Skip to content

ci: adopt dynamic cpu backends on released binaries#1704

Merged
leejet merged 10 commits into
leejet:masterfrom
wbruna:sd_multi_cpu
Jun 26, 2026
Merged

ci: adopt dynamic cpu backends on released binaries#1704
leejet merged 10 commits into
leejet:masterfrom
wbruna:sd_multi_cpu

Conversation

@wbruna

@wbruna wbruna commented Jun 23, 2026

Copy link
Copy Markdown
Contributor

Summary

Switches build options on all released binaries to use dynamic ggml backends, enables multiple CPU variants, and replaces noavx/avx/avx2/avx512 variants with a single one.

This should fix recurrent issues where specific Github CI builders release binaries that assume AVX2 or AVX512, and enable compatibility with several CPU generations in all our builds.

Related Issue / Discussion

To be honest, I'm not used to the Github release process, so there could be bugs with the non-Linux releases, or the Docker image generation. I was only able to test the released Linux binaries; I've made a test build on my own fork some time ago (details on #1589 ).

Related to #1591 .

Should fix: #1343 , #1351 , #1483 , #1588 , #1589 , #1695 .

Additional Information

Checklist

@leejet leejet merged commit 39f7962 into leejet:master Jun 26, 2026
12 checks passed
@wbruna

wbruna commented Jun 26, 2026

Copy link
Copy Markdown
Contributor Author

https://github.com/leejet/stable-diffusion.cpp/actions/runs/28211388692/job/83573270446#step:10:1567 :

cc1: error: invalid feature modifier 'sme' in '-march=armv9.2-a+dotprod+fp16+sve+i8mm+sme'
cc1: note: valid arguments are: fp simd crc lse fp16 rcpc rdma dotprod aes sha2 crypto sha3 sm4 fp16fml sve profile rng memtag sb ssbs predres sve2 sve2-sm4 sve2-aes sve2-sha3 sve2-bitperm tme i8mm f32mm f64mm bf16 flagm pauth ls64 mops cssc; did you mean 'sm4'?
gmake[2]: *** [ggml/src/CMakeFiles/ggml-cpu-armv9.2_1.dir/build.make:76: ggml/src/CMakeFiles/ggml-cpu-armv9.2_1.dir/ggml-cpu/ggml-cpu.c.o] Error 1

@leejet , I've found this: ggml-org/llama.cpp#17403 . Looks like -DGGML_CPU_ALL_VARIANTS=ON on ARM needs either a newer gcc, or a ggml patch.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

[Bug] AVX512 false detection

2 participants