All pull requests

Pull requests

• Rewrite Poseidon's partial rounds into cheaper equivalent steps

  #3656·kpandl opened Nov 3, 2026Collaborator·5 tasks done·

  ·
  1

• Skip Elligator2's checks that cannot fail in release builds

  #3655·kpandl opened Nov 3, 2026Collaborator·6 tasks done·

  ·
  6

• Add a batched G1 MSM on BLS12-377 in AVX-512 IFMA GLV lanes

  ifmaperftx-verifier
  #3653·cbeck88 opened Nov 3, 2026Collaborator·

  ·

• Compute the BLS12-377 Miller loop in AVX-512 IFMA lanes

  ifmaperftx-verifier
  #3652·cbeck88 opened Nov 3, 2026Collaborator·3/3·

  ·

• Compute the cyclotomic exponentiation chain in AVX-512 IFMA lanes

  ifmaperftx-verifier
  #3651·cbeck88 opened Nov 3, 2026Collaborator·2/3·

  ·

• Compute the BLS12-377 Fp12 product in AVX-512 IFMA lanes

  ifmaperftx-verifier
  #3650·cbeck88 opened Nov 3, 2026Collaborator·1/3·

  ·

• remove custom logic in max_available_threads

  bug
  #3649·cbeck88 opened Nov 2, 2026Collaborator·

  ·

• Use each network's own consensus heights

  bug
  #3647·cbeck88 opened Nov 2, 2026Collaborator·

  ·

• Translate public terms in the thread-local Circuit's ConstraintSynthesizer

  bug
  #3646·cbeck88 opened Nov 2, 2026Collaborator·

  ·

• Pair the KZG batch check's adjusted witness with h in the commitments' MSM

  perftx-verifier
  #3645·cbeck88 opened Nov 2, 2026Collaborator·4/4·

  ·

• Decode Edwards points with one sliding-window scalar multiplication

  perfserialization
  #3644·cbeck88 opened Nov 2, 2026Collaborator·

  ·

• Read certificate signatures and parent IDs serially

  perfserialization
  #3640·cbeck88 opened Oct 31, 2026Collaborator·2/2·

  ·

• Cache decoded compute keys across signature reads

  perfserialization
  #3639·cbeck88 opened Oct 31, 2026Collaborator·1/2·

  ·

• Retire the testnet branch

  #3637·cbeck88 opened Oct 29, 2026Collaborator·

  ·

• Retire the mainnet branch

  #3636·cbeck88 opened Oct 29, 2026Collaborator·

  ·

• Prune old block authority data behind prune-authority-data

  #3630·eranrund opened Oct 28, 2026Collaborator·

  ·
  6

• Sum the linear combinations' commitments inside the batch check's MSM per degree bound

  perftx-verifier
  #3622·cbeck88 opened Oct 28, 2026Collaborator·3/4·

  ·

• Run the Poseidon partial rounds with sparse matrices

  perf
  #3621·cbeck88 opened Oct 27, 2026Collaborator·

  ·
  1

• Make Fp384 add_assign, sub_assign and double_in_place branchless

  perf
  #3607·cbeck88 opened Oct 24, 2026Collaborator·

  ·

• Compute the Legendre symbol in Fp256 and Fp384 with safegcd

  perfserialization
  #3605·cbeck88 opened Oct 23, 2026Collaborator·

  ·
  1

• Make sqrt tables required and skip the Legendre test in sqrt

  perfserialization
  #3604·cbeck88 opened Oct 23, 2026Collaborator·

  ·
  1

• Make the CUDA MSM threshold configurable

  #3593·peter941221 opened Oct 21, 2026·

  ·
  1

• Keep Varuna's round-four polynomials on the device through commitment and opening

  cudaperf
  #3588·cbeck88 opened Oct 20, 2026Collaborator·5/5·

  ·

• Keep SRS windows and each circuit's round-4 arrays on the device across cuda proofs

  cudaperf
  #3587·cbeck88 opened Oct 20, 2026Collaborator·4/5·

  ·

• Speed up the cuda prover: faster MSM inputs, Varuna rounds 3 and 4 and openings on the GPU

  cudaperf
  #3579·cbeck88 opened Oct 19, 2026Collaborator·3/5·

  ·
  1