой, и нигде нет умножителя.</p><ol><li><a href="/ru/learn/one-ternary-neuron/">25 Один троичный нейрон</a></li><li><a href="/ru/learn/a-bar-detector/">26 Детектор полосы</a></li><li><a href="/ru/learn/a-tiny-layer/">27 Маленький слой</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/two-flop-sync/index.html b/apps/website/public/ru/learn/two-flop-sync/index.html
index 4d2155efa6..712a6652a9 100644
--- a/apps/website/public/ru/learn/two-flop-sync/index.html
+++ b/apps/website/public/ru/learn/two-flop-sync/index.html
@@ -300,5 +300,6 @@ <h1>Двухтриггерный синхронизатор</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое тактовый сигнал</h3><p>Один фронт, один мир: всё, что делит фронт тактового сигнала, делит и мир; период и джиттер настоящего фронта и место, где тактовый сигнал входит в плату.</p><ol><li><a href="/ru/learn/one-edge-one-world/">01 Один фронт, один мир</a></li><li><a href="/ru/learn/period-and-jitter/">02 Период, фронт и джиттер</a></li><li><a href="/ru/learn/where-the-clock-enters/">03 Где входит тактовый сигнал</a></li></ol></li><li><h3>Модуль 2 · Деревья тактового сигнала</h3><p>Перекос и задержка прохождения, глобальная сеть буферов и ловушка стробирования клока логикой.</p><ol><li><a href="/ru/learn/skew-and-insertion/">04 Перекос и задержка прохождения</a></li><li><a href="/ru/learn/global-buffers/">05 Глобальный буфер</a></li><li><a href="/ru/learn/gated-clock-trap/">06 Ловушка стробированного клока</a></li></ol></li><li><h3>Модуль 3 · PLL и MMCM</h3><p>Умножить и поделить один тактовый сигнал в другой, сдвинуть его фазу шагами VCO и какие тактовые сигналы анализатор считает родственными.</p><ol><li><a href="/ru/learn/multiply-and-divide/">07 Умножить и поделить</a></li><li><a href="/ru/learn/phase-and-lock/">08 Фаза и захват</a></li><li><a href="/ru/learn/related-or-not/">09 Родственные или чужие</a></li></ol></li><li><h3>Модуль 4 · Сбросы</h3><p>Ассертировать асинхронно, снимать синхронно: три вида сброса, конвейер снятия и дерево, которое растит сброс.</p><ol><li><a href="/ru/learn/sync-or-async/">10 Синхронный или асинхронный</a></li><li><a href="/ru/learn/reset-release/">11 Конвейер снятия</a></li><li><a href="/ru/learn/reset-fanout/">12 Разветвление сброса</a></li></ol></li><li><h3>Модуль 5 · Метастабильность</h3><p>Окно setup-hold, среднее время между отказами в целочисленной арифметике и два триггера, которые это чинят.</p><ol><li><a href="/ru/learn/setup-hold-window/">13 Окно setup-hold</a></li><li><a href="/ru/learn/mtbf/">14 Среднее время между отказами</a></li><li class="on" aria-current="page"><a href="/ru/learn/two-flop-sync/">15 Двухтриггерный синхронизатор</a></li></ol></li><li><h3>Модуль 6 · Пересечение многих бит</h3><p>Почему двоичная шина рвётся, почему код Грея нет, и рукопожатие, переносящее импульс между мирами.</p><ol><li><a href="/ru/learn/why-binary-fails/">16 Почему двоичный код рвётся</a></li><li><a href="/ru/learn/gray-code/">17 Код Грея</a></li><li><a href="/ru/learn/handshakes-and-pulses/">18 Рукопожатия и импульсы</a></li></ol></li><li><h3>Модуль 7 · Асинхронный FIFO</h3><p>Указатели, флаги и глубина: буфер, переносящий поток между двумя тактовыми сигналами.</p><ol><li><a href="/ru/learn/pointers-and-flags/">19 Указатели и флаги</a></li><li><a href="/ru/learn/depth-sizing/">20 Выбор глубины</a></li><li><a href="/ru/learn/fifo-testbench/">21 Тестбенч FIFO</a></li></ol></li><li><h3>Модуль 8 · Ограничения</h3><p>Строки, сообщающие анализатору, что такое тактовый сигнал, какие пути не проверять и чему должны соответствовать выводы.</p><ol><li><a href="/ru/learn/create-clock/">22 Строка create_clock</a></li><li><a href="/ru/learn/false-path-and-max-delay/">23 False path и max delay</a></li><li><a href="/ru/learn/io-timing/">24 Тайминг выводов</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Отчёт CDC, одно пересечение, захваченное на триггерах, и дифф битстрима, замыкающий курс.</p><ol><li><a href="/ru/learn/cdc-report/">25 Отчёт CDC</a></li><li><a href="/ru/learn/capture-a-crossing/">26 Захватить пересечение</a></li><li><a href="/ru/learn/capstone-diff/">27 Капстоун: поменяй и сравни</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/two-products-one-sum/index.html b/apps/website/public/ru/learn/two-products-one-sum/index.html
index 90d5de6f8f..0f37312e5a 100644
--- a/apps/website/public/ru/learn/two-products-one-sum/index.html
+++ b/apps/website/public/ru/learn/two-products-one-sum/index.html
@@ -145,5 +145,6 @@ <h1>Два произведения, одна сумма</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Правило и его числа</h3><p>Одно правило делит каждую ширину, отношение, к которому оно стремится, и числа Люка за тройкой 3.</p><ol><li><a href="/ru/learn/a-float-cut-by-phi/">01 Число с плавающей точкой, разрезанное по phi</a></li><li><a href="/ru/learn/phi-as-a-ratio/">02 Phi как отношение</a></li><li><a href="/ru/learn/lucas-numbers-stay-whole/">03 Числа Люка остаются целыми</a></li></ol></li><li><h3>Модуль 2 · Почему phi, почему три</h3><p>Почему деление идёт по phi, почему основание три и как спека проверяет, что GF16 хранит phi.</p><ol><li><a href="/ru/learn/why-the-split-is-phi/">04 Почему деление по phi</a></li><li><a href="/ru/learn/why-base-three/">05 Почему основание три</a></li><li><a href="/ru/learn/phi-in-sixteen-bits/">06 Phi в шестнадцати битах</a></li></ol></li><li><h3>Модуль 3 · Малые ступени: от GF4 до GF8</h3><p>GF4, GF6 и GF8 — меньше всего битов, и округление до целых битов стоит здесь дороже всего.</p><ol><li><a href="/ru/learn/gf4-four-bits/">07 GF4: четыре бита</a></li><li><a href="/ru/learn/gf6-six-bits/">08 GF6: шесть битов</a></li><li><a href="/ru/learn/gf8-one-byte/">09 GF8: один байт</a></li></ol></li><li><h3>Модуль 4 · От десяти до четырнадцати битов</h3><p>GF10, GF12 и GF14 и то, как расстояние до 1 / phi меняется с ростом слова.</p><ol><li><a href="/ru/learn/gf10-ten-bits/">10 GF10: десять битов</a></li><li><a href="/ru/learn/gf12-twelve-bits/">11 GF12: двенадцать битов</a></li><li><a href="/ru/learn/gf14-fourteen-bits/">12 GF14: четырнадцать битов</a></li></ol></li><li><h3>Модуль 5 · GF16 в работе</h3><p>Основной 16-битный формат, скалярное произведение из двух слагаемых в GF-T16, затем GF20 и GF24.</p><ol><li><a href="/ru/learn/gf16-the-primary/">13 GF16: основной формат</a></li><li class="on" aria-current="page"><a href="/ru/learn/two-products-one-sum/">14 Два произведения, одна сумма</a></li><li><a href="/ru/learn/gf24-twenty-four-bits/">15 GF20 и GF24</a></li></ol></li><li><h3>Модуль 6 · От GF32 до GF64</h3><p>GF32 рядом с IEEE single, GF48 без пары в IEEE, GF64 рядом с IEEE double.</p><ol><li><a href="/ru/learn/gf32-a-single/">16 GF32: одинарная точность</a></li><li><a href="/ru/learn/gf48-forty-eight-bits/">17 GF48: сорок восемь битов</a></li><li><a href="/ru/learn/gf64-a-double/">18 GF64: двойная точность</a></li></ol></li><li><h3>Модуль 7 · От GF96 до GF256</h3><p>GF96, GF128 и GF256, где спеки держат раскладку инвариантами.</p><ol><li><a href="/ru/learn/gf96-ninety-six-bits/">19 GF96: девяносто шесть битов</a></li><li><a href="/ru/learn/gf128-a-quad/">20 GF128: четверная точность</a></li><li><a href="/ru/learn/gf256-two-five-six/">21 GF256: двести пятьдесят шесть битов</a></li></ol></li><li><h3>Модуль 8 · Самые широкие ступени, затем триты</h3><p>GF512 и GF1024, две самые широкие ступени, затем GF-T8, где порядок уходит в триты.</p><ol><li><a href="/ru/learn/gf512-five-twelve/">22 GF512: пятьсот двенадцать битов</a></li><li><a href="/ru/learn/gf1024-one-kilobit/">23 GF1024: один килобит</a></li><li><a href="/ru/learn/gft8-exponent-in-trits/">24 GF-T8: порядок в тритах</a></li></ol></li><li><h3>Модуль 9 · Ещё триты, затем декодирование</h3><p>GF-T16 и GF-T32, затем почему фиксированные поля декодируются параллельно, а posit — нет.</p><ol><li><a href="/ru/learn/gft16-six-trits/">25 GF-T16: шесть тритов</a></li><li><a href="/ru/learn/gft32-twelve-trits/">26 GF-T32: двенадцать тритов</a></li><li><a href="/ru/learn/decode-in-one-step/">27 Раскодировать за один шаг</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/two-to-the-x/index.html b/apps/website/public/ru/learn/two-to-the-x/index.html
index 9ee9ddf5a9..99cddcfeb0 100644
--- a/apps/website/public/ru/learn/two-to-the-x/index.html
+++ b/apps/website/public/ru/learn/two-to-the-x/index.html
@@ -137,5 +137,6 @@ <h1>Два в степени x</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Лаборатория: наши исследования</h3><p>Свой формат чисел, честная таблица результатов и таблицы модели, перемноженные на плате.</p><ol><li><a href="/ru/learn/t27-number-format/">01 T27: наш собственный формат чисел</a></li><li><a href="/ru/learn/an-honest-scoreboard/">02 Честная таблица результатов</a></li><li><a href="/ru/learn/a-model-on-the-board/">03 Таблицы модели на плате</a></li></ol></li><li><h3>Модуль 2 · ИИ-числа: блок MX</h3><p>Как ИИ-чипы хранят веса в нескольких битах: один общий масштаб на блок, сам байт масштаба и что один выброс делает с соседями.</p><ol><li><a href="/ru/learn/one-scale-per-block/">04 Один масштаб на блок весов</a></li><li><a href="/ru/learn/scale-byte-native/">05 Байт масштаба на настоящей машине</a></li><li><a href="/ru/learn/one-outlier/">06 Один выброс, много нулей</a></li></ol></li><li><h3>Модуль 3 · Тернарные веса</h3><p>Веса, которые бывают только минус масштаб, ноль или плюс масштаб, пять правил, которые должен пройти тернарный алфавит, и прогон тестов, который ничего не проверил.</p><ol><li><a href="/ru/learn/phi-scaled-ternary/">07 Веса минус фи, ноль и плюс фи</a></li><li><a href="/ru/learn/a-pass-that-checked-nothing/">08 Прогон, который ничего не проверил</a></li><li><a href="/ru/learn/five-rules-for-weights/">09 Пять правил для алфавита весов</a></li></ol></li><li><h3>Модуль 4 · Тернарный формат Ternary Network Float</h3><p>Правило, которое компилятор проверяет до запуска любого теста, и 17-битное число с плавающей точкой, порядок которого — четыре сбалансированных трита.</p><ol><li><a href="/ru/learn/blocked-is-not-failed/">10 Заблокирован — не значит провален</a></li><li><a href="/ru/learn/four-trit-exponent/">11 Число, у которого порядок — четыре трита</a></li><li><a href="/ru/learn/negate-twice/">12 Дважды сменить знак — получить число обратно</a></li></ol></li><li><h3>Модуль 5 · Арифметика чисел со знаком</h3><p>Умножить два числа со знаком, сложить их, когда знаки разные, и сделать то и другое сразу в умножении с накоплением.</p><ol><li><a href="/ru/learn/a-sign-is-one-bit/">13 Знак — это один бит</a></li><li><a href="/ru/learn/opposite-signs-subtract/">14 Разные знаки — вычитание</a></li><li><a href="/ru/learn/multiply-then-add/">15 Умножить, потом сложить</a></li></ol></li><li><h3>Модуль 6 · Части нейрона</h3><p>ReLU с изломом в нуле, степень двойки для softmax и argmax, который называет ответ.</p><ol><li><a href="/ru/learn/a-bend-at-zero/">16 Излом в нуле</a></li><li class="on" aria-current="page"><a href="/ru/learn/two-to-the-x/">17 Два в степени x</a></li><li><a href="/ru/learn/pick-the-largest/">18 Выбрать наибольшее</a></li></ol></li><li><h3>Модуль 7 · Обучение на ошибке</h3><p>Потеря, которая оценивает неверную догадку в битах, один шаг, который сдвигает вес против градиента, и скрытый слой, который нужен XOR.</p><ol><li><a href="/ru/learn/a-loss-in-bits/">19 Потеря в битах</a></li><li><a href="/ru/learn/one-step-downhill/">20 Один шаг вниз по склону</a></li><li><a href="/ru/learn/xor-needs-a-bend/">21 XOR нужен излом</a></li></ol></li><li><h3>Модуль 8 · BitNet: тернарные сети</h3><p>Порог, который сжимает сумму обратно до трёх значений, один нейрон, который становится другой функцией при смене весов, и нейрон, который читает входы по 27 тритов за раз.</p><ol><li><a href="/ru/learn/back-to-three-values/">22 Обратно в три значения</a></li><li><a href="/ru/learn/same-neuron-new-weights/">23 Тот же нейрон, другие веса</a></li><li><a href="/ru/learn/a-neuron-in-chunks/">24 Нейрон по кускам</a></li></ol></li><li><h3>Модуль 9 · Тернарный MAC как чип</h3><p>Скалярное произведение 27 тритов как провода без регистра, та же сумма, которую регистр копит на каждом такте, и небольшая целая сеть в конце курса.</p><ol><li><a href="/ru/learn/a-dot-product-in-wires/">25 Скалярное произведение из проводов</a></li><li><a href="/ru/learn/add-it-up-every-clock/">26 Складывать на каждом такте</a></li><li><a href="/ru/learn/a-whole-network/">27 Целая сеть</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/use-other-modules/index.html b/apps/website/public/ru/learn/use-other-modules/index.html
index 9452161910..c073e8bc94 100644
--- a/apps/website/public/ru/learn/use-other-modules/index.html
+++ b/apps/website/public/ru/learn/use-other-modules/index.html
@@ -83,5 +83,6 @@ <h1>Другие модули</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Файл спеки</h3><p>Что лежит в файле .t27: строка module, строки прозы и комментарии.</p><ol><li><a href="/ru/learn/what-a-spec-is/">01 Что такое спека</a></li><li><a href="/ru/learn/the-module-line/">02 Строка module</a></li><li><a href="/ru/learn/comments-and-prose/">03 Комментарии и строки прозы</a></li></ol></li><li><h3>Модуль 2 · Значения и типы</h3><p>Константы, ширина целого числа, а также true, false и текст.</p><ol><li><a href="/ru/learn/constants/">04 Константы</a></li><li><a href="/ru/learn/whole-numbers/">05 Целые числа и их ширина</a></li><li><a href="/ru/learn/true-false-and-text/">06 Истина, ложь и текст</a></li></ol></li><li><h3>Модуль 3 · Массивы, триты и выражения</h3><p>Списки значений, три значения трита и что вычисляют операторы.</p><ol><li><a href="/ru/learn/arrays/">07 Массивы</a></li><li><a href="/ru/learn/trits/">08 Триты: минус один, ноль, один</a></li><li><a href="/ru/learn/expressions/">09 Выражения</a></li></ol></li><li><h3>Модуль 4 · Тесты и функции</h3><p>Блок test, несколько маленьких тестов в одной спеке и функция с типизированными входами.</p><ol><li><a href="/ru/learn/a-test-block/">10 Блок test</a></li><li><a href="/ru/learn/many-tests/">11 Много тестов в одной спеке</a></li><li><a href="/ru/learn/functions/">12 Функции</a></li></ol></li><li><h3>Модуль 5 · Внутри функции</h3><p>Имена через let и var, выбор через if и switch и циклы.</p><ol><li><a href="/ru/learn/local-names/">13 Локальные имена: let и var</a></li><li><a href="/ru/learn/if-and-else/">14 Выбор: if, else и switch</a></li><li><a href="/ru/learn/loops/">15 Циклы: while и for</a></li></ol></li><li><h3>Модуль 6 · Видимость и свои типы</h3><p>Что pub открывает другим модулям, и структуры и перечисления, которые вы объявляете.</p><ol><li><a href="/ru/learn/pub-or-private/">16 pub или закрытое имя</a></li><li><a href="/ru/learn/structs/">17 Структуры</a></li><li><a href="/ru/learn/enums/">18 Перечисления</a></li></ol></li><li><h3>Модуль 7 · Модули, правила и gen-ts</h3><p>Чтение другого модуля, правило, которое выполняется всегда, и спека, превращённая в TypeScript.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/use-other-modules/">19 Другие модули</a></li><li><a href="/ru/learn/invariants/">20 Инварианты и замеры</a></li><li><a href="/ru/learn/gen-ts/">21 gen-ts: спека превращается в TypeScript</a></li></ol></li><li><h3>Модуль 8 · Компилятор и tri</h3><p>Бэкенды, которые пишет t27c, стадии, в которых он читает спеку, и команда tri.</p><ol><li><a href="/ru/learn/seven-backends/">22 Семь бэкендов</a></li><li><a href="/ru/learn/stages-of-t27c/">23 Стадии t27c</a></li><li><a href="/ru/learn/tri-commands/">24 tri: командная строка</a></li></ol></li><li><h3>Модуль 9 · Ошибки, программа и что дальше</h3><p>Как читать ошибку компилятора, одна маленькая законченная программа и какой курс пройти дальше.</p><ol><li><a href="/ru/learn/reading-errors/">25 Как читать ошибки компилятора</a></li><li><a href="/ru/learn/a-small-program/">26 Маленькая законченная программа</a></li><li><a href="/ru/learn/where-next/">27 Куда дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/vcd-format/index.html b/apps/website/public/ru/learn/vcd-format/index.html
index 8a540e265a..c5229998a5 100644
--- a/apps/website/public/ru/learn/vcd-format/index.html
+++ b/apps/website/public/ru/learn/vcd-format/index.html
