x86_64: Eliminate use_hint 32/88 intrinsics by willieyz · Pull Request #940 · pq-code-package/mldsa-native

willieyz · 2026-02-03T07:41:03Z

Resolves: AVX2: Replace intrinsics implementation of poly_use_hint with assembly #484
In this PR, we replace the AVX2 intrinsics implementation of poly_use_hint_32 and poly_use_hint_88 with a x86_64 assembly version, this is part of the effort to enable HOL-Light proofs.

We also tried unrolling the loops: mld_poly_use_hint_88_avx2_loop and mld_poly_use_hint_32_avx2_loop
in both files. However, the benchmark results showed that this did not provide any performance benefit, so we decided to keep the current version.

bench components
- Δ (%) = (asm − AVX2) / AVX2 × 100

Component	Implementation	Build	ML-DSA-44	ML-DSA-65	ML-DSA-87	Notes
mld_poly_caddq (avg)	AVX2 intrinsics	no-opt	821	781	789
	x86_64 asm	no-opt	847	786	787
	Δ (%)	no-opt	+3.17%	+0.64%	-0.25%
mld_poly_caddq (avg)	AVX2 intrinsics	opt	210	147	143
	x86_64 asm	opt	220	153	155
	x86_64 asm (unroll)	opt	273	154	156	unroll by 4
	Δ (%)	opt	+4.76%	+4.08%	+8.39%
	Δ (%) (unroll)	opt	+30.00%	+4.76%	+9.09%	unroll by 4

bench
- Δ (%) = (asm − AVX2) / AVX2 × 100

Component	Implementation	Build	ML-DSA-44	ML-DSA-65	ML-DSA-87	Notes
keypair cycles (avg)	AVX2 intrinsics	no-opt	127436	218610	360739	baseline (main)
	x86_64 asm	no-opt	127459	217604	367118
	Δ (%)	no-opt	+0.02%	-0.46%	+1.77%
	AVX2 intrinsics	opt	56955	98362	157869	baseline (main)
	x86_64 asm	opt	59747	102961	165706
	x86_64 asm (unroll)	opt	59483	104732	166654
	Δ (%)	opt	+4.90%	+4.68%	+4.96%
	Δ (%) (unroll)	opt	+4.44%	+6.48%	+5.56%	unroll by 4
sign cycles (avg)	AVX2 intrinsics	no-opt	451922	756003	958151	baseline (main)
	x86_64 asm	no-opt	452833	752512	974497
	Δ (%)	no-opt	+0.20%	-0.46%	+1.71%
	AVX2 intrinsics	opt	170370	281545	347924	baseline (main)
	x86_64 asm	opt	178564	294843	362677
	x86_64 asm (unroll)	opt	177251	300667	366158
	Δ (%)	opt	+4.81%	+4.72%	+4.24%
	Δ (%) (unroll)	opt	+4.04%	+6.79%	+5.24%	unroll by 4
verify cycles (avg)	AVX2 intrinsics	no-opt	134113	220671	363234	baseline (main)
	x86_64 asm	no-opt	134633	220015	369763
	Δ (%)	no-opt	+0.39%	-0.30%	+1.80%
	AVX2 intrinsics	opt	60234	98904	156281	baseline (main)
	x86_64 asm	opt	63140	103682	164376
	x86_64 asm (unroll)	opt	62822	105719	164028
	Δ (%)	opt	+4.82%	+4.83%	+5.18%
	Δ (%) (unroll)	opt	+4.30%	+6.89%	+4.96%	unroll by 4

oqs-bot · 2026-02-03T07:58:29Z

CBMC Results (ML-DSA-87)

Full Results (179 proofs)

