01▲MamaBearZKP: A Holistic Co-design of Prime Fields and Proving Stacks for High-Throughput ZKP on Modern CPUs eprint.iacr.org 49-bit field plus AVX-512IFMA co-design speeds HyperPlonk-DeepFold proving past Goldilocks/BabyBear baselines.avx-512ifmacpu-optimizationeprintfftplonkpolynomial-commitmentsimdsnarksum-checkzk1 pt/max_carry/2 days ago/1 comment
02▲Vectorized SQIsign Implementation Using AVX-512 arxiv.org First full AVX-512 IFMA SQIsign, hand-vectorizing the field/isogeny stack for speed the reference code never had.arxivavx-512elliptic-curvesisogenypqcsignaturessimd0 pts/dford/1 day ago/discuss
03▲Optimizing Cryptography with AI blog.zksecurity.xyz Article on using AI to optimize crypto code with correctness checks, plus a Bolt case study and the usual hardware rabbit hole.blogscryptographyformal-verificationgpuhashoptimizationpoly-commitmentproof-systemssimdsnarkzkzksecurity2 pts/nullptr7/5 days ago/1 comment
04▲Tensor Encodings for SIMD HSS eprint.iacr.org Yet another way to squeeze more SIMD into BKS HSS: tensor encodings, authenticated automorphisms, and a near-maximal pack.cryptographyeprinthomomorphic-secret-sharinghsslatticeprivacyrlwesimd0 pts/rkerr/14 days ago/3 comments