@@ -362,5 +362,6 @@ <h1>Формат VCD</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Зачем проверять</h3><p>Дизайны, которые компилируются и ошибаются; модель, которая выносит вердикт; план, записанный до кода.</p><ol><li><a href="/ru/learn/bugs-that-compile/">01 Баги, которые компилируются</a></li><li><a href="/ru/learn/golden-model/">02 Золотая модель</a></li><li><a href="/ru/learn/plan-before-code/">03 Сначала план</a></li></ol></li><li><h3>Модуль 2 · Тестбенчи</h3><p>Стимулы, проверки и вердикт, записанные как один spec рядом с дизайном, который они судят.</p><ol><li><a href="/ru/learn/anatomy-of-a-tb/">04 Анатомия тестбенча</a></li><li><a href="/ru/learn/self-checking/">05 Самопроверка вместо всматривания</a></li><li><a href="/ru/learn/directed-vs-random/">06 Направленные против случайных</a></li></ol></li><li><h3>Модуль 3 · Временные диаграммы</h3><p>Трасса каждого сигнала, прочитанная так, как её читает инженер по железу, и два прогона, сравнённые между собой.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/vcd-format/">07 Формат VCD</a></li><li><a href="/ru/learn/debug-with-waves/">08 Отладка по волнам</a></li><li><a href="/ru/learn/compare-traces/">09 Сравнение двух трасс</a></li></ol></li><li><h3>Модуль 4 · Векторы соответствия</h3><p>Случаи с ответом, записанным рядом, там, откуда их достанет компилятор.</p><ol><li><a href="/ru/learn/what-a-vector-is/">10 Что такое вектор</a></li><li><a href="/ru/learn/golden-ruler-vectors/">11 Векторы как эталонная линейка</a></li><li><a href="/ru/learn/end-to-end-scenarios/">12 Сквозные сценарии</a></li></ol></li><li><h3>Модуль 5 · Косимуляция</h3><p>Spec, симулятор и плата сходятся в одном ответе на стенде Artix-7 XC7A200T, и что делать, когда не сходятся.</p><ol><li><a href="/ru/learn/three-way-agreement/">13 Тройное согласие</a></li><li><a href="/ru/learn/simulator-internals/">14 Внутри симулятора</a></li><li><a href="/ru/learn/when-they-disagree/">15 Когда они не сходятся</a></li></ol></li><li><h3>Модуль 6 · Покрытие</h3><p>Чего коснулись тесты: строки, переключения, состояния — и что прячет это число.</p><ol><li><a href="/ru/learn/line-and-toggle/">16 Строки и переключения</a></li><li><a href="/ru/learn/fsm-coverage/">17 Покрытие автоматов</a></li><li><a href="/ru/learn/coverage-lies/">18 Покрытие лжёт</a></li></ol></li><li><h3>Модуль 7 · Формальные методы</h3><p>Ассерты, верные каждый такт; ограниченный поиск контрпримера; и почему доказательству нужна индукция.</p><ol><li><a href="/ru/learn/assertions/">19 Ассерты</a></li><li><a href="/ru/learn/bounded-model-checking/">20 Ограниченная проверка моделей</a></li><li><a href="/ru/learn/induction/">21 Индукция</a></li></ol></li><li><h3>Модуль 8 · Мутации</h3><p>Ломайте дизайн нарочно и считайте, что заметили тесты.</p><ol><li><a href="/ru/learn/kill-the-mutant/">22 Убить мутанта</a></li><li><a href="/ru/learn/surviving-mutants/">23 Выжившие мутанты</a></li><li><a href="/ru/learn/test-strength/">24 Сила тестов</a></li></ol></li><li><h3>Модуль 9 · Приёмка</h3><p>Одна команда, все квитанции, чистый вердикт, который можно показать.</p><ol><li><a href="/ru/learn/build-verify/">25 Сборка и проверка</a></li><li><a href="/ru/learn/board-selftest/">26 Стенд проверяет себя</a></li><li><a href="/ru/learn/verify-capstone/">27 Капстоун: проверить модуль целиком</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">Этот курс сам является spec t27: прочитайте его</a></p><p>Виджеты сохраняют свои английские слова: каждый живёт в собственном spec t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/verify-capstone/index.html b/apps/website/public/ru/learn/verify-capstone/index.html
index 473bcae9be..fdcd3d5072 100644
--- a/apps/website/public/ru/learn/verify-capstone/index.html
+++ b/apps/website/public/ru/learn/verify-capstone/index.html
@@ -202,5 +202,6 @@ <h1>Капстоун: проверить модуль целиком</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Зачем проверять</h3><p>Дизайны, которые компилируются и ошибаются; модель, которая выносит вердикт; план, записанный до кода.</p><ol><li><a href="/ru/learn/bugs-that-compile/">01 Баги, которые компилируются</a></li><li><a href="/ru/learn/golden-model/">02 Золотая модель</a></li><li><a href="/ru/learn/plan-before-code/">03 Сначала план</a></li></ol></li><li><h3>Модуль 2 · Тестбенчи</h3><p>Стимулы, проверки и вердикт, записанные как один spec рядом с дизайном, который они судят.</p><ol><li><a href="/ru/learn/anatomy-of-a-tb/">04 Анатомия тестбенча</a></li><li><a href="/ru/learn/self-checking/">05 Самопроверка вместо всматривания</a></li><li><a href="/ru/learn/directed-vs-random/">06 Направленные против случайных</a></li></ol></li><li><h3>Модуль 3 · Временные диаграммы</h3><p>Трасса каждого сигнала, прочитанная так, как её читает инженер по железу, и два прогона, сравнённые между собой.</p><ol><li><a href="/ru/learn/vcd-format/">07 Формат VCD</a></li><li><a href="/ru/learn/debug-with-waves/">08 Отладка по волнам</a></li><li><a href="/ru/learn/compare-traces/">09 Сравнение двух трасс</a></li></ol></li><li><h3>Модуль 4 · Векторы соответствия</h3><p>Случаи с ответом, записанным рядом, там, откуда их достанет компилятор.</p><ol><li><a href="/ru/learn/what-a-vector-is/">10 Что такое вектор</a></li><li><a href="/ru/learn/golden-ruler-vectors/">11 Векторы как эталонная линейка</a></li><li><a href="/ru/learn/end-to-end-scenarios/">12 Сквозные сценарии</a></li></ol></li><li><h3>Модуль 5 · Косимуляция</h3><p>Spec, симулятор и плата сходятся в одном ответе на стенде Artix-7 XC7A200T, и что делать, когда не сходятся.</p><ol><li><a href="/ru/learn/three-way-agreement/">13 Тройное согласие</a></li><li><a href="/ru/learn/simulator-internals/">14 Внутри симулятора</a></li><li><a href="/ru/learn/when-they-disagree/">15 Когда они не сходятся</a></li></ol></li><li><h3>Модуль 6 · Покрытие</h3><p>Чего коснулись тесты: строки, переключения, состояния — и что прячет это число.</p><ol><li><a href="/ru/learn/line-and-toggle/">16 Строки и переключения</a></li><li><a href="/ru/learn/fsm-coverage/">17 Покрытие автоматов</a></li><li><a href="/ru/learn/coverage-lies/">18 Покрытие лжёт</a></li></ol></li><li><h3>Модуль 7 · Формальные методы</h3><p>Ассерты, верные каждый такт; ограниченный поиск контрпримера; и почему доказательству нужна индукция.</p><ol><li><a href="/ru/learn/assertions/">19 Ассерты</a></li><li><a href="/ru/learn/bounded-model-checking/">20 Ограниченная проверка моделей</a></li><li><a href="/ru/learn/induction/">21 Индукция</a></li></ol></li><li><h3>Модуль 8 · Мутации</h3><p>Ломайте дизайн нарочно и считайте, что заметили тесты.</p><ol><li><a href="/ru/learn/kill-the-mutant/">22 Убить мутанта</a></li><li><a href="/ru/learn/surviving-mutants/">23 Выжившие мутанты</a></li><li><a href="/ru/learn/test-strength/">24 Сила тестов</a></li></ol></li><li><h3>Модуль 9 · Приёмка</h3><p>Одна команда, все квитанции, чистый вердикт, который можно показать.</p><ol><li><a href="/ru/learn/build-verify/">25 Сборка и проверка</a></li><li><a href="/ru/learn/board-selftest/">26 Стенд проверяет себя</a></li><li class="on" aria-current="page"><a href="/ru/learn/verify-capstone/">27 Капстоун: проверить модуль целиком</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">Этот курс сам является spec t27: прочитайте его</a></p><p>Виджеты сохраняют свои английские слова: каждый живёт в собственном spec t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/verify-hardware/index.html b/apps/website/public/ru/learn/verify-hardware/index.html
index fc281bb8e9..0fce245718 100644
--- a/apps/website/public/ru/learn/verify-hardware/index.html
+++ b/apps/website/public/ru/learn/verify-hardware/index.html
@@ -67,5 +67,6 @@ <h1>Проверка железа с t27: тестбенчи, временные
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Зачем проверять</h3><p>Дизайны, которые компилируются и ошибаются; модель, которая выносит вердикт; план, записанный до кода.</p><ol><li><a href="/ru/learn/bugs-that-compile/">01 Баги, которые компилируются</a></li><li><a href="/ru/learn/golden-model/">02 Золотая модель</a></li><li><a href="/ru/learn/plan-before-code/">03 Сначала план</a></li></ol></li><li><h3>Модуль 2 · Тестбенчи</h3><p>Стимулы, проверки и вердикт, записанные как один spec рядом с дизайном, который они судят.</p><ol><li><a href="/ru/learn/anatomy-of-a-tb/">04 Анатомия тестбенча</a></li><li><a href="/ru/learn/self-checking/">05 Самопроверка вместо всматривания</a></li><li><a href="/ru/learn/directed-vs-random/">06 Направленные против случайных</a></li></ol></li><li><h3>Модуль 3 · Временные диаграммы</h3><p>Трасса каждого сигнала, прочитанная так, как её читает инженер по железу, и два прогона, сравнённые между собой.</p><ol><li><a href="/ru/learn/vcd-format/">07 Формат VCD</a></li><li><a href="/ru/learn/debug-with-waves/">08 Отладка по волнам</a></li><li><a href="/ru/learn/compare-traces/">09 Сравнение двух трасс</a></li></ol></li><li><h3>Модуль 4 · Векторы соответствия</h3><p>Случаи с ответом, записанным рядом, там, откуда их достанет компилятор.</p><ol><li><a href="/ru/learn/what-a-vector-is/">10 Что такое вектор</a></li><li><a href="/ru/learn/golden-ruler-vectors/">11 Векторы как эталонная линейка</a></li><li><a href="/ru/learn/end-to-end-scenarios/">12 Сквозные сценарии</a></li></ol></li><li><h3>Модуль 5 · Косимуляция</h3><p>Spec, симулятор и плата сходятся в одном ответе на стенде Artix-7 XC7A200T, и что делать, когда не сходятся.</p><ol><li><a href="/ru/learn/three-way-agreement/">13 Тройное согласие</a></li><li><a href="/ru/learn/simulator-internals/">14 Внутри симулятора</a></li><li><a href="/ru/learn/when-they-disagree/">15 Когда они не сходятся</a></li></ol></li><li><h3>Модуль 6 · Покрытие</h3><p>Чего коснулись тесты: строки, переключения, состояния — и что прячет это число.</p><ol><li><a href="/ru/learn/line-and-toggle/">16 Строки и переключения</a></li><li><a href="/ru/learn/fsm-coverage/">17 Покрытие автоматов</a></li><li><a href="/ru/learn/coverage-lies/">18 Покрытие лжёт</a></li></ol></li><li><h3>Модуль 7 · Формальные методы</h3><p>Ассерты, верные каждый такт; ограниченный поиск контрпримера; и почему доказательству нужна индукция.</p><ol><li><a href="/ru/learn/assertions/">19 Ассерты</a></li><li><a href="/ru/learn/bounded-model-checking/">20 Ограниченная проверка моделей</a></li><li><a href="/ru/learn/induction/">21 Индукция</a></li></ol></li><li><h3>Модуль 8 · Мутации</h3><p>Ломайте дизайн нарочно и считайте, что заметили тесты.</p><ol><li><a href="/ru/learn/kill-the-mutant/">22 Убить мутанта</a></li><li><a href="/ru/learn/surviving-mutants/">23 Выжившие мутанты</a></li><li><a href="/ru/learn/test-strength/">24 Сила тестов</a></li></ol></li><li><h3>Модуль 9 · Приёмка</h3><p>Одна команда, все квитанции, чистый вердикт, который можно показать.</p><ol><li><a href="/ru/learn/build-verify/">25 Сборка и проверка</a></li><li><a href="/ru/learn/board-selftest/">26 Стенд проверяет себя</a></li><li><a href="/ru/learn/verify-capstone/">27 Капстоун: проверить модуль целиком</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">Этот курс сам является spec t27: прочитайте его</a></p><p>Виджеты сохраняют свои английские слова: каждый живёт в собственном spec t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/what-a-spec-is/index.html b/apps/website/public/ru/learn/what-a-spec-is/index.html
index e91355ffd6..35dcd06f05 100644
--- a/apps/website/public/ru/learn/what-a-spec-is/index.html
+++ b/apps/website/public/ru/learn/what-a-spec-is/index.html
@@ -83,5 +83,6 @@ <h1>Что такое спека</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Файл спеки</h3><p>Что лежит в файле .t27: строка module, строки прозы и комментарии.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/what-a-spec-is/">01 Что такое спека</a></li><li><a href="/ru/learn/the-module-line/">02 Строка module</a></li><li><a href="/ru/learn/comments-and-prose/">03 Комментарии и строки прозы</a></li></ol></li><li><h3>Модуль 2 · Значения и типы</h3><p>Константы, ширина целого числа, а также true, false и текст.</p><ol><li><a href="/ru/learn/constants/">04 Константы</a></li><li><a href="/ru/learn/whole-numbers/">05 Целые числа и их ширина</a></li><li><a href="/ru/learn/true-false-and-text/">06 Истина, ложь и текст</a></li></ol></li><li><h3>Модуль 3 · Массивы, триты и выражения</h3><p>Списки значений, три значения трита и что вычисляют операторы.</p><ol><li><a href="/ru/learn/arrays/">07 Массивы</a></li><li><a href="/ru/learn/trits/">08 Триты: минус один, ноль, один</a></li><li><a href="/ru/learn/expressions/">09 Выражения</a></li></ol></li><li><h3>Модуль 4 · Тесты и функции</h3><p>Блок test, несколько маленьких тестов в одной спеке и функция с типизированными входами.</p><ol><li><a href="/ru/learn/a-test-block/">10 Блок test</a></li><li><a href="/ru/learn/many-tests/">11 Много тестов в одной спеке</a></li><li><a href="/ru/learn/functions/">12 Функции</a></li></ol></li><li><h3>Модуль 5 · Внутри функции</h3><p>Имена через let и var, выбор через if и switch и циклы.</p><ol><li><a href="/ru/learn/local-names/">13 Локальные имена: let и var</a></li><li><a href="/ru/learn/if-and-else/">14 Выбор: if, else и switch</a></li><li><a href="/ru/learn/loops/">15 Циклы: while и for</a></li></ol></li><li><h3>Модуль 6 · Видимость и свои типы</h3><p>Что pub открывает другим модулям, и структуры и перечисления, которые вы объявляете.</p><ol><li><a href="/ru/learn/pub-or-private/">16 pub или закрытое имя</a></li><li><a href="/ru/learn/structs/">17 Структуры</a></li><li><a href="/ru/learn/enums/">18 Перечисления</a></li></ol></li><li><h3>Модуль 7 · Модули, правила и gen-ts</h3><p>Чтение другого модуля, правило, которое выполняется всегда, и спека, превращённая в TypeScript.</p><ol><li><a href="/ru/learn/use-other-modules/">19 Другие модули</a></li><li><a href="/ru/learn/invariants/">20 Инварианты и замеры</a></li><li><a href="/ru/learn/gen-ts/">21 gen-ts: спека превращается в TypeScript</a></li></ol></li><li><h3>Модуль 8 · Компилятор и tri</h3><p>Бэкенды, которые пишет t27c, стадии, в которых он читает спеку, и команда tri.</p><ol><li><a href="/ru/learn/seven-backends/">22 Семь бэкендов</a></li><li><a href="/ru/learn/stages-of-t27c/">23 Стадии t27c</a></li><li><a href="/ru/learn/tri-commands/">24 tri: командная строка</a></li></ol></li><li><h3>Модуль 9 · Ошибки, программа и что дальше</h3><p>Как читать ошибку компилятора, одна маленькая законченная программа и какой курс пройти дальше.</p><ol><li><a href="/ru/learn/reading-errors/">25 Как читать ошибки компилятора</a></li><li><a href="/ru/learn/a-small-program/">26 Маленькая законченная программа</a></li><li><a href="/ru/learn/where-next/">27 Куда дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/what-a-vector-is/index.html b/apps/website/public/ru/learn/what-a-vector-is/index.html
index 4d28d3fdba..e7810f7160 100644
--- a/apps/website/public/ru/learn/what-a-vector-is/index.html
+++ b/apps/website/public/ru/learn/what-a-vector-is/index.html