Proof	Status	Current	Previous	Change
`TOTAL`	✅	2655s	2394s	+10.9%
`polyvecl_pointwise_acc_montgomery_c`	✅	308s	260s	+18%
`mld_attempt_signature_generation`	✅	233s	221s	+5%
`sign_verify_internal`	✅	217s	203s	+7%
`polyvec_matrix_expand`	✅	192s	176s	+9%
`poly_pointwise_montgomery_c`	✅	182s	142s	+28%
`rej_uniform_native`	✅	156s	138s	+13%
`mld_invntt_layer`	✅	104s	92s	+13%
`mld_ct_memcmp`	✅	84s	72s	+17%
`polyvec_matrix_expand_serial`	✅	82s	80s	+2%
`polyveck_decompose`	✅	59s	55s	+7%
`mld_ntt_layer`	✅	57s	52s	+10%
`polymat_permute_bitrev_to_custom`	✅	47s	46s	+2%
`sign_signature_internal`	✅	40s	38s	+5%
`mld_compute_t0_t1_tr_from_sk_components`	✅	25s	26s	-4%
`rej_uniform`	✅	23s	25s	-8%
`fqmul`	✅	21s	18s	+17%
`poly_chknorm_c`	✅	19s	19s	+0%
`polyeta_unpack`	✅	19s	17s	+12%
`poly_uniform_4x`	✅	16s	13s	+23%
`poly_uniform_eta_4x`	✅	16s	15s	+7%
`polyt0_unpack`	✅	16s	14s	+14%
`keccakf1600x4_permute_native`	✅	15s	12s	+25%
`rej_uniform_c`	✅	14s	13s	+8%
`mld_ntt_butterfly_block`	✅	13s	11s	+18%
`mld_sample_s1_s2`	✅	13s	6s	+117%
`poly_add`	✅	13s	11s	+18%
`polyvec_matrix_pointwise_montgomery`	✅	13s	12s	+8%
`polyz_unpack_c`	✅	12s	7s	+71%
`sign_pk_from_sk`	✅	12s	7s	+71%
`keccak_absorb_once_x4`	✅	11s	12s	-8%
`polyvecl_ntt`	✅	11s	9s	+22%
`mld_polyvecl_permute_bitrev_to_custom_native`	✅	10s	14s	-29%
`polyveck_add`	✅	10s	9s	+11%
`polyveck_power2round`	✅	10s	6s	+67%
`keccak_squeezeblocks_x4`	✅	9s	8s	+12%
`keccakf1600_permute_native`	✅	9s	10s	-10%
`poly_decompose_c`	✅	9s	8s	+12%
`polyveck_pointwise_poly_montgomery_s2`	✅	9s	5s	+80%
`polyveck_reduce`	✅	9s	9s	+0%
`polyveck_sub`	✅	9s	7s	+29%
`polyveck_use_hint`	✅	9s	8s	+12%
`unpack_sk`	✅	9s	9s	+0%
`keccakf1600_permute`	✅	8s	8s	+0%
`poly_invntt_tomont_c`	✅	8s	6s	+33%
`polyveck_caddq`	✅	8s	7s	+14%
`polyveck_ntt`	✅	8s	7s	+14%
`polyveck_pointwise_poly_montgomery`	✅	8s	9s	-11%
`sign`	✅	8s	6s	+33%
`mld_check_pct`	✅	7s	9s	-22%
`mld_sample_s1_s2_serial`	✅	7s	8s	-12%
`poly_ntt_c`	✅	7s	2s	+250%
`poly_uniform_eta`	✅	7s	7s	+0%
`polyveck_pointwise_poly_montgomery_t0`	✅	7s	4s	+75%
`unpack_hints`	✅	7s	6s	+17%
`keccak_absorb`	✅	6s	6s	+0%