@@ -565,5 +565,6 @@ <h1>Что такое вектор</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Зачем проверять</h3><p>Дизайны, которые компилируются и ошибаются; модель, которая выносит вердикт; план, записанный до кода.</p><ol><li><a href="/ru/learn/bugs-that-compile/">01 Баги, которые компилируются</a></li><li><a href="/ru/learn/golden-model/">02 Золотая модель</a></li><li><a href="/ru/learn/plan-before-code/">03 Сначала план</a></li></ol></li><li><h3>Модуль 2 · Тестбенчи</h3><p>Стимулы, проверки и вердикт, записанные как один spec рядом с дизайном, который они судят.</p><ol><li><a href="/ru/learn/anatomy-of-a-tb/">04 Анатомия тестбенча</a></li><li><a href="/ru/learn/self-checking/">05 Самопроверка вместо всматривания</a></li><li><a href="/ru/learn/directed-vs-random/">06 Направленные против случайных</a></li></ol></li><li><h3>Модуль 3 · Временные диаграммы</h3><p>Трасса каждого сигнала, прочитанная так, как её читает инженер по железу, и два прогона, сравнённые между собой.</p><ol><li><a href="/ru/learn/vcd-format/">07 Формат VCD</a></li><li><a href="/ru/learn/debug-with-waves/">08 Отладка по волнам</a></li><li><a href="/ru/learn/compare-traces/">09 Сравнение двух трасс</a></li></ol></li><li><h3>Модуль 4 · Векторы соответствия</h3><p>Случаи с ответом, записанным рядом, там, откуда их достанет компилятор.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/what-a-vector-is/">10 Что такое вектор</a></li><li><a href="/ru/learn/golden-ruler-vectors/">11 Векторы как эталонная линейка</a></li><li><a href="/ru/learn/end-to-end-scenarios/">12 Сквозные сценарии</a></li></ol></li><li><h3>Модуль 5 · Косимуляция</h3><p>Spec, симулятор и плата сходятся в одном ответе на стенде Artix-7 XC7A200T, и что делать, когда не сходятся.</p><ol><li><a href="/ru/learn/three-way-agreement/">13 Тройное согласие</a></li><li><a href="/ru/learn/simulator-internals/">14 Внутри симулятора</a></li><li><a href="/ru/learn/when-they-disagree/">15 Когда они не сходятся</a></li></ol></li><li><h3>Модуль 6 · Покрытие</h3><p>Чего коснулись тесты: строки, переключения, состояния — и что прячет это число.</p><ol><li><a href="/ru/learn/line-and-toggle/">16 Строки и переключения</a></li><li><a href="/ru/learn/fsm-coverage/">17 Покрытие автоматов</a></li><li><a href="/ru/learn/coverage-lies/">18 Покрытие лжёт</a></li></ol></li><li><h3>Модуль 7 · Формальные методы</h3><p>Ассерты, верные каждый такт; ограниченный поиск контрпримера; и почему доказательству нужна индукция.</p><ol><li><a href="/ru/learn/assertions/">19 Ассерты</a></li><li><a href="/ru/learn/bounded-model-checking/">20 Ограниченная проверка моделей</a></li><li><a href="/ru/learn/induction/">21 Индукция</a></li></ol></li><li><h3>Модуль 8 · Мутации</h3><p>Ломайте дизайн нарочно и считайте, что заметили тесты.</p><ol><li><a href="/ru/learn/kill-the-mutant/">22 Убить мутанта</a></li><li><a href="/ru/learn/surviving-mutants/">23 Выжившие мутанты</a></li><li><a href="/ru/learn/test-strength/">24 Сила тестов</a></li></ol></li><li><h3>Модуль 9 · Приёмка</h3><p>Одна команда, все квитанции, чистый вердикт, который можно показать.</p><ol><li><a href="/ru/learn/build-verify/">25 Сборка и проверка</a></li><li><a href="/ru/learn/board-selftest/">26 Стенд проверяет себя</a></li><li><a href="/ru/learn/verify-capstone/">27 Капстоун: проверить модуль целиком</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">Этот курс сам является spec t27: прочитайте его</a></p><p>Виджеты сохраняют свои английские слова: каждый живёт в собственном spec t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/what-arithmetic-costs/index.html b/apps/website/public/ru/learn/what-arithmetic-costs/index.html
index dac281ed03..b574b02d73 100644
--- a/apps/website/public/ru/learn/what-arithmetic-costs/index.html
+++ b/apps/website/public/ru/learn/what-arithmetic-costs/index.html
@@ -185,5 +185,6 @@ <h1>Сколько стоит арифметика</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Чип</h3><p>Что такое FPGA, какой чип у нас и как его выводы встречаются с платой.</p><ol><li><a href="/ru/learn/fpga-is-a-grid/">01 FPGA — это сетка, которую вы настраиваете</a></li><li><a href="/ru/learn/which-chip/">02 Какой это чип?</a></li><li><a href="/ru/learn/pins-and-package/">03 Выводы, шарики и корпус</a></li></ol></li><li><h3>Модуль 2 · Числа в железе</h3><p>Биты, триты и форматы чисел, и сколько стоит арифметика в логике.</p><ol><li><a href="/ru/learn/bits-and-trits/">04 Биты, триты и GF16</a></li><li class="on" aria-current="page"><a href="/ru/learn/what-arithmetic-costs/">05 Сколько стоит арифметика</a></li><li><a href="/ru/learn/how-much-fits/">06 Сколько помещается на чип</a></li></ol></li><li><h3>Модуль 3 · Ваша программа на t27</h3><p>Напишите спеку, проверьте её и увидьте, почему «компилируется» не значит «верно».</p><ol><li><a href="/ru/learn/hello-t27/">07 Привет, t27</a></li><li><a href="/ru/learn/tests-are-the-spec/">08 Тесты — часть спеки</a></li><li><a href="/ru/learn/compiles-is-not-correct/">09 Компилируется — не значит верно</a></li></ol></li><li><h3>Модуль 4 · Внутри t27c</h3><p>Как компилятор читает спеку и что он пишет, включая нативный t27b.</p><ol><li><a href="/ru/learn/compiler-stages/">10 Стадии t27c</a></li><li><a href="/ru/learn/t27c-on-a-real-machine/">11 t27c на настоящей машине</a></li><li><a href="/ru/learn/seven-backends-and-t27b/">12 Семь бэкендов и t27b</a></li></ol></li><li><h3>Модуль 5 · От спеки к железу</h3><p>Verilog, который пишет t27c, ячейки, в которые он превращается, и что делает одна LUT.</p><ol><li><a href="/ru/learn/spec-to-verilog/">13 От спеки к Verilog</a></li><li><a href="/ru/learn/count-the-cells/">14 Посчитайте ячейки</a></li><li><a href="/ru/learn/one-lut-up-close/">15 Одна LUT вблизи</a></li></ol></li><li><h3>Модуль 6 · Читаем синтез</h3><p>Что yosys сообщает о вашем дизайне и какие предупреждения важны.</p><ol><li><a href="/ru/learn/synthesis-warnings/">16 Важные предупреждения синтеза</a></li><li><a href="/ru/learn/state-machines/">17 Конечные автоматы</a></li><li><a href="/ru/learn/where-the-luts-go/">18 Куда уходят LUT</a></li></ol></li><li><h3>Модуль 7 · Размещение, трассировка, тайминг</h3><p>Где ячейки оказываются на кристалле и успевает ли тактовая частота.</p><ol><li><a href="/ru/learn/placement/">19 Размещение: где оказываются ячейки</a></li><li><a href="/ru/learn/logic-depth/">20 Глубина логики</a></li><li><a href="/ru/learn/timing-slack/">21 Тайминг и запас</a></li></ol></li><li><h3>Модуль 8 · Битстрим</h3><p>Как оттрассированный дизайн становится битами, которые загружает чип, без Vivado.</p><ol><li><a href="/ru/learn/fasm-to-frames/">22 От FASM к фреймам</a></li><li><a href="/ru/learn/a-build-receipt/">23 Сборка по пунктам</a></li><li><a href="/ru/learn/read-a-bitstream/">24 Читаем битстрим</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Спросите чип, кто он, загрузите биты и проверьте их.</p><ol><li><a href="/ru/learn/jtag-asks-the-chip/">25 JTAG спрашивает чип</a></li><li><a href="/ru/learn/same-bits-on-the-board/">26 Те же биты — на плате</a></li><li><a href="/ru/learn/capstone/">27 Итог: переугадайте yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/when-they-disagree/index.html b/apps/website/public/ru/learn/when-they-disagree/index.html
index b1c4e5f41c..2a48eeb147 100644
--- a/apps/website/public/ru/learn/when-they-disagree/index.html
+++ b/apps/website/public/ru/learn/when-they-disagree/index.html
@@ -107,5 +107,6 @@ <h1>Когда они не сходятся</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Зачем проверять</h3><p>Дизайны, которые компилируются и ошибаются; модель, которая выносит вердикт; план, записанный до кода.</p><ol><li><a href="/ru/learn/bugs-that-compile/">01 Баги, которые компилируются</a></li><li><a href="/ru/learn/golden-model/">02 Золотая модель</a></li><li><a href="/ru/learn/plan-before-code/">03 Сначала план</a></li></ol></li><li><h3>Модуль 2 · Тестбенчи</h3><p>Стимулы, проверки и вердикт, записанные как один spec рядом с дизайном, который они судят.</p><ol><li><a href="/ru/learn/anatomy-of-a-tb/">04 Анатомия тестбенча</a></li><li><a href="/ru/learn/self-checking/">05 Самопроверка вместо всматривания</a></li><li><a href="/ru/learn/directed-vs-random/">06 Направленные против случайных</a></li></ol></li><li><h3>Модуль 3 · Временные диаграммы</h3><p>Трасса каждого сигнала, прочитанная так, как её читает инженер по железу, и два прогона, сравнённые между собой.</p><ol><li><a href="/ru/learn/vcd-format/">07 Формат VCD</a></li><li><a href="/ru/learn/debug-with-waves/">08 Отладка по волнам</a></li><li><a href="/ru/learn/compare-traces/">09 Сравнение двух трасс</a></li></ol></li><li><h3>Модуль 4 · Векторы соответствия</h3><p>Случаи с ответом, записанным рядом, там, откуда их достанет компилятор.</p><ol><li><a href="/ru/learn/what-a-vector-is/">10 Что такое вектор</a></li><li><a href="/ru/learn/golden-ruler-vectors/">11 Векторы как эталонная линейка</a></li><li><a href="/ru/learn/end-to-end-scenarios/">12 Сквозные сценарии</a></li></ol></li><li><h3>Модуль 5 · Косимуляция</h3><p>Spec, симулятор и плата сходятся в одном ответе на стенде Artix-7 XC7A200T, и что делать, когда не сходятся.</p><ol><li><a href="/ru/learn/three-way-agreement/">13 Тройное согласие</a></li><li><a href="/ru/learn/simulator-internals/">14 Внутри симулятора</a></li><li class="on" aria-current="page"><a href="/ru/learn/when-they-disagree/">15 Когда они не сходятся</a></li></ol></li><li><h3>Модуль 6 · Покрытие</h3><p>Чего коснулись тесты: строки, переключения, состояния — и что прячет это число.</p><ol><li><a href="/ru/learn/line-and-toggle/">16 Строки и переключения</a></li><li><a href="/ru/learn/fsm-coverage/">17 Покрытие автоматов</a></li><li><a href="/ru/learn/coverage-lies/">18 Покрытие лжёт</a></li></ol></li><li><h3>Модуль 7 · Формальные методы</h3><p>Ассерты, верные каждый такт; ограниченный поиск контрпримера; и почему доказательству нужна индукция.</p><ol><li><a href="/ru/learn/assertions/">19 Ассерты</a></li><li><a href="/ru/learn/bounded-model-checking/">20 Ограниченная проверка моделей</a></li><li><a href="/ru/learn/induction/">21 Индукция</a></li></ol></li><li><h3>Модуль 8 · Мутации</h3><p>Ломайте дизайн нарочно и считайте, что заметили тесты.</p><ol><li><a href="/ru/learn/kill-the-mutant/">22 Убить мутанта</a></li><li><a href="/ru/learn/surviving-mutants/">23 Выжившие мутанты</a></li><li><a href="/ru/learn/test-strength/">24 Сила тестов</a></li></ol></li><li><h3>Модуль 9 · Приёмка</h3><p>Одна команда, все квитанции, чистый вердикт, который можно показать.</p><ol><li><a href="/ru/learn/build-verify/">25 Сборка и проверка</a></li><li><a href="/ru/learn/board-selftest/">26 Стенд проверяет себя</a></li><li><a href="/ru/learn/verify-capstone/">27 Капстоун: проверить модуль целиком</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/verify-hardware.t27">Этот курс сам является spec t27: прочитайте его</a></p><p>Виджеты сохраняют свои английские слова: каждый живёт в собственном spec t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/where-next/index.html b/apps/website/public/ru/learn/where-next/index.html
index d9975ae682..6065acc61e 100644
--- a/apps/website/public/ru/learn/where-next/index.html
+++ b/apps/website/public/ru/learn/where-next/index.html
@@ -84,5 +84,6 @@ <h1>Куда дальше</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Файл спеки</h3><p>Что лежит в файле .t27: строка module, строки прозы и комментарии.</p><ol><li><a href="/ru/learn/what-a-spec-is/">01 Что такое спека</a></li><li><a href="/ru/learn/the-module-line/">02 Строка module</a></li><li><a href="/ru/learn/comments-and-prose/">03 Комментарии и строки прозы</a></li></ol></li><li><h3>Модуль 2 · Значения и типы</h3><p>Константы, ширина целого числа, а также true, false и текст.</p><ol><li><a href="/ru/learn/constants/">04 Константы</a></li><li><a href="/ru/learn/whole-numbers/">05 Целые числа и их ширина</a></li><li><a href="/ru/learn/true-false-and-text/">06 Истина, ложь и текст</a></li></ol></li><li><h3>Модуль 3 · Массивы, триты и выражения</h3><p>Списки значений, три значения трита и что вычисляют операторы.</p><ol><li><a href="/ru/learn/arrays/">07 Массивы</a></li><li><a href="/ru/learn/trits/">08 Триты: минус один, ноль, один</a></li><li><a href="/ru/learn/expressions/">09 Выражения</a></li></ol></li><li><h3>Модуль 4 · Тесты и функции</h3><p>Блок test, несколько маленьких тестов в одной спеке и функция с типизированными входами.</p><ol><li><a href="/ru/learn/a-test-block/">10 Блок test</a></li><li><a href="/ru/learn/many-tests/">11 Много тестов в одной спеке</a></li><li><a href="/ru/learn/functions/">12 Функции</a></li></ol></li><li><h3>Модуль 5 · Внутри функции</h3><p>Имена через let и var, выбор через if и switch и циклы.</p><ol><li><a href="/ru/learn/local-names/">13 Локальные имена: let и var</a></li><li><a href="/ru/learn/if-and-else/">14 Выбор: if, else и switch</a></li><li><a href="/ru/learn/loops/">15 Циклы: while и for</a></li></ol></li><li><h3>Модуль 6 · Видимость и свои типы</h3><p>Что pub открывает другим модулям, и структуры и перечисления, которые вы объявляете.</p><ol><li><a href="/ru/learn/pub-or-private/">16 pub или закрытое имя</a></li><li><a href="/ru/learn/structs/">17 Структуры</a></li><li><a href="/ru/learn/enums/">18 Перечисления</a></li></ol></li><li><h3>Модуль 7 · Модули, правила и gen-ts</h3><p>Чтение другого модуля, правило, которое выполняется всегда, и спека, превращённая в TypeScript.</p><ol><li><a href="/ru/learn/use-other-modules/">19 Другие модули</a></li><li><a href="/ru/learn/invariants/">20 Инварианты и замеры</a></li><li><a href="/ru/learn/gen-ts/">21 gen-ts: спека превращается в TypeScript</a></li></ol></li><li><h3>Модуль 8 · Компилятор и tri</h3><p>Бэкенды, которые пишет t27c, стадии, в которых он читает спеку, и команда tri.</p><ol><li><a href="/ru/learn/seven-backends/">22 Семь бэкендов</a></li><li><a href="/ru/learn/stages-of-t27c/">23 Стадии t27c</a></li><li><a href="/ru/learn/tri-commands/">24 tri: командная строка</a></li></ol></li><li><h3>Модуль 9 · Ошибки, программа и что дальше</h3><p>Как читать ошибку компилятора, одна маленькая законченная программа и какой курс пройти дальше.</p><ol><li><a href="/ru/learn/reading-errors/">25 Как читать ошибки компилятора</a></li><li><a href="/ru/learn/a-small-program/">26 Маленькая законченная программа</a></li><li class="on" aria-current="page"><a href="/ru/learn/where-next/">27 Куда дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/where-the-clock-enters/index.html b/apps/website/public/ru/learn/where-the-clock-enters/index.html
index 400b67c8aa..23253a3d23 100644
--- a/apps/website/public/ru/learn/where-the-clock-enters/index.html
+++ b/apps/website/public/ru/learn/where-the-clock-enters/index.html
@@ -367,5 +367,6 @@ <h1>Где входит тактовый сигнал</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое тактовый сигнал</h3><p>Один фронт, один мир: всё, что делит фронт тактового сигнала, делит и мир; период и джиттер настоящего фронта и место, где тактовый сигнал входит в плату.</p><ol><li><a href="/ru/learn/one-edge-one-world/">01 Один фронт, один мир</a></li><li><a href="/ru/learn/period-and-jitter/">02 Период, фронт и джиттер</a></li><li class="on" aria-current="page"><a href="/ru/learn/where-the-clock-enters/">03 Где входит тактовый сигнал</a></li></ol></li><li><h3>Модуль 2 · Деревья тактового сигнала</h3><p>Перекос и задержка прохождения, глобальная сеть буферов и ловушка стробирования клока логикой.</p><ol><li><a href="/ru/learn/skew-and-insertion/">04 Перекос и задержка прохождения</a></li><li><a href="/ru/learn/global-buffers/">05 Глобальный буфер</a></li><li><a href="/ru/learn/gated-clock-trap/">06 Ловушка стробированного клока</a></li></ol></li><li><h3>Модуль 3 · PLL и MMCM</h3><p>Умножить и поделить один тактовый сигнал в другой, сдвинуть его фазу шагами VCO и какие тактовые сигналы анализатор считает родственными.</p><ol><li><a href="/ru/learn/multiply-and-divide/">07 Умножить и поделить</a></li><li><a href="/ru/learn/phase-and-lock/">08 Фаза и захват</a></li><li><a href="/ru/learn/related-or-not/">09 Родственные или чужие</a></li></ol></li><li><h3>Модуль 4 · Сбросы</h3><p>Ассертировать асинхронно, снимать синхронно: три вида сброса, конвейер снятия и дерево, которое растит сброс.</p><ol><li><a href="/ru/learn/sync-or-async/">10 Синхронный или асинхронный</a></li><li><a href="/ru/learn/reset-release/">11 Конвейер снятия</a></li><li><a href="/ru/learn/reset-fanout/">12 Разветвление сброса</a></li></ol></li><li><h3>Модуль 5 · Метастабильность</h3><p>Окно setup-hold, среднее время между отказами в целочисленной арифметике и два триггера, которые это чинят.</p><ol><li><a href="/ru/learn/setup-hold-window/">13 Окно setup-hold</a></li><li><a href="/ru/learn/mtbf/">14 Среднее время между отказами</a></li><li><a href="/ru/learn/two-flop-sync/">15 Двухтриггерный синхронизатор</a></li></ol></li><li><h3>Модуль 6 · Пересечение многих бит</h3><p>Почему двоичная шина рвётся, почему код Грея нет, и рукопожатие, переносящее импульс между мирами.</p><ol><li><a href="/ru/learn/why-binary-fails/">16 Почему двоичный код рвётся</a></li><li><a href="/ru/learn/gray-code/">17 Код Грея</a></li><li><a href="/ru/learn/handshakes-and-pulses/">18 Рукопожатия и импульсы</a></li></ol></li><li><h3>Модуль 7 · Асинхронный FIFO</h3><p>Указатели, флаги и глубина: буфер, переносящий поток между двумя тактовыми сигналами.</p><ol><li><a href="/ru/learn/pointers-and-flags/">19 Указатели и флаги</a></li><li><a href="/ru/learn/depth-sizing/">20 Выбор глубины</a></li><li><a href="/ru/learn/fifo-testbench/">21 Тестбенч FIFO</a></li></ol></li><li><h3>Модуль 8 · Ограничения</h3><p>Строки, сообщающие анализатору, что такое тактовый сигнал, какие пути не проверять и чему должны соответствовать выводы.</p><ol><li><a href="/ru/learn/create-clock/">22 Строка create_clock</a></li><li><a href="/ru/learn/false-path-and-max-delay/">23 False path и max delay</a></li><li><a href="/ru/learn/io-timing/">24 Тайминг выводов</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Отчёт CDC, одно пересечение, захваченное на триггерах, и дифф битстрима, замыкающий курс.</p><ol><li><a href="/ru/learn/cdc-report/">25 Отчёт CDC</a></li><li><a href="/ru/learn/capture-a-crossing/">26 Захватить пересечение</a></li><li><a href="/ru/learn/capstone-diff/">27 Капстоун: поменяй и сравни</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/where-the-luts-go/index.html b/apps/website/public/ru/learn/where-the-luts-go/index.html
index d837d532d1..f760c66154 100644
--- a/apps/website/public/ru/learn/where-the-luts-go/index.html
+++ b/apps/website/public/ru/learn/where-the-luts-go/index.html
@@ -330,5 +330,6 @@ <h1>Куда уходят LUT</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Чип</h3><p>Что такое FPGA, какой чип у нас и как его выводы встречаются с платой.</p><ol><li><a href="/ru/learn/fpga-is-a-grid/">01 FPGA — это сетка, которую вы настраиваете</a></li><li><a href="/ru/learn/which-chip/">02 Какой это чип?</a></li><li><a href="/ru/learn/pins-and-package/">03 Выводы, шарики и корпус</a></li></ol></li><li><h3>Модуль 2 · Числа в железе</h3><p>Биты, триты и форматы чисел, и сколько стоит арифметика в логике.</p><ol><li><a href="/ru/learn/bits-and-trits/">04 Биты, триты и GF16</a></li><li><a href="/ru/learn/what-arithmetic-costs/">05 Сколько стоит арифметика</a></li><li><a href="/ru/learn/how-much-fits/">06 Сколько помещается на чип</a></li></ol></li><li><h3>Модуль 3 · Ваша программа на t27</h3><p>Напишите спеку, проверьте её и увидьте, почему «компилируется» не значит «верно».</p><ol><li><a href="/ru/learn/hello-t27/">07 Привет, t27</a></li><li><a href="/ru/learn/tests-are-the-spec/">08 Тесты — часть спеки</a></li><li><a href="/ru/learn/compiles-is-not-correct/">09 Компилируется — не значит верно</a></li></ol></li><li><h3>Модуль 4 · Внутри t27c</h3><p>Как компилятор читает спеку и что он пишет, включая нативный t27b.</p><ol><li><a href="/ru/learn/compiler-stages/">10 Стадии t27c</a></li><li><a href="/ru/learn/t27c-on-a-real-machine/">11 t27c на настоящей машине</a></li><li><a href="/ru/learn/seven-backends-and-t27b/">12 Семь бэкендов и t27b</a></li></ol></li><li><h3>Модуль 5 · От спеки к железу</h3><p>Verilog, который пишет t27c, ячейки, в которые он превращается, и что делает одна LUT.</p><ol><li><a href="/ru/learn/spec-to-verilog/">13 От спеки к Verilog</a></li><li><a href="/ru/learn/count-the-cells/">14 Посчитайте ячейки</a></li><li><a href="/ru/learn/one-lut-up-close/">15 Одна LUT вблизи</a></li></ol></li><li><h3>Модуль 6 · Читаем синтез</h3><p>Что yosys сообщает о вашем дизайне и какие предупреждения важны.</p><ol><li><a href="/ru/learn/synthesis-warnings/">16 Важные предупреждения синтеза</a></li><li><a href="/ru/learn/state-machines/">17 Конечные автоматы</a></li><li class="on" aria-current="page"><a href="/ru/learn/where-the-luts-go/">18 Куда уходят LUT</a></li></ol></li><li><h3>Модуль 7 · Размещение, трассировка, тайминг</h3><p>Где ячейки оказываются на кристалле и успевает ли тактовая частота.</p><ol><li><a href="/ru/learn/placement/">19 Размещение: где оказываются ячейки</a></li><li><a href="/ru/learn/logic-depth/">20 Глубина логики</a></li><li><a href="/ru/learn/timing-slack/">21 Тайминг и запас</a></li></ol></li><li><h3>Модуль 8 · Битстрим</h3><p>Как оттрассированный дизайн становится битами, которые загружает чип, без Vivado.</p><ol><li><a href="/ru/learn/fasm-to-frames/">22 От FASM к фреймам</a></li><li><a href="/ru/learn/a-build-receipt/">23 Сборка по пунктам</a></li><li><a href="/ru/learn/read-a-bitstream/">24 Читаем битстрим</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Спросите чип, кто он, загрузите биты и проверьте их.</p><ol><li><a href="/ru/learn/jtag-asks-the-chip/">25 JTAG спрашивает чип</a></li><li><a href="/ru/learn/same-bits-on-the-board/">26 Те же биты — на плате</a></li><li><a href="/ru/learn/capstone/">27 Итог: переугадайте yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/which-chip/index.html b/apps/website/public/ru/learn/which-chip/index.html
index 34fc0a9f63..051b8e0b8b 100644
--- a/apps/website/public/ru/learn/which-chip/index.html
+++ b/apps/website/public/ru/learn/which-chip/index.html
@@ -113,5 +113,6 @@ <h1>Какой это чип?</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Чип</h3><p>Что такое FPGA, какой чип у нас и как его выводы встречаются с платой.</p><ol><li><a href="/ru/learn/fpga-is-a-grid/">01 FPGA — это сетка, которую вы настраиваете</a></li><li class="on" aria-current="page"><a href="/ru/learn/which-chip/">02 Какой это чип?</a></li><li><a href="/ru/learn/pins-and-package/">03 Выводы, шарики и корпус</a></li></ol></li><li><h3>Модуль 2 · Числа в железе</h3><p>Биты, триты и форматы чисел, и сколько стоит арифметика в логике.</p><ol><li><a href="/ru/learn/bits-and-trits/">04 Биты, триты и GF16</a></li><li><a href="/ru/learn/what-arithmetic-costs/">05 Сколько стоит арифметика</a></li><li><a href="/ru/learn/how-much-fits/">06 Сколько помещается на чип</a></li></ol></li><li><h3>Модуль 3 · Ваша программа на t27</h3><p>Напишите спеку, проверьте её и увидьте, почему «компилируется» не значит «верно».</p><ol><li><a href="/ru/learn/hello-t27/">07 Привет, t27</a></li><li><a href="/ru/learn/tests-are-the-spec/">08 Тесты — часть спеки</a></li><li><a href="/ru/learn/compiles-is-not-correct/">09 Компилируется — не значит верно</a></li></ol></li><li><h3>Модуль 4 · Внутри t27c</h3><p>Как компилятор читает спеку и что он пишет, включая нативный t27b.</p><ol><li><a href="/ru/learn/compiler-stages/">10 Стадии t27c</a></li><li><a href="/ru/learn/t27c-on-a-real-machine/">11 t27c на настоящей машине</a></li><li><a href="/ru/learn/seven-backends-and-t27b/">12 Семь бэкендов и t27b</a></li></ol></li><li><h3>Модуль 5 · От спеки к железу</h3><p>Verilog, который пишет t27c, ячейки, в которые он превращается, и что делает одна LUT.</p><ol><li><a href="/ru/learn/spec-to-verilog/">13 От спеки к Verilog</a></li><li><a href="/ru/learn/count-the-cells/">14 Посчитайте ячейки</a></li><li><a href="/ru/learn/one-lut-up-close/">15 Одна LUT вблизи</a></li></ol></li><li><h3>Модуль 6 · Читаем синтез</h3><p>Что yosys сообщает о вашем дизайне и какие предупреждения важны.</p><ol><li><a href="/ru/learn/synthesis-warnings/">16 Важные предупреждения синтеза</a></li><li><a href="/ru/learn/state-machines/">17 Конечные автоматы</a></li><li><a href="/ru/learn/where-the-luts-go/">18 Куда уходят LUT</a></li></ol></li><li><h3>Модуль 7 · Размещение, трассировка, тайминг</h3><p>Где ячейки оказываются на кристалле и успевает ли тактовая частота.</p><ol><li><a href="/ru/learn/placement/">19 Размещение: где оказываются ячейки</a></li><li><a href="/ru/learn/logic-depth/">20 Глубина логики</a></li><li><a href="/ru/learn/timing-slack/">21 Тайминг и запас</a></li></ol></li><li><h3>Модуль 8 · Битстрим</h3><p>Как оттрассированный дизайн становится битами, которые загружает чип, без Vivado.</p><ol><li><a href="/ru/learn/fasm-to-frames/">22 От FASM к фреймам</a></li><li><a href="/ru/learn/a-build-receipt/">23 Сборка по пунктам</a></li><li><a href="/ru/learn/read-a-bitstream/">24 Читаем битстрим</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Спросите чип, кто он, загрузите биты и проверьте их.</p><ol><li><a href="/ru/learn/jtag-asks-the-chip/">25 JTAG спрашивает чип</a></li><li><a href="/ru/learn/same-bits-on-the-board/">26 Те же биты — на плате</a></li><li><a href="/ru/learn/capstone/">27 Итог: переугадайте yosys</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/course.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/who-holds-the-io/index.html b/apps/website/public/ru/learn/who-holds-the-io/index.html
index 26eb1b0a77..c0b7f75eae 100644
--- a/apps/website/public/ru/learn/who-holds-the-io/index.html
+++ b/apps/website/public/ru/learn/who-holds-the-io/index.html
@@ -109,5 +109,6 @@ <h1>Кто держит IO</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое шина</h3><p>Зачем вообще существует шина: разговор по проводам, с кадрами и адресами, и кому позволено говорить.</p><ol><li><a href="/ru/learn/wires-that-talk/">01 Провода, которые говорят</a></li><li><a href="/ru/learn/framing-a-conversation/">02 Кадрирование разговора</a></li><li><a href="/ru/learn/master-and-servant/">03 Хозяин и слуга</a></li></ol></li><li><h3>Модуль 2 · Разговор по UART</h3><p>Двухпроводная шина без такта: кадр, делитель, задающий скорость, и статус, который опрашивает драйвер.</p><ol><li><a href="/ru/learn/the-frame/">04 Кадр: старт, данные, стоп</a></li><li><a href="/ru/learn/baud-and-the-divisor/">05 Бод и делитель</a></li><li><a href="/ru/learn/status-and-fifos/">06 Статус и FIFO</a></li></ol></li><li><h3>Модуль 3 · SPI под тактом</h3><p>Разговор под тактом: четыре режима, лестница предделителей и выбор кристалла на каждого подчинённого.</p><ol><li><a href="/ru/learn/modes-cpol-cpha/">07 Режимы: CPOL и CPHA</a></li><li><a href="/ru/learn/the-prescaler-ladder/">08 Лестница предделителей</a></li><li><a href="/ru/learn/chip-select-and-widths/">09 Выбор кристалла и ширины</a></li></ol></li><li><h3>Модуль 4 · Регистровая шина APB</h3><p>Шина регистров: PSEL и PENABLE, строобы и ожидания, и сколько адресных бит стоит количество периферии.</p><ol><li><a href="/ru/learn/psel-and-penable/">10 PSEL и PENABLE</a></li><li><a href="/ru/learn/strobes-and-waits/">11 Строобы и ожидания</a></li><li><a href="/ru/learn/address-decode/">12 Дешифрация адреса</a></li></ol></li><li><h3>Модуль 5 · Пять каналов AXI4</h3><p>Пять каналов: адрес, данные и отклик в обе стороны, lite или full, пакеты и идентификаторы.</p><ol><li><a href="/ru/learn/five-channels/">13 Пять каналов</a></li><li><a href="/ru/learn/lite-or-full/">14 Lite или full</a></li><li><a href="/ru/learn/bursts-and-ids/">15 Пакеты и идентификаторы</a></li></ol></li><li><h3>Модуль 6 · Память</h3><p>Что находится на дальнем конце каждой шины: карты памяти, типы портов и задержка, которую должно покрывать ожидание.</p><ol><li><a href="/ru/learn/memory-maps/">16 Карты памяти</a></li><li><a href="/ru/learn/ports-and-kinds/">17 Порты и типы</a></li><li><a href="/ru/learn/latency/">18 Задержка</a></li></ol></li><li><h3>Модуль 7 · Мосты</h3><p>Почему в дизайне появляется больше одной шины и пакетный мост, переносящий работу между ними.</p><ol><li><a href="/ru/learn/why-bridges/">19 Зачем мосты</a></li><li><a href="/ru/learn/the-packet-bridge/">20 Пакетный мост</a></li><li><a href="/ru/learn/everything-in-one-bench/">21 Всё в одном стенде</a></li></ol></li><li><h3>Модуль 8 · Ethernet: кадры и тайминг</h3><p>Кадры, контрольная сумма кадра, тайминг RGMII и заранее зарегистрированные шаги настоящего запуска.</p><ol><li><a href="/ru/learn/frames-and-the-fcs/">22 Кадры и контрольная сумма</a></li><li><a href="/ru/learn/rgmii-timing/">23 Тайминг RGMII</a></li><li><a href="/ru/learn/bring-up-on-the-bench/">24 Запуск на стенде</a></li></ol></li><li><h3>Модуль 9 · Стенд</h3><p>Дисциплина, оберегающая настоящее железо: кто держит IO, как взять и вернуть захват, и что запускается дальше.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/who-holds-the-io/">25 Кто держит IO</a></li><li><a href="/ru/learn/claim-and-release/">26 Захват и возврат</a></li><li><a href="/ru/learn/capstone-what-runs-next/">27 Капстоун: что запускается дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/whole-numbers/index.html b/apps/website/public/ru/learn/whole-numbers/index.html
index 69db09a7e7..71e433ad76 100644
--- a/apps/website/public/ru/learn/whole-numbers/index.html
+++ b/apps/website/public/ru/learn/whole-numbers/index.html
@@ -85,5 +85,6 @@ <h1>Целые числа и их ширина</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Файл спеки</h3><p>Что лежит в файле .t27: строка module, строки прозы и комментарии.</p><ol><li><a href="/ru/learn/what-a-spec-is/">01 Что такое спека</a></li><li><a href="/ru/learn/the-module-line/">02 Строка module</a></li><li><a href="/ru/learn/comments-and-prose/">03 Комментарии и строки прозы</a></li></ol></li><li><h3>Модуль 2 · Значения и типы</h3><p>Константы, ширина целого числа, а также true, false и текст.</p><ol><li><a href="/ru/learn/constants/">04 Константы</a></li><li class="on" aria-current="page"><a href="/ru/learn/whole-numbers/">05 Целые числа и их ширина</a></li><li><a href="/ru/learn/true-false-and-text/">06 Истина, ложь и текст</a></li></ol></li><li><h3>Модуль 3 · Массивы, триты и выражения</h3><p>Списки значений, три значения трита и что вычисляют операторы.</p><ol><li><a href="/ru/learn/arrays/">07 Массивы</a></li><li><a href="/ru/learn/trits/">08 Триты: минус один, ноль, один</a></li><li><a href="/ru/learn/expressions/">09 Выражения</a></li></ol></li><li><h3>Модуль 4 · Тесты и функции</h3><p>Блок test, несколько маленьких тестов в одной спеке и функция с типизированными входами.</p><ol><li><a href="/ru/learn/a-test-block/">10 Блок test</a></li><li><a href="/ru/learn/many-tests/">11 Много тестов в одной спеке</a></li><li><a href="/ru/learn/functions/">12 Функции</a></li></ol></li><li><h3>Модуль 5 · Внутри функции</h3><p>Имена через let и var, выбор через if и switch и циклы.</p><ol><li><a href="/ru/learn/local-names/">13 Локальные имена: let и var</a></li><li><a href="/ru/learn/if-and-else/">14 Выбор: if, else и switch</a></li><li><a href="/ru/learn/loops/">15 Циклы: while и for</a></li></ol></li><li><h3>Модуль 6 · Видимость и свои типы</h3><p>Что pub открывает другим модулям, и структуры и перечисления, которые вы объявляете.</p><ol><li><a href="/ru/learn/pub-or-private/">16 pub или закрытое имя</a></li><li><a href="/ru/learn/structs/">17 Структуры</a></li><li><a href="/ru/learn/enums/">18 Перечисления</a></li></ol></li><li><h3>Модуль 7 · Модули, правила и gen-ts</h3><p>Чтение другого модуля, правило, которое выполняется всегда, и спека, превращённая в TypeScript.</p><ol><li><a href="/ru/learn/use-other-modules/">19 Другие модули</a></li><li><a href="/ru/learn/invariants/">20 Инварианты и замеры</a></li><li><a href="/ru/learn/gen-ts/">21 gen-ts: спека превращается в TypeScript</a></li></ol></li><li><h3>Модуль 8 · Компилятор и tri</h3><p>Бэкенды, которые пишет t27c, стадии, в которых он читает спеку, и команда tri.</p><ol><li><a href="/ru/learn/seven-backends/">22 Семь бэкендов</a></li><li><a href="/ru/learn/stages-of-t27c/">23 Стадии t27c</a></li><li><a href="/ru/learn/tri-commands/">24 tri: командная строка</a></li></ol></li><li><h3>Модуль 9 · Ошибки, программа и что дальше</h3><p>Как читать ошибку компилятора, одна маленькая законченная программа и какой курс пройти дальше.</p><ol><li><a href="/ru/learn/reading-errors/">25 Как читать ошибки компилятора</a></li><li><a href="/ru/learn/a-small-program/">26 Маленькая законченная программа</a></li><li><a href="/ru/learn/where-next/">27 Куда дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/t27-basics.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/why-base-three/index.html b/apps/website/public/ru/learn/why-base-three/index.html
index b6239a8d92..32f88b20fd 100644
--- a/apps/website/public/ru/learn/why-base-three/index.html
+++ b/apps/website/public/ru/learn/why-base-three/index.html