`pack_pk`	✅	6s	4s	+50%
`polyeta_pack`	✅	6s	2s	+200%
`polyvecl_chknorm`	✅	6s	4s	+50%
`rej_eta_native`	✅	6s	4s	+50%
`intt_native_x86_64`	✅	5s	6s	-17%
`mld_compute_pack_z`	✅	5s	7s	-29%
`mld_ct_sel_int32`	✅	5s	2s	+150%
`mld_h`	✅	5s	4s	+25%
`pack_sk`	✅	5s	4s	+25%
`poly_challenge`	✅	5s	8s	-38%
`poly_chknorm`	✅	5s	3s	+67%
`poly_power2round`	✅	5s	8s	-38%
`polyveck_shiftl`	✅	5s	7s	-29%
`polyvecl_permute_bitrev_to_custom`	✅	5s	3s	+67%
`polyvecl_uniform_gamma1`	✅	5s	2s	+150%
`polyvecl_uniform_gamma1_serial`	✅	5s	6s	-17%
`polyvecl_unpack_z`	✅	5s	4s	+25%
`polyw1_pack`	✅	5s	2s	+150%
`sign_keypair_internal`	✅	5s	6s	-17%
`sign_open`	✅	5s	5s	+0%
`sign_signature_extmu`	✅	5s	4s	+25%
`sign_signature_pre_hash_internal`	✅	5s	4s	+25%
`sign_verify`	✅	5s	4s	+25%
`caddq`	✅	4s	3s	+33%
`make_hint`	✅	4s	2s	+100%
`mld_ct_cmask_nonzero_u32`	✅	4s	2s	+100%
`mld_prepare_domain_separation_prefix`	✅	4s	4s	+0%
`ntt_native_x86_64`	✅	4s	2s	+100%
`poly_caddq_c`	✅	4s	5s	-20%
`poly_caddq_native`	✅	4s	2s	+100%
`poly_chknorm_native_aarch64`	✅	4s	1s	+300%
`poly_invntt_tomont`	✅	4s	2s	+100%
`poly_reduce`	✅	4s	4s	+0%
`poly_shiftl`	✅	4s	3s	+33%
`poly_sub`	✅	4s	3s	+33%
`poly_uniform`	✅	4s	3s	+33%
`poly_uniform_gamma1_4x`	✅	4s	3s	+33%
`poly_use_hint_c`	✅	4s	3s	+33%
`poly_use_hint_native`	✅	4s	3s	+33%
`polyt0_pack`	✅	4s	2s	+100%
`polyt1_unpack`	✅	4s	2s	+100%
`polyveck_chknorm`	✅	4s	6s	-33%
`polyveck_invntt_tomont`	✅	4s	4s	+0%
`polyveck_make_hint`	✅	4s	5s	-20%
`polyveck_pack_eta`	✅	4s	2s	+100%
`polyveck_unpack_eta`	✅	4s	4s	+0%
`polyvecl_pointwise_acc_montgomery_native`	✅	4s	1s	+300%
`polyz_unpack`	✅	4s	2s	+100%
`polyz_unpack_native`	✅	4s	3s	+33%
`power2round`	✅	4s	3s	+33%
`rej_eta_c`	✅	4s	3s	+33%
`shake128_init`	✅	4s	4s	+0%
`shake256_init`	✅	4s	4s	+0%
`sign_signature`	✅	4s	5s	-20%
`sign_verify_extmu`	✅	4s	3s	+33%
`sign_verify_pre_hash_shake256`	✅	4s	4s	+0%
`sys_check_capability`	✅	4s	4s	+0%
`unpack_sig`	✅	4s	3s	+33%
`use_hint`	✅	4s	4s	+0%
`decompose`	✅	3s	2s	+50%
`fqscale`	✅	3s	4s	-25%
`keccakf1600x4_extract_bytes`	✅	3s	1s	+200%
`keccakf1600x4_xor_bytes`	✅	3s	3s	+0%
`mld_ct_cmask_nonzero_u8`	✅	3s	3s	+0%