@@ -414,5 +414,6 @@ <h1>Почему основание три</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Правило и его числа</h3><p>Одно правило делит каждую ширину, отношение, к которому оно стремится, и числа Люка за тройкой 3.</p><ol><li><a href="/ru/learn/a-float-cut-by-phi/">01 Число с плавающей точкой, разрезанное по phi</a></li><li><a href="/ru/learn/phi-as-a-ratio/">02 Phi как отношение</a></li><li><a href="/ru/learn/lucas-numbers-stay-whole/">03 Числа Люка остаются целыми</a></li></ol></li><li><h3>Модуль 2 · Почему phi, почему три</h3><p>Почему деление идёт по phi, почему основание три и как спека проверяет, что GF16 хранит phi.</p><ol><li><a href="/ru/learn/why-the-split-is-phi/">04 Почему деление по phi</a></li><li class="on" aria-current="page"><a href="/ru/learn/why-base-three/">05 Почему основание три</a></li><li><a href="/ru/learn/phi-in-sixteen-bits/">06 Phi в шестнадцати битах</a></li></ol></li><li><h3>Модуль 3 · Малые ступени: от GF4 до GF8</h3><p>GF4, GF6 и GF8 — меньше всего битов, и округление до целых битов стоит здесь дороже всего.</p><ol><li><a href="/ru/learn/gf4-four-bits/">07 GF4: четыре бита</a></li><li><a href="/ru/learn/gf6-six-bits/">08 GF6: шесть битов</a></li><li><a href="/ru/learn/gf8-one-byte/">09 GF8: один байт</a></li></ol></li><li><h3>Модуль 4 · От десяти до четырнадцати битов</h3><p>GF10, GF12 и GF14 и то, как расстояние до 1 / phi меняется с ростом слова.</p><ol><li><a href="/ru/learn/gf10-ten-bits/">10 GF10: десять битов</a></li><li><a href="/ru/learn/gf12-twelve-bits/">11 GF12: двенадцать битов</a></li><li><a href="/ru/learn/gf14-fourteen-bits/">12 GF14: четырнадцать битов</a></li></ol></li><li><h3>Модуль 5 · GF16 в работе</h3><p>Основной 16-битный формат, скалярное произведение из двух слагаемых в GF-T16, затем GF20 и GF24.</p><ol><li><a href="/ru/learn/gf16-the-primary/">13 GF16: основной формат</a></li><li><a href="/ru/learn/two-products-one-sum/">14 Два произведения, одна сумма</a></li><li><a href="/ru/learn/gf24-twenty-four-bits/">15 GF20 и GF24</a></li></ol></li><li><h3>Модуль 6 · От GF32 до GF64</h3><p>GF32 рядом с IEEE single, GF48 без пары в IEEE, GF64 рядом с IEEE double.</p><ol><li><a href="/ru/learn/gf32-a-single/">16 GF32: одинарная точность</a></li><li><a href="/ru/learn/gf48-forty-eight-bits/">17 GF48: сорок восемь битов</a></li><li><a href="/ru/learn/gf64-a-double/">18 GF64: двойная точность</a></li></ol></li><li><h3>Модуль 7 · От GF96 до GF256</h3><p>GF96, GF128 и GF256, где спеки держат раскладку инвариантами.</p><ol><li><a href="/ru/learn/gf96-ninety-six-bits/">19 GF96: девяносто шесть битов</a></li><li><a href="/ru/learn/gf128-a-quad/">20 GF128: четверная точность</a></li><li><a href="/ru/learn/gf256-two-five-six/">21 GF256: двести пятьдесят шесть битов</a></li></ol></li><li><h3>Модуль 8 · Самые широкие ступени, затем триты</h3><p>GF512 и GF1024, две самые широкие ступени, затем GF-T8, где порядок уходит в триты.</p><ol><li><a href="/ru/learn/gf512-five-twelve/">22 GF512: пятьсот двенадцать битов</a></li><li><a href="/ru/learn/gf1024-one-kilobit/">23 GF1024: один килобит</a></li><li><a href="/ru/learn/gft8-exponent-in-trits/">24 GF-T8: порядок в тритах</a></li></ol></li><li><h3>Модуль 9 · Ещё триты, затем декодирование</h3><p>GF-T16 и GF-T32, затем почему фиксированные поля декодируются параллельно, а posit — нет.</p><ol><li><a href="/ru/learn/gft16-six-trits/">25 GF-T16: шесть тритов</a></li><li><a href="/ru/learn/gft32-twelve-trits/">26 GF-T32: двенадцать тритов</a></li><li><a href="/ru/learn/decode-in-one-step/">27 Раскодировать за один шаг</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/why-binary-fails/index.html b/apps/website/public/ru/learn/why-binary-fails/index.html
index fc51c2fae8..3943287d0a 100644
--- a/apps/website/public/ru/learn/why-binary-fails/index.html
+++ b/apps/website/public/ru/learn/why-binary-fails/index.html
@@ -434,5 +434,6 @@ <h1>Почему двоичный код рвётся</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое тактовый сигнал</h3><p>Один фронт, один мир: всё, что делит фронт тактового сигнала, делит и мир; период и джиттер настоящего фронта и место, где тактовый сигнал входит в плату.</p><ol><li><a href="/ru/learn/one-edge-one-world/">01 Один фронт, один мир</a></li><li><a href="/ru/learn/period-and-jitter/">02 Период, фронт и джиттер</a></li><li><a href="/ru/learn/where-the-clock-enters/">03 Где входит тактовый сигнал</a></li></ol></li><li><h3>Модуль 2 · Деревья тактового сигнала</h3><p>Перекос и задержка прохождения, глобальная сеть буферов и ловушка стробирования клока логикой.</p><ol><li><a href="/ru/learn/skew-and-insertion/">04 Перекос и задержка прохождения</a></li><li><a href="/ru/learn/global-buffers/">05 Глобальный буфер</a></li><li><a href="/ru/learn/gated-clock-trap/">06 Ловушка стробированного клока</a></li></ol></li><li><h3>Модуль 3 · PLL и MMCM</h3><p>Умножить и поделить один тактовый сигнал в другой, сдвинуть его фазу шагами VCO и какие тактовые сигналы анализатор считает родственными.</p><ol><li><a href="/ru/learn/multiply-and-divide/">07 Умножить и поделить</a></li><li><a href="/ru/learn/phase-and-lock/">08 Фаза и захват</a></li><li><a href="/ru/learn/related-or-not/">09 Родственные или чужие</a></li></ol></li><li><h3>Модуль 4 · Сбросы</h3><p>Ассертировать асинхронно, снимать синхронно: три вида сброса, конвейер снятия и дерево, которое растит сброс.</p><ol><li><a href="/ru/learn/sync-or-async/">10 Синхронный или асинхронный</a></li><li><a href="/ru/learn/reset-release/">11 Конвейер снятия</a></li><li><a href="/ru/learn/reset-fanout/">12 Разветвление сброса</a></li></ol></li><li><h3>Модуль 5 · Метастабильность</h3><p>Окно setup-hold, среднее время между отказами в целочисленной арифметике и два триггера, которые это чинят.</p><ol><li><a href="/ru/learn/setup-hold-window/">13 Окно setup-hold</a></li><li><a href="/ru/learn/mtbf/">14 Среднее время между отказами</a></li><li><a href="/ru/learn/two-flop-sync/">15 Двухтриггерный синхронизатор</a></li></ol></li><li><h3>Модуль 6 · Пересечение многих бит</h3><p>Почему двоичная шина рвётся, почему код Грея нет, и рукопожатие, переносящее импульс между мирами.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/why-binary-fails/">16 Почему двоичный код рвётся</a></li><li><a href="/ru/learn/gray-code/">17 Код Грея</a></li><li><a href="/ru/learn/handshakes-and-pulses/">18 Рукопожатия и импульсы</a></li></ol></li><li><h3>Модуль 7 · Асинхронный FIFO</h3><p>Указатели, флаги и глубина: буфер, переносящий поток между двумя тактовыми сигналами.</p><ol><li><a href="/ru/learn/pointers-and-flags/">19 Указатели и флаги</a></li><li><a href="/ru/learn/depth-sizing/">20 Выбор глубины</a></li><li><a href="/ru/learn/fifo-testbench/">21 Тестбенч FIFO</a></li></ol></li><li><h3>Модуль 8 · Ограничения</h3><p>Строки, сообщающие анализатору, что такое тактовый сигнал, какие пути не проверять и чему должны соответствовать выводы.</p><ol><li><a href="/ru/learn/create-clock/">22 Строка create_clock</a></li><li><a href="/ru/learn/false-path-and-max-delay/">23 False path и max delay</a></li><li><a href="/ru/learn/io-timing/">24 Тайминг выводов</a></li></ol></li><li><h3>Модуль 9 · На плате</h3><p>Отчёт CDC, одно пересечение, захваченное на триггерах, и дифф битстрима, замыкающий курс.</p><ol><li><a href="/ru/learn/cdc-report/">25 Отчёт CDC</a></li><li><a href="/ru/learn/capture-a-crossing/">26 Захватить пересечение</a></li><li><a href="/ru/learn/capstone-diff/">27 Капстоун: поменяй и сравни</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/clocks-and-cdc.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/why-bridges/index.html b/apps/website/public/ru/learn/why-bridges/index.html
index 95e0fe4db4..f193a0bfee 100644
--- a/apps/website/public/ru/learn/why-bridges/index.html
+++ b/apps/website/public/ru/learn/why-bridges/index.html
@@ -593,5 +593,6 @@ <h1>Зачем мосты</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое шина</h3><p>Зачем вообще существует шина: разговор по проводам, с кадрами и адресами, и кому позволено говорить.</p><ol><li><a href="/ru/learn/wires-that-talk/">01 Провода, которые говорят</a></li><li><a href="/ru/learn/framing-a-conversation/">02 Кадрирование разговора</a></li><li><a href="/ru/learn/master-and-servant/">03 Хозяин и слуга</a></li></ol></li><li><h3>Модуль 2 · Разговор по UART</h3><p>Двухпроводная шина без такта: кадр, делитель, задающий скорость, и статус, который опрашивает драйвер.</p><ol><li><a href="/ru/learn/the-frame/">04 Кадр: старт, данные, стоп</a></li><li><a href="/ru/learn/baud-and-the-divisor/">05 Бод и делитель</a></li><li><a href="/ru/learn/status-and-fifos/">06 Статус и FIFO</a></li></ol></li><li><h3>Модуль 3 · SPI под тактом</h3><p>Разговор под тактом: четыре режима, лестница предделителей и выбор кристалла на каждого подчинённого.</p><ol><li><a href="/ru/learn/modes-cpol-cpha/">07 Режимы: CPOL и CPHA</a></li><li><a href="/ru/learn/the-prescaler-ladder/">08 Лестница предделителей</a></li><li><a href="/ru/learn/chip-select-and-widths/">09 Выбор кристалла и ширины</a></li></ol></li><li><h3>Модуль 4 · Регистровая шина APB</h3><p>Шина регистров: PSEL и PENABLE, строобы и ожидания, и сколько адресных бит стоит количество периферии.</p><ol><li><a href="/ru/learn/psel-and-penable/">10 PSEL и PENABLE</a></li><li><a href="/ru/learn/strobes-and-waits/">11 Строобы и ожидания</a></li><li><a href="/ru/learn/address-decode/">12 Дешифрация адреса</a></li></ol></li><li><h3>Модуль 5 · Пять каналов AXI4</h3><p>Пять каналов: адрес, данные и отклик в обе стороны, lite или full, пакеты и идентификаторы.</p><ol><li><a href="/ru/learn/five-channels/">13 Пять каналов</a></li><li><a href="/ru/learn/lite-or-full/">14 Lite или full</a></li><li><a href="/ru/learn/bursts-and-ids/">15 Пакеты и идентификаторы</a></li></ol></li><li><h3>Модуль 6 · Память</h3><p>Что находится на дальнем конце каждой шины: карты памяти, типы портов и задержка, которую должно покрывать ожидание.</p><ol><li><a href="/ru/learn/memory-maps/">16 Карты памяти</a></li><li><a href="/ru/learn/ports-and-kinds/">17 Порты и типы</a></li><li><a href="/ru/learn/latency/">18 Задержка</a></li></ol></li><li><h3>Модуль 7 · Мосты</h3><p>Почему в дизайне появляется больше одной шины и пакетный мост, переносящий работу между ними.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/why-bridges/">19 Зачем мосты</a></li><li><a href="/ru/learn/the-packet-bridge/">20 Пакетный мост</a></li><li><a href="/ru/learn/everything-in-one-bench/">21 Всё в одном стенде</a></li></ol></li><li><h3>Модуль 8 · Ethernet: кадры и тайминг</h3><p>Кадры, контрольная сумма кадра, тайминг RGMII и заранее зарегистрированные шаги настоящего запуска.</p><ol><li><a href="/ru/learn/frames-and-the-fcs/">22 Кадры и контрольная сумма</a></li><li><a href="/ru/learn/rgmii-timing/">23 Тайминг RGMII</a></li><li><a href="/ru/learn/bring-up-on-the-bench/">24 Запуск на стенде</a></li></ol></li><li><h3>Модуль 9 · Стенд</h3><p>Дисциплина, оберегающая настоящее железо: кто держит IO, как взять и вернуть захват, и что запускается дальше.</p><ol><li><a href="/ru/learn/who-holds-the-io/">25 Кто держит IO</a></li><li><a href="/ru/learn/claim-and-release/">26 Захват и возврат</a></li><li><a href="/ru/learn/capstone-what-runs-next/">27 Капстоун: что запускается дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/why-the-split-is-phi/index.html b/apps/website/public/ru/learn/why-the-split-is-phi/index.html
index 82d8f5e38c..9199aabf00 100644
--- a/apps/website/public/ru/learn/why-the-split-is-phi/index.html
+++ b/apps/website/public/ru/learn/why-the-split-is-phi/index.html
@@ -428,5 +428,6 @@ <h1>Почему деление по phi</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Правило и его числа</h3><p>Одно правило делит каждую ширину, отношение, к которому оно стремится, и числа Люка за тройкой 3.</p><ol><li><a href="/ru/learn/a-float-cut-by-phi/">01 Число с плавающей точкой, разрезанное по phi</a></li><li><a href="/ru/learn/phi-as-a-ratio/">02 Phi как отношение</a></li><li><a href="/ru/learn/lucas-numbers-stay-whole/">03 Числа Люка остаются целыми</a></li></ol></li><li><h3>Модуль 2 · Почему phi, почему три</h3><p>Почему деление идёт по phi, почему основание три и как спека проверяет, что GF16 хранит phi.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/why-the-split-is-phi/">04 Почему деление по phi</a></li><li><a href="/ru/learn/why-base-three/">05 Почему основание три</a></li><li><a href="/ru/learn/phi-in-sixteen-bits/">06 Phi в шестнадцати битах</a></li></ol></li><li><h3>Модуль 3 · Малые ступени: от GF4 до GF8</h3><p>GF4, GF6 и GF8 — меньше всего битов, и округление до целых битов стоит здесь дороже всего.</p><ol><li><a href="/ru/learn/gf4-four-bits/">07 GF4: четыре бита</a></li><li><a href="/ru/learn/gf6-six-bits/">08 GF6: шесть битов</a></li><li><a href="/ru/learn/gf8-one-byte/">09 GF8: один байт</a></li></ol></li><li><h3>Модуль 4 · От десяти до четырнадцати битов</h3><p>GF10, GF12 и GF14 и то, как расстояние до 1 / phi меняется с ростом слова.</p><ol><li><a href="/ru/learn/gf10-ten-bits/">10 GF10: десять битов</a></li><li><a href="/ru/learn/gf12-twelve-bits/">11 GF12: двенадцать битов</a></li><li><a href="/ru/learn/gf14-fourteen-bits/">12 GF14: четырнадцать битов</a></li></ol></li><li><h3>Модуль 5 · GF16 в работе</h3><p>Основной 16-битный формат, скалярное произведение из двух слагаемых в GF-T16, затем GF20 и GF24.</p><ol><li><a href="/ru/learn/gf16-the-primary/">13 GF16: основной формат</a></li><li><a href="/ru/learn/two-products-one-sum/">14 Два произведения, одна сумма</a></li><li><a href="/ru/learn/gf24-twenty-four-bits/">15 GF20 и GF24</a></li></ol></li><li><h3>Модуль 6 · От GF32 до GF64</h3><p>GF32 рядом с IEEE single, GF48 без пары в IEEE, GF64 рядом с IEEE double.</p><ol><li><a href="/ru/learn/gf32-a-single/">16 GF32: одинарная точность</a></li><li><a href="/ru/learn/gf48-forty-eight-bits/">17 GF48: сорок восемь битов</a></li><li><a href="/ru/learn/gf64-a-double/">18 GF64: двойная точность</a></li></ol></li><li><h3>Модуль 7 · От GF96 до GF256</h3><p>GF96, GF128 и GF256, где спеки держат раскладку инвариантами.</p><ol><li><a href="/ru/learn/gf96-ninety-six-bits/">19 GF96: девяносто шесть битов</a></li><li><a href="/ru/learn/gf128-a-quad/">20 GF128: четверная точность</a></li><li><a href="/ru/learn/gf256-two-five-six/">21 GF256: двести пятьдесят шесть битов</a></li></ol></li><li><h3>Модуль 8 · Самые широкие ступени, затем триты</h3><p>GF512 и GF1024, две самые широкие ступени, затем GF-T8, где порядок уходит в триты.</p><ol><li><a href="/ru/learn/gf512-five-twelve/">22 GF512: пятьсот двенадцать битов</a></li><li><a href="/ru/learn/gf1024-one-kilobit/">23 GF1024: один килобит</a></li><li><a href="/ru/learn/gft8-exponent-in-trits/">24 GF-T8: порядок в тритах</a></li></ol></li><li><h3>Модуль 9 · Ещё триты, затем декодирование</h3><p>GF-T16 и GF-T32, затем почему фиксированные поля декодируются параллельно, а posit — нет.</p><ol><li><a href="/ru/learn/gft16-six-trits/">25 GF-T16: шесть тритов</a></li><li><a href="/ru/learn/gft32-twelve-trits/">26 GF-T32: двенадцать тритов</a></li><li><a href="/ru/learn/decode-in-one-step/">27 Раскодировать за один шаг</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/goldenfloat.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/why-three/index.html b/apps/website/public/ru/learn/why-three/index.html
index 1fd1a5c42f..afa8a55dda 100644
--- a/apps/website/public/ru/learn/why-three/index.html
+++ b/apps/website/public/ru/learn/why-three/index.html
@@ -184,5 +184,6 @@ <h1>Почему три</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Три значения</h3><p>Почему три, как сбалансированная троичная система пишет любое число без знака и что дают переворот и отсечение тритов.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/why-three/">01 Почему три</a></li><li><a href="/ru/learn/balanced-ternary/">02 Сбалансированная троичная</a></li><li><a href="/ru/learn/flip-and-cut/">03 Переворот вместо минуса, отсечение вместо округления</a></li></ol></li><li><h3>Модуль 2 · Логика с «не знаю»</h3><p>Трёхзначные вентили Клини, два вентиля без двоичного двойника и сложение как пара таблиц.</p><ol><li><a href="/ru/learn/kleene-logic/">04 Трёхзначная логика</a></li><li><a href="/ru/learn/consensus-and-any/">05 Консенсус и «любой»</a></li><li><a href="/ru/learn/sum-and-carry/">06 Сумма и перенос</a></li></ol></li><li><h3>Модуль 3 · Складываем триты</h3><p>Полусумматор, полный сумматор и перенос, бегущий влево разряд за разрядом.</p><ol><li><a href="/ru/learn/half-adder/">07 Полусумматор</a></li><li><a href="/ru/learn/full-adder/">08 Полный сумматор</a></li><li><a href="/ru/learn/ripple-carry/">09 Сквозной перенос</a></li></ol></li><li><h3>Модуль 4 · Умножение без умножителя</h3><p>Копия, ноль или переворот: произведение на один трит, умножение столбиком и MAC, который только складывает.</p><ol><li><a href="/ru/learn/times-a-trit/">10 Умножить на один трит</a></li><li><a href="/ru/learn/long-multiplication/">11 Умножение столбиком</a></li><li><a href="/ru/learn/ternary-mac/">12 Троичный MAC</a></li></ol></li><li><h3>Модуль 5 · Триты в двоичной памяти</h3><p>Два бита на трит, пять тритов на байт и сколько бит нужно слову из 27 тритов.</p><ol><li><a href="/ru/learn/two-bits-per-trit/">13 Два бита на трит</a></li><li><a href="/ru/learn/five-trits-a-byte/">14 Пять тритов в байте</a></li><li><a href="/ru/learn/a-word-of-27/">15 Слово из 27 тритов</a></li></ol></li><li><h3>Модуль 6 · Командное слово TRI-27</h3><p>32-битное слово, которое декодирует эмулятор Trinity: его поля, 47 опкодов и 15-битное непосредственное значение.</p><ol><li><a href="/ru/learn/the-instruction-word/">16 Командное слово</a></li><li><a href="/ru/learn/the-opcode-map/">17 Карта опкодов</a></li><li><a href="/ru/learn/the-immediate/">18 Непосредственное значение</a></li></ol></li><li><h3>Модуль 7 · Троичная машина</h3><p>АЛУ из сумматоров этого курса, переход на три стороны и программа, которую можно пройти по шагам.</p><ol><li><a href="/ru/learn/a-ternary-alu/">19 Троичное АЛУ</a></li><li><a href="/ru/learn/three-way-jump/">20 Переход на три стороны</a></li><li><a href="/ru/learn/run-a-program/">21 Запустить программу</a></li></ol></li><li><h3>Модуль 8 · Три ответа</h3><p>Сравнение по старшему различному триту, поиск числа по третям и сортировка трёхзначными сравнениями.</p><ol><li><a href="/ru/learn/compare-at-the-top/">22 Сравнение сверху</a></li><li><a href="/ru/learn/questions-in-thirds/">23 Вопросы по третям</a></li><li><a href="/ru/learn/sort-by-three/">24 Сортировка трёхзначными сравнениями</a></li></ol></li><li><h3>Модуль 9 · Троичные нейроны</h3><p>Веса -1, 0, +1: один нейрон, детектор и маленький слой, и нигде нет умножителя.</p><ol><li><a href="/ru/learn/one-ternary-neuron/">25 Один троичный нейрон</a></li><li><a href="/ru/learn/a-bar-detector/">26 Детектор полосы</a></li><li><a href="/ru/learn/a-tiny-layer/">27 Маленький слой</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ternary-computing.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/wires-that-talk/index.html b/apps/website/public/ru/learn/wires-that-talk/index.html
index 58bb8fd95d..311e3a721a 100644
--- a/apps/website/public/ru/learn/wires-that-talk/index.html
+++ b/apps/website/public/ru/learn/wires-that-talk/index.html
@@ -299,5 +299,6 @@ <h1>Провода, которые говорят</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Что такое шина</h3><p>Зачем вообще существует шина: разговор по проводам, с кадрами и адресами, и кому позволено говорить.</p><ol><li class="on" aria-current="page"><a href="/ru/learn/wires-that-talk/">01 Провода, которые говорят</a></li><li><a href="/ru/learn/framing-a-conversation/">02 Кадрирование разговора</a></li><li><a href="/ru/learn/master-and-servant/">03 Хозяин и слуга</a></li></ol></li><li><h3>Модуль 2 · Разговор по UART</h3><p>Двухпроводная шина без такта: кадр, делитель, задающий скорость, и статус, который опрашивает драйвер.</p><ol><li><a href="/ru/learn/the-frame/">04 Кадр: старт, данные, стоп</a></li><li><a href="/ru/learn/baud-and-the-divisor/">05 Бод и делитель</a></li><li><a href="/ru/learn/status-and-fifos/">06 Статус и FIFO</a></li></ol></li><li><h3>Модуль 3 · SPI под тактом</h3><p>Разговор под тактом: четыре режима, лестница предделителей и выбор кристалла на каждого подчинённого.</p><ol><li><a href="/ru/learn/modes-cpol-cpha/">07 Режимы: CPOL и CPHA</a></li><li><a href="/ru/learn/the-prescaler-ladder/">08 Лестница предделителей</a></li><li><a href="/ru/learn/chip-select-and-widths/">09 Выбор кристалла и ширины</a></li></ol></li><li><h3>Модуль 4 · Регистровая шина APB</h3><p>Шина регистров: PSEL и PENABLE, строобы и ожидания, и сколько адресных бит стоит количество периферии.</p><ol><li><a href="/ru/learn/psel-and-penable/">10 PSEL и PENABLE</a></li><li><a href="/ru/learn/strobes-and-waits/">11 Строобы и ожидания</a></li><li><a href="/ru/learn/address-decode/">12 Дешифрация адреса</a></li></ol></li><li><h3>Модуль 5 · Пять каналов AXI4</h3><p>Пять каналов: адрес, данные и отклик в обе стороны, lite или full, пакеты и идентификаторы.</p><ol><li><a href="/ru/learn/five-channels/">13 Пять каналов</a></li><li><a href="/ru/learn/lite-or-full/">14 Lite или full</a></li><li><a href="/ru/learn/bursts-and-ids/">15 Пакеты и идентификаторы</a></li></ol></li><li><h3>Модуль 6 · Память</h3><p>Что находится на дальнем конце каждой шины: карты памяти, типы портов и задержка, которую должно покрывать ожидание.</p><ol><li><a href="/ru/learn/memory-maps/">16 Карты памяти</a></li><li><a href="/ru/learn/ports-and-kinds/">17 Порты и типы</a></li><li><a href="/ru/learn/latency/">18 Задержка</a></li></ol></li><li><h3>Модуль 7 · Мосты</h3><p>Почему в дизайне появляется больше одной шины и пакетный мост, переносящий работу между ними.</p><ol><li><a href="/ru/learn/why-bridges/">19 Зачем мосты</a></li><li><a href="/ru/learn/the-packet-bridge/">20 Пакетный мост</a></li><li><a href="/ru/learn/everything-in-one-bench/">21 Всё в одном стенде</a></li></ol></li><li><h3>Модуль 8 · Ethernet: кадры и тайминг</h3><p>Кадры, контрольная сумма кадра, тайминг RGMII и заранее зарегистрированные шаги настоящего запуска.</p><ol><li><a href="/ru/learn/frames-and-the-fcs/">22 Кадры и контрольная сумма</a></li><li><a href="/ru/learn/rgmii-timing/">23 Тайминг RGMII</a></li><li><a href="/ru/learn/bring-up-on-the-bench/">24 Запуск на стенде</a></li></ol></li><li><h3>Модуль 9 · Стенд</h3><p>Дисциплина, оберегающая настоящее железо: кто держит IO, как взять и вернуть захват, и что запускается дальше.</p><ol><li><a href="/ru/learn/who-holds-the-io/">25 Кто держит IO</a></li><li><a href="/ru/learn/claim-and-release/">26 Захват и возврат</a></li><li><a href="/ru/learn/capstone-what-runs-next/">27 Капстоун: что запускается дальше</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/buses-and-peripherals.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты хранят свои английские слова: у каждого своя спека t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/public/ru/learn/xor-needs-a-bend/index.html b/apps/website/public/ru/learn/xor-needs-a-bend/index.html
index e137351a00..d21c5be3a6 100644
--- a/apps/website/public/ru/learn/xor-needs-a-bend/index.html
+++ b/apps/website/public/ru/learn/xor-needs-a-bend/index.html
@@ -215,5 +215,6 @@ <h1>XOR нужен излом</h1>
 <section><h2>Все уроки</h2><ul class="outline"><li><h3>Модуль 1 · Лаборатория: наши исследования</h3><p>Свой формат чисел, честная таблица результатов и таблицы модели, перемноженные на плате.</p><ol><li><a href="/ru/learn/t27-number-format/">01 T27: наш собственный формат чисел</a></li><li><a href="/ru/learn/an-honest-scoreboard/">02 Честная таблица результатов</a></li><li><a href="/ru/learn/a-model-on-the-board/">03 Таблицы модели на плате</a></li></ol></li><li><h3>Модуль 2 · ИИ-числа: блок MX</h3><p>Как ИИ-чипы хранят веса в нескольких битах: один общий масштаб на блок, сам байт масштаба и что один выброс делает с соседями.</p><ol><li><a href="/ru/learn/one-scale-per-block/">04 Один масштаб на блок весов</a></li><li><a href="/ru/learn/scale-byte-native/">05 Байт масштаба на настоящей машине</a></li><li><a href="/ru/learn/one-outlier/">06 Один выброс, много нулей</a></li></ol></li><li><h3>Модуль 3 · Тернарные веса</h3><p>Веса, которые бывают только минус масштаб, ноль или плюс масштаб, пять правил, которые должен пройти тернарный алфавит, и прогон тестов, который ничего не проверил.</p><ol><li><a href="/ru/learn/phi-scaled-ternary/">07 Веса минус фи, ноль и плюс фи</a></li><li><a href="/ru/learn/a-pass-that-checked-nothing/">08 Прогон, который ничего не проверил</a></li><li><a href="/ru/learn/five-rules-for-weights/">09 Пять правил для алфавита весов</a></li></ol></li><li><h3>Модуль 4 · Тернарный формат Ternary Network Float</h3><p>Правило, которое компилятор проверяет до запуска любого теста, и 17-битное число с плавающей точкой, порядок которого — четыре сбалансированных трита.</p><ol><li><a href="/ru/learn/blocked-is-not-failed/">10 Заблокирован — не значит провален</a></li><li><a href="/ru/learn/four-trit-exponent/">11 Число, у которого порядок — четыре трита</a></li><li><a href="/ru/learn/negate-twice/">12 Дважды сменить знак — получить число обратно</a></li></ol></li><li><h3>Модуль 5 · Арифметика чисел со знаком</h3><p>Умножить два числа со знаком, сложить их, когда знаки разные, и сделать то и другое сразу в умножении с накоплением.</p><ol><li><a href="/ru/learn/a-sign-is-one-bit/">13 Знак — это один бит</a></li><li><a href="/ru/learn/opposite-signs-subtract/">14 Разные знаки — вычитание</a></li><li><a href="/ru/learn/multiply-then-add/">15 Умножить, потом сложить</a></li></ol></li><li><h3>Модуль 6 · Части нейрона</h3><p>ReLU с изломом в нуле, степень двойки для softmax и argmax, который называет ответ.</p><ol><li><a href="/ru/learn/a-bend-at-zero/">16 Излом в нуле</a></li><li><a href="/ru/learn/two-to-the-x/">17 Два в степени x</a></li><li><a href="/ru/learn/pick-the-largest/">18 Выбрать наибольшее</a></li></ol></li><li><h3>Модуль 7 · Обучение на ошибке</h3><p>Потеря, которая оценивает неверную догадку в битах, один шаг, который сдвигает вес против градиента, и скрытый слой, который нужен XOR.</p><ol><li><a href="/ru/learn/a-loss-in-bits/">19 Потеря в битах</a></li><li><a href="/ru/learn/one-step-downhill/">20 Один шаг вниз по склону</a></li><li class="on" aria-current="page"><a href="/ru/learn/xor-needs-a-bend/">21 XOR нужен излом</a></li></ol></li><li><h3>Модуль 8 · BitNet: тернарные сети</h3><p>Порог, который сжимает сумму обратно до трёх значений, один нейрон, который становится другой функцией при смене весов, и нейрон, который читает входы по 27 тритов за раз.</p><ol><li><a href="/ru/learn/back-to-three-values/">22 Обратно в три значения</a></li><li><a href="/ru/learn/same-neuron-new-weights/">23 Тот же нейрон, другие веса</a></li><li><a href="/ru/learn/a-neuron-in-chunks/">24 Нейрон по кускам</a></li></ol></li><li><h3>Модуль 9 · Тернарный MAC как чип</h3><p>Скалярное произведение 27 тритов как провода без регистра, та же сумма, которую регистр копит на каждом такте, и небольшая целая сеть в конце курса.</p><ol><li><a href="/ru/learn/a-dot-product-in-wires/">25 Скалярное произведение из проводов</a></li><li><a href="/ru/learn/add-it-up-every-clock/">26 Складывать на каждом такте</a></li><li><a href="/ru/learn/a-whole-network/">27 Целая сеть</a></li></ol></li></ul></section>
 <footer><p><a href="/learn/ai-numbers.t27">Этот курс сам является спекой t27: прочитайте её</a></p><p>Виджеты говорят по-английски: их слова живут в их собственных спеках t27.</p></footer>
 </main>
+<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>
 </body>
 </html>
diff --git a/apps/website/scripts/course-pages.mjs b/apps/website/scripts/course-pages.mjs
index 4bcac55ea7..68a9853897 100644
--- a/apps/website/scripts/course-pages.mjs
+++ b/apps/website/scripts/course-pages.mjs
@@ -68,6 +68,11 @@ export function loadCourses(file = join(SITE, 'src/lib/course.generated.ts')) {
 /** Site path of a page: the course's share path, or learn/<lesson>/; under ru/ for Russian. */
 export const pathOf = (C, lang, id) => `${lang === 'ru' ? 'ru/' : ''}${id ? `${BASE}/${id}/` : C.share}`
 const urlOf = (C, lang, id) => `${ORIGIN}/${pathOf(C, lang, id)}`
+// A shared link carries its utm tags (src/lib/shareTags.ts) to this page, and the app counts
+// them on arrival (src/lib/trafficArrival.ts) -- but this page is not the app. So its links
+// into the app take the query along, before the '#', where the app reads it.
+const CARRY_UTM = `<script>(function(){var q=location.search;if(!/[?&]utm_source=/.test(q))return;document.querySelectorAll('a[href^="/#/"]').forEach(function(a){a.setAttribute('href','/'+q+a.getAttribute('href').slice(1))})})()</script>`
+
 export const appOf = (C, id) => `/#/${C.route}${id ? `/${id}` : ''}`
 /** A neighbouring course's lesson page: lesson pages live under learn/ whatever the course. */
 const lessonPath = (lang, id) => `${lang === 'ru' ? 'ru/' : ''}${BASE}/${id}/`
@@ -248,6 +253,7 @@ ${body}
 <section><h2>${esc(say.ALL)}</h2><ul class="outline">${outline}</ul></section>
 <footer><p><a href="/${C.source.publicSpec.replace(/^public\//, '')}">${esc(say.SOURCE)}</a></p><p>${esc(say.WIDGET_LANG)}</p></footer>
 </main>
+${CARRY_UTM}
 </body>
 </html>
 `
diff --git a/apps/website/specs/ladder/course-ladder.t27 b/apps/website/specs/ladder/course-ladder.t27
index 4578715925..03c17cb5a1 100644
--- a/apps/website/specs/ladder/course-ladder.t27
+++ b/apps/website/specs/ladder/course-ladder.t27
@@ -8,7 +8,7 @@
 ; i18n/course-ladder.ru.json, every key, no extra) and where the render answers.
 ; Owner, 2026-10-11: any one course free, the second for the Telegram bot, the third free,
 ; the fourth for the club; every reader past the bot tracked in the CRM.
-; Owner, 2026-10-13: any three courses free, then payment; the bot rung lost every reader
+; Owner, 2026-10-13: any three courses free, with no sign-up, then payment; the bot rung lost every reader
 ; (rules VERSION 2), so no rung sends a reader to the bot for a key any more.
 ; What is honest here: the lesson words are in this site's bundle and on its share pages, so
 ; the ladder orders the reader's way, it does not hide the words (CONTENT_HIDDEN in the rules).
@@ -28,7 +28,7 @@ pub const RENDER_BASE : str = "https://vibee-render-production.up.railway.app";
 pub const OPEN_KEY : str = "t27-course-open";
 pub const KEYS_KEY : str = "t27-course-keys";
 
-pub const SAY_LADDER : str = "The ladder: your first three courses are free, the fourth and every one after it open with the club. You choose which courses are your three.";
+pub const SAY_LADDER : str = "The ladder: your first three courses are free with no sign-up, the fourth and every one after it open with the club. You choose which courses are your three.";
 pub const SAY_RUNG : str = "Rung {0} of your ladder";
 pub const SAY_FREE_TITLE : str = "This course is free for you";
 pub const SAY_FREE_LINE : str = "Open it with one press. Nothing is asked and nothing leaves this browser.";
@@ -64,7 +64,7 @@ pub const SAY_PAY_OR_BOT : str = "Or pay in Telegram Stars in our bot";
 ; was too hard to follow).
 pub const SAY_COURSES : str = "All courses";
 pub const SAY_CATALOG_TITLE : str = "Courses";
-pub const SAY_CATALOG_LEAD : str = "Every course is 27 short lessons in the browser, each with a live widget or a spec you can run. Pick any three and they are free; the club opens the rest.";
+pub const SAY_CATALOG_LEAD : str = "Every course is 27 short lessons in the browser, each with a live widget or a spec you can run. Pick any three and they are free, with no sign-up and no email; the club opens the rest.";
 pub const SAY_FREE_LEFT : str = "Free courses left: {0} of 3";
 pub const SAY_FREE_NONE : str = "Your three free courses are chosen. Every other course opens with the club.";
 pub const SAY_IN_CLUB : str = "You are in the club: every course is open.";
@@ -77,6 +77,18 @@ pub const SAY_DONE_OF : str = "{0} of {1} done";
 pub const SAY_FREE_SLOT : str = "Opening it uses free course {0} of 3.";
 pub const SAY_LAST_FREE : str = "This is your last free course: after it, courses open with the club.";
 pub const SAY_NOT_NOW : str = "Not now: back to the courses";
+; Selling where readers already are (owner, 2026-10-13: other courses at the foot of every
+; course page, a course advert in the footer and on every blog post). Internal links carry
+; ?from=<placement> -- never utm_*, which would overwrite the channel the reader came from --
+; and the placement rides on to the order, so a payment names the shelf that sold it.