`mld_ct_get_optblocker_u32`	✅	3s	1s	+200%
`mld_ct_get_optblocker_u8`	✅	3s	3s	+0%
`pack_sig_z`	✅	3s	5s	-40%
`poly_caddq`	✅	3s	4s	-25%
`poly_caddq_native_aarch64`	✅	3s	4s	-25%
`poly_chknorm_native`	✅	3s	3s	+0%
`poly_decompose`	✅	3s	2s	+50%
`poly_decompose_native`	✅	3s	2s	+50%
`poly_ntt`	✅	3s	4s	-25%
`poly_ntt_native`	✅	3s	2s	+50%
`poly_pointwise_montgomery`	✅	3s	3s	+0%
`poly_pointwise_montgomery_native`	✅	3s	2s	+50%
`polyt1_pack`	✅	3s	3s	+0%
`polyveck_pack_t0`	✅	3s	3s	+0%
`polyveck_pack_w1`	✅	3s	2s	+50%
`polyveck_unpack_t0`	✅	3s	2s	+50%
`polyvecl_pointwise_acc_montgomery`	✅	3s	2s	+50%
`polyvecl_unpack_eta`	✅	3s	2s	+50%
`shake128x4_absorb_once`	✅	3s	2s	+50%
`shake256_release`	✅	3s	1s	+200%
`shake256x4_absorb_once`	✅	3s	2s	+50%
`shake256x4_squeezeblocks`	✅	3s	3s	+0%
`sign_keypair`	✅	3s	3s	+0%
`sign_verify_pre_hash_internal`	✅	3s	4s	-25%
`keccak_finalize`	✅	2s	3s	-33%
`keccak_init`	✅	2s	3s	-33%
`keccakf1600_extract_bytes (big endian)`	✅	2s	2s	+0%
`keccakf1600x4_permute`	✅	2s	1s	+100%
`mld_ct_abs_i32`	✅	2s	2s	+0%
`mld_value_barrier_i64`	✅	2s	1s	+100%
`mld_value_barrier_u32`	✅	2s	1s	+100%
`mld_value_barrier_u8`	✅	2s	2s	+0%
`montgomery_reduce`	✅	2s	2s	+0%
`ntt_native_aarch64`	✅	2s	4s	-50%
`pack_sig_c_h`	✅	2s	4s	-50%
`poly_invntt_tomont_native`	✅	2s	9s	-78%
`poly_uniform_gamma1`	✅	2s	2s	+0%
`poly_use_hint`	✅	2s	3s	-33%
`polyvecl_pack_eta`	✅	2s	5s	-60%
`polyz_pack`	✅	2s	4s	-50%
`reduce32`	✅	2s	2s	+0%
`rej_eta`	✅	2s	2s	+0%
`shake128_absorb`	✅	2s	4s	-50%
`shake128_release`	✅	2s	3s	-33%
`shake128_squeeze`	✅	2s	1s	+100%
`shake128x4_squeezeblocks`	✅	2s	3s	-33%
`shake256`	✅	2s	2s	+0%
`shake256_absorb`	✅	2s	2s	+0%
`shake256_finalize`	✅	2s	4s	-50%
`shake256_squeeze`	✅	2s	2s	+0%
`sign_signature_pre_hash_shake256`	✅	2s	3s	-33%
`unpack_pk`	✅	2s	4s	-50%
`keccak_squeeze`	✅	1s	3s	-67%
`keccakf1600_xor_bytes`	✅	1s	3s	-67%
`keccakf1600_xor_bytes (big endian)`	✅	1s	2s	-50%
`mld_ct_cmask_neg_i32`	✅	1s	3s	-67%
`mld_ct_get_optblocker_i64`	✅	1s	2s	-50%
`mld_keccakf1600_extract_bytes`	✅	1s	2s	-50%
`poly_make_hint`	✅	1s	3s	-67%
`shake128_finalize`	✅	1s	3s	-67%