+pub const SAY_MORE_TITLE : str = "More courses";
+pub const SAY_PROMO_TITLE : str = "Three courses free, no sign-up";
+pub const SAY_PROMO_LINE : str = "Every lesson runs in the browser, with a live widget or a spec you can run. Any three courses are free, no sign-up.";
+pub const SAY_PROMO_PRESS : str = "See the courses";
+; A reader who joined the club in the bot signs in here with the bot's button: the bot hands
+; back https://t27.ai/#/courses?unlock=<key> and the key stays in this browser.
+pub const SAY_SIGN_IN : str = "Already in the club? Sign in via Telegram";
+pub const SAY_SIGNED_IN : str = "You are signed in: the club key is saved in this browser.";
 pub const SAY_CHECKING : str = "Checking your key...";
 pub const SAY_KEY_BAD : str = "This key did not open the course ({0}). Press the button again for a fresh one.";
 pub const SAY_SHARED : str = "Your finished lessons are shared with the course author.";
diff --git a/apps/website/src/components/CourseShelf.tsx b/apps/website/src/components/CourseShelf.tsx
new file mode 100644
index 0000000000..e952f571b4
--- /dev/null
+++ b/apps/website/src/components/CourseShelf.tsx
@@ -0,0 +1,62 @@
+import { Link } from 'react-router-dom'
+import { COURSES } from '../lib/course.generated'
+import { LADDER, type LadderSay } from '../lib/courseLadder.generated'
+import { STEP, courseState, promoTo } from '../lib/courseLadder'
+import { pickShelf, useLadderView } from '../lib/useLadderView'
+import '../pages/Course.css'
+
+// THE COURSE CARDS, wherever a course is sold: the catalog (#/courses), the foot of every course
+// page, every blog post (owner, 2026-10-13: other courses at the bottom of each course page and
+// adverts across the blog). Each card says before the press what the press will do: yours,
+// free, or club. Every word is a spec's: the course specs' and specs/ladder/course-ladder.t27.
+
+type Lang = 'en' | 'ru'
+type CourseT = (typeof COURSES)[number]
+const fmt = (s: string, ...args: (string | number)[]) =>
+  args.reduce<string>((out, a, i) => out.split(`{${i}}`).join(String(a)), s)
+
+export function CourseCard({ c, lang, view, placement }: { c: CourseT; lang: Lang; view: ReturnType<typeof useLadderView>; placement: string }) {
+  const say = LADDER.say[lang] as LadderSay
+  const s = courseState(c.id, view.opened, view.valid)
+  const total = c.lessons.length
+  const finished = c.lessons.filter((l) => view.done.includes(l.id)).length
+  const badge = s.known && s.open ? say.BADGE_OPEN : view.valid.paid || s.step !== STEP.paid ? say.BADGE_FREE : say.BADGE_CLUB
+  const cls = s.known && s.open ? 'is-open' : badge === say.BADGE_CLUB ? 'is-club' : 'is-free'
+  return (
+    <li className={`course-card ${cls}`}>
+      <Link to={promoTo(c.route, placement)} className="course-card-link">
+        <span className="course-badge">{badge}</span>
+        <span className="course-card-title">{c.say[lang].TITLE}</span>
+        <span className="course-card-desc">{c.say[lang].DESCRIPTION}</span>
+        <span className="course-card-foot">{finished > 0 ? fmt(say.DONE_OF, finished, total) : fmt(say.LESSONS, total)}</span>
+        {finished > 0 ? (
+          <span className="course-card-bar" aria-hidden="true">
+            <span style={{ width: `${(100 * finished) / total}%` }} />
+          </span>
+        ) : null}
+      </Link>
+    </li>
+  )
+}
+
+export default function CourseShelf({ lang, placement, exclude, limit = 3, title }: { lang: Lang; placement: string; exclude?: string; limit?: number; title?: string }) {
+  const say = LADDER.say[lang] as LadderSay
+  const view = useLadderView()
+  const picks = pickShelf(exclude, view.opened, limit)
+  if (!picks.length) return null
+  return (
+    <section className="course-shelf" aria-label={title ?? say.MORE_TITLE}>
+      <div className="course-shelf-head">
+        <h2 className="course-shelf-title">{title ?? say.MORE_TITLE}</h2>
+        <Link className="course-link" to={promoTo('courses', placement)}>
+          {say.COURSES} →
+        </Link>
+      </div>
+      <ul className="course-catalog">
+        {picks.map((c) => (
+          <CourseCard key={c.id} c={c} lang={lang} view={view} placement={placement} />
+        ))}
+      </ul>
+    </section>
+  )
+}
diff --git a/apps/website/src/components/Footer.tsx b/apps/website/src/components/Footer.tsx
index 43124e25ce..b9476bf8bb 100644
--- a/apps/website/src/components/Footer.tsx
+++ b/apps/website/src/components/Footer.tsx
@@ -2,6 +2,8 @@
 import { motion } from 'framer-motion'
 import { Link } from 'react-router-dom'
 import { useI18n } from '../i18n/context'
+import { LADDER } from '../lib/courseLadder.generated'
+import { promoTo } from '../lib/courseLadder'
 
 // Smooth scrolling is animation-driven, so it silently does nothing when
 // animations are not running — a hidden or backgrounded tab, or a reader who has
@@ -32,7 +34,51 @@ function goToSection(e: React.MouseEvent, id: string) {
   tick()
 }
 
-export default function Footer() {
+/**
+ * The course advert on every page that is not already selling a course (owner, 2026-10-13:
+ * adverts in the footer). Only the ladder's words, which the header already loads: the course
+ * list itself (course.generated.ts, ~850 KB) stays out of the bundle every page pays for.
+ */
+function CoursePromo({ lang }: { lang: string }) {
+  const say = LADDER.say[lang === 'ru' ? 'ru' : 'en']
+  return (
+    <aside
+      aria-label={say.CATALOG_TITLE}
+      style={{
+        display: 'flex',
+        flexWrap: 'wrap',
+        alignItems: 'center',
+        justifyContent: 'space-between',
+        gap: '1rem 2rem',
+        padding: 'clamp(1rem, 4vw, 1.5rem)',
+        marginBottom: 'clamp(1.5rem, 5vw, 3rem)',
+        border: '1px solid var(--border)',
+        borderRadius: '12px',
+      }}
+    >
+      <div style={{ flex: '1 1 320px' }}>
+        <p style={{ margin: 0, fontWeight: 700, fontSize: '1.05rem', color: 'var(--text)' }}>{say.PROMO_TITLE}</p>
+        <p style={{ margin: '0.4rem 0 0', color: 'var(--muted)', fontSize: '0.9rem', lineHeight: 1.6 }}>{say.PROMO_LINE}</p>
+      </div>
+      <Link
+        to={promoTo('courses', 'footer')}
+        style={{
+          padding: '0.7rem 1.2rem',
+          borderRadius: '999px',
+          background: 'var(--accent)',
+          color: '#000',
+          fontWeight: 700,
+          textDecoration: 'none',
+          whiteSpace: 'nowrap',
+        }}
+      >
+        {say.PROMO_PRESS} →
+      </Link>
+    </aside>
+  )
+}
+
+export default function Footer({ promo = true }: { promo?: boolean } = {}) {
   /* lang нужен подписи «Инвестиции» в списке ссылок: ключа nav[9] в локалях
      нет, и на русской странице там висело английское «Invest». */
   const { t, lang } = useI18n()
@@ -49,6 +95,7 @@ export default function Footer() {
       aria-label="Site footer"
     >
       <div style={{ maxWidth: '1200px', margin: '0 auto' }}>
+        {promo ? <CoursePromo lang={lang} /> : null}
         {/* Main Footer Content */}
         <div style={{ 
           display: 'grid', 
diff --git a/apps/website/src/lib/courseLadder.generated.ts b/apps/website/src/lib/courseLadder.generated.ts
index c13a924c60..72c9cc116d 100644
--- a/apps/website/src/lib/courseLadder.generated.ts
+++ b/apps/website/src/lib/courseLadder.generated.ts
@@ -9,7 +9,7 @@ export const LADDER = {
   "keysKey": "t27-course-keys",
   "say": {
     "en": {
-      "LADDER": "The ladder: your first three courses are free, the fourth and every one after it open with the club. You choose which courses are your three.",
+      "LADDER": "The ladder: your first three courses are free with no sign-up, the fourth and every one after it open with the club. You choose which courses are your three.",
       "RUNG": "Rung {0} of your ladder",
       "FREE_TITLE": "This course is free for you",
       "FREE_LINE": "Open it with one press. Nothing is asked and nothing leaves this browser.",
@@ -36,7 +36,7 @@ export const LADDER = {
       "PAY_OR_BOT": "Or pay in Telegram Stars in our bot",
       "COURSES": "All courses",
       "CATALOG_TITLE": "Courses",
-      "CATALOG_LEAD": "Every course is 27 short lessons in the browser, each with a live widget or a spec you can run. Pick any three and they are free; the club opens the rest.",
+      "CATALOG_LEAD": "Every course is 27 short lessons in the browser, each with a live widget or a spec you can run. Pick any three and they are free, with no sign-up and no email; the club opens the rest.",
       "FREE_LEFT": "Free courses left: {0} of 3",
       "FREE_NONE": "Your three free courses are chosen. Every other course opens with the club.",
       "IN_CLUB": "You are in the club: every course is open.",
@@ -49,12 +49,18 @@ export const LADDER = {
       "FREE_SLOT": "Opening it uses free course {0} of 3.",
       "LAST_FREE": "This is your last free course: after it, courses open with the club.",
       "NOT_NOW": "Not now: back to the courses",
+      "MORE_TITLE": "More courses",
+      "PROMO_TITLE": "Three courses free, no sign-up",
+      "PROMO_LINE": "Every lesson runs in the browser, with a live widget or a spec you can run. Any three courses are free, no sign-up.",
+      "PROMO_PRESS": "See the courses",
+      "SIGN_IN": "Already in the club? Sign in via Telegram",
+      "SIGNED_IN": "You are signed in: the club key is saved in this browser.",
       "CHECKING": "Checking your key...",
       "KEY_BAD": "This key did not open the course ({0}). Press the button again for a fresh one.",
       "SHARED": "Your finished lessons are shared with the course author."
     },
     "ru": {
-      "LADDER": "Лестница: первые три курса бесплатные, четвёртый и все следующие открываются с клубом. Какие три — выбираете вы.",
+      "LADDER": "Лестница: первые три курса бесплатные и без регистрации, четвёртый и все следующие открываются с клубом. Какие три — выбираете вы.",
       "RUNG": "Ступень {0} вашей лестницы",
       "FREE_TITLE": "Этот курс для вас бесплатный",
       "FREE_LINE": "Откройте одним нажатием. Ничего не спрашиваем, из браузера ничего не уходит.",
@@ -81,7 +87,7 @@ export const LADDER = {
       "PAY_OR_BOT": "Или оплатить звёздами Telegram в нашем боте",
       "COURSES": "Все курсы",
       "CATALOG_TITLE": "Курсы",
-      "CATALOG_LEAD": "Каждый курс — 27 коротких уроков в браузере, в каждом живой виджет или спека, которую можно запустить. Любые три — бесплатно, остальные открывает клуб.",
+      "CATALOG_LEAD": "Каждый курс — 27 коротких уроков в браузере, в каждом живой виджет или спека, которую можно запустить. Любые три — бесплатно, без регистрации и почты; остальные открывает клуб.",
       "FREE_LEFT": "Осталось бесплатных курсов: {0} из 3",
       "FREE_NONE": "Три бесплатных курса выбраны. Все остальные открываются с клубом.",
       "IN_CLUB": "Вы в клубе: открыты все курсы.",
@@ -94,6 +100,12 @@ export const LADDER = {
       "FREE_SLOT": "Он станет вашим бесплатным курсом {0} из 3.",
       "LAST_FREE": "Это ваш последний бесплатный курс: дальше курсы открываются с клубом.",
       "NOT_NOW": "Не сейчас: к списку курсов",
+      "MORE_TITLE": "Другие курсы",
+      "PROMO_TITLE": "Три курса бесплатно, без регистрации",
+      "PROMO_LINE": "Каждый урок идёт прямо в браузере, с живым виджетом или спецификацией, которую можно запустить. Любые три курса бесплатно, без регистрации.",
+      "PROMO_PRESS": "Смотреть курсы",
+      "SIGN_IN": "Уже в клубе? Войти через Telegram",
+      "SIGNED_IN": "Вы вошли: ключ клуба сохранён в этом браузере.",
       "CHECKING": "Проверяем ключ...",
       "KEY_BAD": "Этот ключ не открыл курс ({0}). Нажмите кнопку ещё раз — бот выдаст новый.",
       "SHARED": "Ваши пройденные уроки видит автор курса."
diff --git a/apps/website/src/lib/courseLadder.ts b/apps/website/src/lib/courseLadder.ts
index 9a1855afd5..b3fd0b7e20 100644
--- a/apps/website/src/lib/courseLadder.ts
+++ b/apps/website/src/lib/courseLadder.ts
@@ -9,6 +9,8 @@
 import * as RULES from './courseAccess.gen'
 import { mayOpen, mayReport, staysOpen, stepOf } from './courseAccess.logic'
 import { LADDER } from './courseLadder.generated'
+// The utm_source this session arrived with, as trafficArrival.ts keeps it.
+import { SOURCE_KEY } from './trafficArrival'
 
 /** The courses this reader opened, in the order they were opened, with the rung each took. */
 export type Opened = { id: string; step: number }[]
@@ -122,7 +124,8 @@ export async function startCheckout(email: string, route: string, lang: string):
       method: 'POST',
       credentials: 'omit',
       headers: { 'Content-Type': 'application/json' },
-      body: JSON.stringify({ email, route, lang }),
+      // The shelf that sold it and the channel the reader came from: the owner's notice names both.
+      body: JSON.stringify({ email, route, lang, from: readPromo(), source: readSource() }),
     })
     const body = await r.json()
     return body && body.ok === true && typeof body.url === 'string' && body.url.startsWith('https://') ? body : { ok: false, reason: body?.reason ?? `http_${r.status}` }
@@ -143,6 +146,45 @@ export async function readOrder(id: string): Promise<Order> {
 
 export const ORDER_PARAM = 'order'
 
+/**
+ * WHICH SHELF SOLD IT. An internal link to a course carries ?from=<placement> (footer,
+ * blog:<slug>, course:<route>, catalog) inside the hash. Never utm_*: an internal utm tag
+ * overwrites the channel the reader really came from. The first placement of this session is
+ * kept (blog -> catalog -> course names the blog, not the catalog) and sent with an order, so a
+ * payment names the shelf that sold it.
+ */
+export const PROMO_PARAM = 'from'
+const PROMO_KEY = 'traffic:promo'
+
+/** A course link from a shelf: /<route>?from=<placement>. Pure. */
+export function promoTo(route: string, placement: string): string {
+  return `/${route}?${PROMO_PARAM}=${encodeURIComponent(placement)}`
+}
+
+/** Keep the placement this address names, if any. Never throws. */
+export function notePromo(search: string) {
+  const from = new URLSearchParams(search).get(PROMO_PARAM)
+  if (!from || !/^[a-z0-9:_-]{1,80}$/i.test(from)) return
+  try {
+    if (!window.sessionStorage.getItem(PROMO_KEY)) window.sessionStorage.setItem(PROMO_KEY, from)
+  } catch {
+    // no storage: the order goes without its shelf
+  }
+}
+
+function readSession(key: string): string | undefined {
+  try {
+    return window.sessionStorage.getItem(key) ?? undefined
+  } catch {
+    return undefined
+  }
+}
+export const readPromo = () => readSession(PROMO_KEY)
+export const readSource = () => readSession(SOURCE_KEY)
+
+/** The catalog's address in the bot: a club member signs in here with the key it hands back. */
+export const SIGN_IN_ROUTE = 'courses'
+
 /** A finished lesson, sent with the reader's key -- and never without one (may_report). */
 export async function reportDone(key: string | undefined, course: string, lesson: string): Promise<void> {
   if (!mayReport(Boolean(key))) return
diff --git a/apps/website/src/lib/shareTags.ts b/apps/website/src/lib/shareTags.ts
new file mode 100644
index 0000000000..cca2484b29
--- /dev/null
+++ b/apps/website/src/lib/shareTags.ts
@@ -0,0 +1,16 @@
+/**
+ * A SHARED LINK SAYS WHERE IT WAS SHARED. Owner, 2026-10-13: utm tags everywhere, to see the
+ * traffic. The share buttons are where our links leave for other sites, so that is where the
+ * tags go: utm_source=<platform>, utm_medium=social, utm_campaign=<blog|course>,
+ * utm_content=<slug>. The query sits before any '#', where trafficArrival.ts reads it on arrival.
+ *
+ * Internal links never carry utm_* (an internal tag overwrites the channel the reader came
+ * from); they carry ?from=<placement> instead (courseLadder.ts promoTo).
+ */
+export function shareTagged(url: string, source: string, campaign: string, content: string): string {
+  const hash = url.indexOf('#')
+  const base = hash >= 0 ? url.slice(0, hash) : url
+  const tail = hash >= 0 ? url.slice(hash) : ''
+  const q = new URLSearchParams({ utm_source: source, utm_medium: 'social', utm_campaign: campaign, utm_content: content })
+  return `${base}${base.includes('?') ? '&' : '?'}${q.toString()}${tail}`
+}
diff --git a/apps/website/src/lib/useLadderView.ts b/apps/website/src/lib/useLadderView.ts
new file mode 100644
index 0000000000..264b4a0c50
--- /dev/null
+++ b/apps/website/src/lib/useLadderView.ts
@@ -0,0 +1,75 @@
+import { useEffect, useMemo, useState } from 'react'
+import { useLocation, useNavigate } from 'react-router-dom'
+import { COURSE, COURSES } from './course.generated'
+import { KEY_PARAM, STEP, checkKey, readKeys, readOpened, saveKeys, type Keys, type Opened } from './courseLadder'
+
+// What a course card needs to know about this reader, shared by the catalog (#/courses) and the
+// course shelves (components/CourseShelf.tsx).
+
+type CourseT = (typeof COURSES)[number]
+
+function readDone(): string[] {
+  try {
+    const list = JSON.parse(window.localStorage.getItem(COURSE.progressKey) ?? '[]')
+    return Array.isArray(list) ? list.filter((x): x is string => typeof x === 'string') : []
+  } catch {
+    return []
+  }
+}
+
+/**
+ * What this reader holds: the courses opened, the lessons done, and whether the keys still hold.