oqs-bot · 2026-02-03T07:59:14Z

CBMC Results (ML-DSA-44)

Full Results (179 proofs)

Proof	Status	Current	Previous	Change
`TOTAL`	✅	1995s	2009s	-0.7%
`mld_attempt_signature_generation`	✅	245s	251s	-2%
`polyvecl_pointwise_acc_montgomery_c`	✅	202s	204s	-1%
`sign_verify_internal`	✅	182s	182s	+0%
`poly_pointwise_montgomery_c`	✅	146s	157s	-7%
`rej_uniform_native`	✅	139s	144s	-3%
`mld_invntt_layer`	✅	84s	86s	-2%
`mld_ct_memcmp`	✅	71s	75s	-5%
`mld_ntt_layer`	✅	53s	53s	+0%
`poly_chknorm_c`	✅	21s	19s	+11%
`rej_uniform`	✅	21s	23s	-9%
`sign_signature_internal`	✅	21s	22s	-5%
`fqmul`	✅	20s	18s	+11%
`polyvec_matrix_expand`	✅	20s	19s	+5%
`poly_uniform_eta_4x`	✅	18s	18s	+0%
`poly_uniform_4x`	✅	16s	16s	+0%
`polyeta_unpack`	✅	16s	18s	-11%
`rej_uniform_c`	✅	14s	14s	+0%
`polymat_permute_bitrev_to_custom`	✅	13s	13s	+0%
`polyt0_unpack`	✅	13s	13s	+0%
`polyvec_matrix_pointwise_montgomery`	✅	13s	13s	+0%
`polyveck_power2round`	✅	13s	14s	-7%
`keccak_absorb_once_x4`	✅	12s	10s	+20%
`keccakf1600x4_permute_native`	✅	12s	13s	-8%
`mld_compute_t0_t1_tr_from_sk_components`	✅	12s	17s	-29%
`polyz_unpack_c`	✅	12s	12s	+0%
`mld_ntt_butterfly_block`	✅	11s	12s	-8%
`mld_polyvecl_permute_bitrev_to_custom_native`	✅	11s	8s	+38%
`poly_add`	✅	11s	10s	+10%
`polyvec_matrix_expand_serial`	✅	11s	13s	-15%
`unpack_sk`	✅	9s	6s	+50%
`keccakf1600_permute`	✅	8s	9s	-11%
`keccakf1600_permute_native`	✅	8s	7s	+14%
`polyveck_add`	✅	8s	7s	+14%
`polyveck_decompose`	✅	8s	5s	+60%
`polyveck_use_hint`	✅	8s	6s	+33%
`sign`	✅	8s	8s	+0%
`mld_prepare_domain_separation_prefix`	✅	7s	5s	+40%
`polyveck_chknorm`	✅	7s	6s	+17%
`sign_verify_pre_hash_shake256`	✅	7s	4s	+75%
`keccak_absorb`	✅	6s	7s	-14%
`keccak_squeezeblocks_x4`	✅	6s	5s	+20%
`mld_check_pct`	✅	6s	7s	-14%
`mld_compute_pack_z`	✅	6s	7s	-14%
`poly_invntt_tomont_c`	✅	6s	5s	+20%
`poly_ntt_native`	✅	6s	4s	+50%
`poly_power2round`	✅	6s	4s	+50%
`polyveck_caddq`	✅	6s	3s	+100%
`polyveck_invntt_tomont`	✅	6s	6s	+0%
`sign_open`	✅	6s	4s	+50%
`sign_signature_pre_hash_internal`	✅	6s	5s	+20%
`sign_verify`	✅	6s	4s	+50%
`decompose`	✅	5s	3s	+67%
`mld_h`	✅	5s	2s	+150%
`mld_sample_s1_s2`	✅	5s	5s	+0%
`mld_sample_s1_s2_serial`	✅	5s	7s	-29%
`poly_caddq_c`	✅	5s	5s	+0%
`poly_use_hint_c`	✅	5s	5s	+0%
`polyveck_make_hint`	✅	5s	5s	+0%
`polyveck_pointwise_poly_montgomery_t0`	✅	5s	5s	+0%
`polyveck_shiftl`	✅	5s	6s	-17%
`polyveck_sub`	✅	5s	5s	+0%
`polyvecl_uniform_gamma1_serial`	✅	5s	2s	+150%
`polyvecl_unpack_z`	✅	5s	3s	+67%
`power2round`	✅	5s	2s	+150%
`rej_eta_native`	✅	5s	6s	-17%
`sign_pk_from_sk`	✅	5s	10s	-50%
`sign_verify_extmu`	✅	5s	3s	+67%
`unpack_hints`	✅	5s	6s	-17%
`use_hint`	✅	5s	3s	+67%
`intt_native_x86_64`	✅	4s	2s	+100%
`keccakf1600_extract_bytes (big endian)`	✅	4s	2s	+100%
`keccakf1600_xor_bytes (big endian)`	✅	4s	3s	+33%
`mld_ct_get_optblocker_u8`	✅	4s	3s	+33%
`mld_value_barrier_u32`	✅	4s	2s	+100%
`montgomery_reduce`	✅	4s	5s	-20%
`pack_sig_z`	✅	4s	4s	+0%
`poly_caddq_native`	✅	4s	2s	+100%
`poly_challenge`	✅	4s	4s	+0%
`poly_chknorm_native_aarch64`	✅	4s	7s	-43%