+ * With takeKey, a key in the address (?unlock=<key>, the bot's sign-in button) is checked, kept
+ * and taken out of the address, so a club member who joined in Telegram is signed in here.
+ */
+export function useLadderView(takeKey = false) {
+  const done = useMemo(() => (typeof window === 'undefined' ? [] : readDone()), [])
+  const known = useMemo(() => COURSES.map((c) => c.id as string), [])
+  const withProgress = useMemo(() => COURSES.filter((c) => c.lessons.some((l) => done.includes(l.id))).map((c) => c.id as string), [done])
+  const opened: Opened = useMemo(() => (typeof window === 'undefined' ? [] : readOpened(known, withProgress)), [known, withProgress])
+  const [valid, setValid] = useState({ bot: false, paid: false })
+  const [signedIn, setSignedIn] = useState(false)
+  const { search, pathname } = useLocation()
+  const navigate = useNavigate()
+
+  useEffect(() => {
+    let live = true
+    const fresh = takeKey ? new URLSearchParams(search).get(KEY_PARAM) : null
+    void (async () => {
+      const held: Keys = { ...readKeys() }
+      if (fresh) {
+        const r = await checkKey(fresh)
+        if (r.ok && (r.step === STEP.bot || r.step === STEP.paid)) {
+          if (r.step === STEP.bot) held.bot = fresh
+          else held.paid = fresh
+          saveKeys(held)
+          if (live) setSignedIn(r.step === STEP.paid)
+        }
+      }
+      const [b, p] = await Promise.all([held.bot ? checkKey(held.bot) : { ok: false }, held.paid ? checkKey(held.paid) : { ok: false }])
+      if (!live) return
+      setValid({ bot: Boolean(b.ok), paid: Boolean(p.ok) })
+      if (fresh) navigate(pathname, { replace: true })
+    })()
+    return () => {
+      live = false
+    }
+    // Once per arrival: a new key in the address.
+    // eslint-disable-next-line react-hooks/exhaustive-deps
+  }, [search])
+
+  return { done, opened, valid, signedIn }
+}
+
+/**
+ * A few courses to go on with, at the foot of a page. The next courses after the one being read
+ * come first, so each course page shows different ones; a course not opened yet goes before one
+ * already opened, because the shelf is there to show the reader something new.
+ */
+export function pickShelf(exclude: string | undefined, opened: Opened, limit: number): CourseT[] {
+  const at = Math.max(0, COURSES.findIndex((c) => c.id === exclude))
+  const ring = [...COURSES.slice(at + 1), ...COURSES.slice(0, at + 1)].filter((c) => c.id !== exclude)
+  const isOpened = (c: CourseT) => opened.some((o) => o.id === c.id)
+  return [...ring.filter((c) => !isOpened(c)), ...ring.filter(isOpened)].slice(0, limit)
+}
+
diff --git a/apps/website/src/pages/Blog.tsx b/apps/website/src/pages/Blog.tsx
index 938b353853..21decb78b1 100644
--- a/apps/website/src/pages/Blog.tsx
+++ b/apps/website/src/pages/Blog.tsx
@@ -1,4 +1,4 @@
-import { useEffect, useState } from 'react'
+import { Suspense, lazy, useEffect, useState } from 'react'
 import { Link, useParams } from 'react-router-dom'
 import Navigation from '../components/Navigation'
 import Footer from '../components/Footer'
@@ -7,6 +7,11 @@ import TerminalCast from '../components/TerminalCast'
 import { publishedPosts, postBySlug } from '../data/blog/index'
 import type { Block, Post, PostBody, PostMeta } from '../data/blog/types'
 import { useI18n } from '../i18n/context'
+import { shareTagged } from '../lib/shareTags'
+
+// The courses under every post (owner, 2026-10-13: adverts across the blog). Lazy: the course
+// list is ~850 KB, so it loads after the article and never delays it.
+const CourseShelf = lazy(() => import('../components/CourseShelf'))
 
 // The blog was English-only whatever the language switcher said. These are the
 // page's own words; a post's words travel with the post, in its `ru` field.
@@ -486,18 +491,18 @@ function Share({ post, lang }: { post: PostMeta; lang: string }) {
   const t = ui(lang)
   const [copied, setCopied] = useState(false)
   const url = postUrl(post.slug, lang)
-  const u = encodeURIComponent(url)
+  const tag = (source: string) => encodeURIComponent(shareTagged(url, source, 'blog', post.slug))
   const title = encodeURIComponent(post.title)
   const hashtags = post.tags.map(blogHashtag)
   const xHashtag = encodeURIComponent(hashtags[0].slice(1))
   const telegramText = encodeURIComponent(`${post.title}\n\n${hashtags.slice(0, 3).join(' ')}`)
 
   const targets = [
-    { name: 'X', href: `https://twitter.com/intent/tweet?url=${u}&text=${title}&hashtags=${xHashtag}` },
-    { name: 'LinkedIn', href: `https://www.linkedin.com/sharing/share-offsite/?url=${u}` },
-    { name: 'Telegram', href: `https://t.me/share/url?url=${u}&text=${telegramText}` },
-    { name: 'Hacker News', href: `https://news.ycombinator.com/submitlink?u=${u}&t=${title}` },
-    { name: 'Reddit', href: `https://www.reddit.com/submit?url=${u}&title=${title}` },
+    { name: 'X', href: `https://twitter.com/intent/tweet?url=${tag('x')}&text=${title}&hashtags=${xHashtag}` },
+    { name: 'LinkedIn', href: `https://www.linkedin.com/sharing/share-offsite/?url=${tag('linkedin')}` },
+    { name: 'Telegram', href: `https://t.me/share/url?url=${tag('telegram')}&text=${telegramText}` },
+    { name: 'Hacker News', href: `https://news.ycombinator.com/submitlink?u=${tag('hackernews')}&t=${title}` },
+    { name: 'Reddit', href: `https://www.reddit.com/submit?url=${tag('reddit')}&title=${title}` },
   ]
 
   return (
@@ -518,7 +523,7 @@ function Share({ post, lang }: { post: PostMeta; lang: string }) {
         <button
           type="button"
           onClick={() => {
-            navigator.clipboard?.writeText(url).then(
+            navigator.clipboard?.writeText(shareTagged(url, 'copy', 'blog', post.slug)).then(
               () => {
                 setCopied(true)
                 window.setTimeout(() => setCopied(false), 2000)
@@ -715,6 +720,9 @@ export function BlogPost() {
         {post.body.map(renderBlock)}
         <Receipts post={post} lang={lang} />
         <Share post={post} lang={lang} />
+        <Suspense fallback={null}>
+          <CourseShelf lang={lang === 'ru' ? 'ru' : 'en'} placement={`blog:${post.slug}`} />
+        </Suspense>
 
         <p style={{ marginTop: '3em' }}>
           <Link to="/blog" className="blog-link">{t.allPosts}</Link>
diff --git a/apps/website/src/pages/Course.css b/apps/website/src/pages/Course.css
index 929b013a5e..1a22c4709a 100644
--- a/apps/website/src/pages/Course.css
+++ b/apps/website/src/pages/Course.css
@@ -735,3 +735,24 @@
   opacity: 0.6;
   cursor: progress;
 }
+
+/* Other courses, at the foot of a course page or a blog post. */
+.course-shelf {
+  margin: 64px 0 0;
+  padding-top: 32px;
+  border-top: 1px solid var(--border, rgba(255, 255, 255, 0.12));
+}
+.course-shelf-head {
+  display: flex;
+  flex-wrap: wrap;
+  align-items: baseline;
+  justify-content: space-between;
+  gap: 8px 16px;
+}
+.course-shelf-title {
+  margin: 0;
+  font-size: 1.25rem;
+}
+.course-shelf .course-catalog {
+  margin-top: 20px;
+}
diff --git a/apps/website/src/pages/Course.tsx b/apps/website/src/pages/Course.tsx
index cb3f902f70..b27365a969 100644
--- a/apps/website/src/pages/Course.tsx
+++ b/apps/website/src/pages/Course.tsx
@@ -4,9 +4,11 @@ import { usePageMeta } from '../hooks/usePageMeta'
 import { useI18n } from '../i18n/context'
 import Navigation from '../components/Navigation'
 import Footer from '../components/Footer'
+import CourseShelf from '../components/CourseShelf'
+import { shareTagged } from '../lib/shareTags'
 import { COURSE, COURSES, type Course as CourseT, type CourseLesson } from '../lib/course.generated'
 import { LADDER, type LadderSay } from '../lib/courseLadder.generated'
-import { KEY_PARAM, ORDER_PARAM, STEP, checkKey, checkoutOffer, courseState, freeLeft, openCourse, readOrder, startCheckout, type Offer, readKeys, readOpened, reportDone, saveKeys, startLink, type Keys, type Opened } from '../lib/courseLadder'
+import { KEY_PARAM, ORDER_PARAM, STEP, notePromo, checkKey, checkoutOffer, courseState, freeLeft, openCourse, readOrder, startCheckout, type Offer, readKeys, readOpened, reportDone, saveKeys, startLink, type Keys, type Opened } from '../lib/courseLadder'
 import './Course.css'
 
 // Every word, the order of the courses and lessons and the widget each one opens
@@ -168,12 +170,13 @@ const shareOf = (C: CourseT, lang: Lang, id?: string) => `${lang === 'ru' ? '/ru
 function CourseShare({ C, lang, say, id, title }: { C: CourseT; lang: Lang; say: Say; id?: string; title: string }) {
   const path = shareOf(C, lang, id)
   const url = `https://t27.ai${path}`
-  const u = encodeURIComponent(url)
+  // Each button's link says where it was shared (utm_source=<platform>, lib/shareTags.ts).
+  const tag = (source: string) => encodeURIComponent(shareTagged(url, source, 'course', C.route))
   const text = encodeURIComponent(title)
   const [copied, setCopied] = useState(false)
   const copy = async () => {
     try {
-      await navigator.clipboard.writeText(url)
+      await navigator.clipboard.writeText(shareTagged(url, 'copy', 'course', C.route))
       setCopied(true)
       window.setTimeout(() => setCopied(false), 1400)
     } catch {
@@ -189,13 +192,13 @@ function CourseShare({ C, lang, say, id, title }: { C: CourseT; lang: Lang; say:
       <button type="button" className="course-share-pill" onClick={() => void copy()} title={say.SHARE} aria-label={say.SHARE}>
         {copied ? '✓' : '⧉'}
       </button>
-      <a className="course-share-pill" href={`https://twitter.com/intent/tweet?url=${u}&text=${text}`} target="_blank" rel="noopener noreferrer">
+      <a className="course-share-pill" href={`https://twitter.com/intent/tweet?url=${tag('x')}&text=${text}`} target="_blank" rel="noopener noreferrer">
         X
       </a>
-      <a className="course-share-pill" href={`https://t.me/share/url?url=${u}&text=${text}`} target="_blank" rel="noopener noreferrer">
+      <a className="course-share-pill" href={`https://t.me/share/url?url=${tag('telegram')}&text=${text}`} target="_blank" rel="noopener noreferrer">
         Telegram
       </a>
-      <a className="course-share-pill" href={`https://www.linkedin.com/sharing/share-offsite/?url=${u}`} target="_blank" rel="noopener noreferrer">
+      <a className="course-share-pill" href={`https://www.linkedin.com/sharing/share-offsite/?url=${tag('linkedin')}`} target="_blank" rel="noopener noreferrer">
         LinkedIn
       </a>
     </div>
@@ -409,6 +412,8 @@ function useLadder(C: CourseT, allDone: string[]) {
   // 'slow' after half a minute of it, 'done' once the key is in hand.
   const [paying, setPaying] = useState<'idle' | 'waiting' | 'slow' | 'done'>('idle')
 
+  useEffect(() => notePromo(search), [search])
+
   useEffect(() => {
     let live = true
     void checkoutOffer().then((o) => live && setOffer(o))
@@ -620,6 +625,13 @@ function LadderGate({ C, lang, ladder }: { C: CourseT; lang: Lang; ladder: Retur
           </Link>
         ) : null}
       </div>
+      {step === STEP.paid && !checking ? (
+        <p className="course-desc">
+          <a className="course-link" href={startLink(STEP.paid, C.route)} target="_blank" rel="noopener noreferrer">
+            {say.SIGN_IN}
+          </a>
+        </p>
+      ) : null}
       {step !== STEP.free ? <p className="course-desc">{say.BOT_SHARES}</p> : null}
       {step === STEP.paid && !rub ? <p className="course-desc">{say.PAID_LATER}</p> : null}
       <p className="course-shape">{say.LADDER}</p>
@@ -689,8 +701,10 @@ export default function Course() {
             <Overview C={C} lang={lang} say={say} done={done} shared={shared} />
           </>
         )}
+        {/* Other courses at the foot of every course page (owner, 2026-10-13). */}
+        <CourseShelf lang={lang} placement={`course:${C.route}`} exclude={C.id} />
       </main>
-      <Footer />
+      <Footer promo={false} />
     </div>
   )
 }
diff --git a/apps/website/src/pages/Courses.tsx b/apps/website/src/pages/Courses.tsx
index 514b39162d..5ecab136c0 100644
--- a/apps/website/src/pages/Courses.tsx
+++ b/apps/website/src/pages/Courses.tsx
@@ -1,12 +1,14 @@
-import { useEffect, useMemo, useState } from 'react'
-import { Link } from 'react-router-dom'
+import { Link, useLocation } from 'react-router-dom'
+import { useEffect } from 'react'
 import { usePageMeta } from '../hooks/usePageMeta'
 import { useI18n } from '../i18n/context'
 import Navigation from '../components/Navigation'
 import Footer from '../components/Footer'
-import { COURSE, COURSES, type CourseLesson } from '../lib/course.generated'
+import { CourseCard } from '../components/CourseShelf'
+import { useLadderView } from '../lib/useLadderView'
+import { COURSES, type CourseLesson } from '../lib/course.generated'
 import { LADDER, type LadderSay } from '../lib/courseLadder.generated'
-import { STEP, checkKey, courseState, freeLeft, readKeys, readOpened } from '../lib/courseLadder'
+import { SIGN_IN_ROUTE, STEP, freeLeft, notePromo, startLink } from '../lib/courseLadder'
 import './Course.css'
 
 // THE CATALOG (#/courses). Owner, 2026-10-13: the navigation was too hard to follow -- the
@@ -19,37 +21,15 @@ type Lang = 'en' | 'ru'
 const fmt = (s: string, ...args: (string | number)[]) =>
   args.reduce<string>((out, a, i) => out.split(`{${i}}`).join(String(a)), s)
 
-function readDone(): string[] {
-  try {
-    const list = JSON.parse(window.localStorage.getItem(COURSE.progressKey) ?? '[]')
-    return Array.isArray(list) ? list.filter((x): x is string => typeof x === 'string') : []
-  } catch {
-    return []
-  }
-}
-
 export default function Courses() {
   const { lang: siteLang } = useI18n()
   const lang: Lang = siteLang === 'ru' ? 'ru' : 'en'
   const say = LADDER.say[lang] as LadderSay
   usePageMeta(say.CATALOG_TITLE, say.CATALOG_LEAD)
-  const done = useMemo(() => (typeof window === 'undefined' ? [] : readDone()), [])
-  const known = useMemo(() => COURSES.map((c) => c.id as string), [])
-  const withProgress = useMemo(() => COURSES.filter((c) => c.lessons.some((l) => done.includes(l.id))).map((c) => c.id as string), [done])
-  const opened = useMemo(() => (typeof window === 'undefined' ? [] : readOpened(known, withProgress)), [known, withProgress])
-  const [valid, setValid] = useState({ bot: false, paid: false })
-
-  useEffect(() => {
-    let live = true
-    const held = readKeys()
-    void Promise.all([held.bot ? checkKey(held.bot) : { ok: false }, held.paid ? checkKey(held.paid) : { ok: false }]).then(([b, p]) => {
-      if (live) setValid({ bot: Boolean(b.ok), paid: Boolean(p.ok) })
-    })
-    return () => {
-      live = false
-    }
-  }, [])
-
+  const view = useLadderView(true)
+  const { done, opened, valid } = view
+  const { search } = useLocation()
+  useEffect(() => notePromo(search), [search])
   const left = freeLeft(opened)
   // The course to continue: the latest one opened that still has a lesson to do.
   const resume = [...opened]
@@ -68,6 +48,7 @@ export default function Courses() {
           <p className="course-counter" aria-live="polite">
             {valid.paid ? say.IN_CLUB : left > 0 ? fmt(say.FREE_LEFT, left) : say.FREE_NONE}
           </p>
+          {view.signedIn ? <p className="course-desc">{say.SIGNED_IN}</p> : null}
           {resume && resumeLesson ? (
             <div className="course-hero-actions">
               <Link className="course-btn is-primary" to={`/${resume.route}/${resumeLesson.id}`}>
@@ -75,35 +56,21 @@ export default function Courses() {
               </Link>
             </div>
           ) : null}
+          {!valid.paid ? (
+            <p className="course-desc">
+              <a className="course-link" href={startLink(STEP.paid, SIGN_IN_ROUTE)} target="_blank" rel="noopener noreferrer">
+                {say.SIGN_IN}
+              </a>
+            </p>
+          ) : null}
         </div>
         <ul className="course-catalog">
-          {COURSES.map((c) => {
-            const s = courseState(c.id, opened, valid)
-            const total = c.lessons.length
-            const finished = c.lessons.filter((l) => done.includes(l.id)).length
-            const badge = s.known && s.open ? say.BADGE_OPEN : valid.paid || s.step !== STEP.paid ? say.BADGE_FREE : say.BADGE_CLUB
-            const cls = s.known && s.open ? 'is-open' : badge === say.BADGE_CLUB ? 'is-club' : 'is-free'
-            return (
-              <li key={c.id} className={`course-card ${cls}`}>
-                <Link to={`/${c.route}`} className="course-card-link">
-                  <span className="course-badge">{badge}</span>
-                  <span className="course-card-title">{c.say[lang].TITLE}</span>
-                  <span className="course-card-desc">{c.say[lang].DESCRIPTION}</span>
-                  <span className="course-card-foot">
-                    {finished > 0 ? fmt(say.DONE_OF, finished, total) : fmt(say.LESSONS, total)}
-                  </span>
-                  {finished > 0 ? (
-                    <span className="course-card-bar" aria-hidden="true">
-                      <span style={{ width: `${(100 * finished) / total}%` }} />
-                    </span>
-                  ) : null}
-                </Link>
-              </li>
-            )
-          })}
+          {COURSES.map((c) => (
+            <CourseCard key={c.id} c={c} lang={lang} view={view} placement="catalog" />
+          ))}
         </ul>
       </main>
-      <Footer />
+      <Footer promo={false} />
     </div>
   )
 }