`poly_decompose`	✅	4s	3s	+33%
`poly_decompose_c`	✅	4s	4s	+0%
`poly_ntt`	✅	4s	4s	+0%
`poly_pointwise_montgomery`	✅	4s	3s	+33%
`poly_pointwise_montgomery_native`	✅	4s	2s	+100%
`poly_shiftl`	✅	4s	5s	-20%
`poly_uniform_eta`	✅	4s	5s	-20%
`poly_uniform_gamma1`	✅	4s	3s	+33%
`poly_use_hint_native`	✅	4s	3s	+33%
`polyt0_pack`	✅	4s	5s	-20%
`polyveck_ntt`	✅	4s	4s	+0%
`polyveck_pack_eta`	✅	4s	4s	+0%
`polyveck_pack_t0`	✅	4s	2s	+100%
`polyveck_pointwise_poly_montgomery_s2`	✅	4s	4s	+0%
`polyveck_reduce`	✅	4s	5s	-20%
`polyveck_unpack_eta`	✅	4s	2s	+100%
`polyveck_unpack_t0`	✅	4s	4s	+0%
`polyvecl_ntt`	✅	4s	3s	+33%
`polyvecl_pointwise_acc_montgomery`	✅	4s	2s	+100%
`polyw1_pack`	✅	4s	1s	+300%
`polyz_unpack`	✅	4s	2s	+100%
`polyz_unpack_native`	✅	4s	2s	+100%
`reduce32`	✅	4s	4s	+0%
`shake256_absorb`	✅	4s	2s	+100%
`shake256x4_squeezeblocks`	✅	4s	4s	+0%
`sign_keypair_internal`	✅	4s	7s	-43%
`sign_signature`	✅	4s	5s	-20%
`unpack_sig`	✅	4s	3s	+33%
`fqscale`	✅	3s	3s	+0%
`keccak_finalize`	✅	3s	3s	+0%
`keccak_init`	✅	3s	2s	+50%
`keccakf1600x4_extract_bytes`	✅	3s	2s	+50%
`keccakf1600x4_permute`	✅	3s	3s	+0%
`mld_ct_abs_i32`	✅	3s	3s	+0%
`mld_ct_cmask_nonzero_u32`	✅	3s	4s	-25%
`mld_ct_cmask_nonzero_u8`	✅	3s	2s	+50%
`mld_keccakf1600_extract_bytes`	✅	3s	2s	+50%
`ntt_native_x86_64`	✅	3s	3s	+0%
`pack_sk`	✅	3s	5s	-40%
`poly_caddq`	✅	3s	3s	+0%
`poly_caddq_native_aarch64`	✅	3s	2s	+50%
`poly_decompose_native`	✅	3s	6s	-50%
`poly_make_hint`	✅	3s	4s	-25%
`poly_ntt_c`	✅	3s	3s	+0%
`poly_sub`	✅	3s	3s	+0%
`poly_uniform`	✅	3s	6s	-50%
`poly_uniform_gamma1_4x`	✅	3s	5s	-40%
`poly_use_hint`	✅	3s	2s	+50%
`polyt1_pack`	✅	3s	2s	+50%
`polyt1_unpack`	✅	3s	3s	+0%
`polyveck_pointwise_poly_montgomery`	✅	3s	4s	-25%
`polyvecl_chknorm`	✅	3s	3s	+0%
`polyvecl_pack_eta`	✅	3s	3s	+0%
`polyvecl_pointwise_acc_montgomery_native`	✅	3s	3s	+0%
`polyvecl_unpack_eta`	✅	3s	2s	+50%
`polyz_pack`	✅	3s	3s	+0%
`rej_eta`	✅	3s	1s	+200%
`shake128x4_absorb_once`	✅	3s	2s	+50%
`shake128x4_squeezeblocks`	✅	3s	3s	+0%
`shake256`	✅	3s	3s	+0%
`shake256_finalize`	✅	3s	2s	+50%
`sign_keypair`	✅	3s	3s	+0%
`sign_signature_extmu`	✅	3s	3s	+0%
`sign_signature_pre_hash_shake256`	✅	3s	3s	+0%
`sign_verify_pre_hash_internal`	✅	3s	2s	+50%
`sys_check_capability`	✅	3s	3s	+0%
`unpack_pk`	✅	3s	3s	+0%
`keccakf1600_xor_bytes`	✅	2s	3s	-33%
`keccakf1600x4_xor_bytes`	✅	2s	1s	+100%
`make_hint`	✅	2s	3s	-33%
`mld_ct_cmask_neg_i32`	✅	2s	3s	-33%
`mld_ct_get_optblocker_i64`	✅	2s	2s	+0%
`mld_ct_sel_int32`	✅	2s	2s	+0%
`ntt_native_aarch64`	✅	2s	3s	-33%
`pack_pk`	✅	2s	3s	-33%
`pack_sig_c_h`	✅	2s	3s	-33%
`poly_chknorm_native`	✅	2s	3s	-33%
`poly_invntt_tomont_native`	✅	2s	2s	+0%
`polyveck_pack_w1`	✅	2s	2s	+0%
`polyvecl_permute_bitrev_to_custom`	✅	2s	2s	+0%
`polyvecl_uniform_gamma1`	✅	2s	2s	+0%
`rej_eta_c`	✅	2s	4s	-50%
`shake128_absorb`	✅	2s	3s	-33%
`shake128_init`	✅	2s	3s	-33%
`shake128_release`	✅	2s	2s	+0%
`shake128_squeeze`	✅	2s	4s	-50%
`shake256_init`	✅	2s	1s	+100%
`caddq`	✅	1s	3s	-67%
`keccak_squeeze`	✅	1s	3s	-67%
`mld_ct_get_optblocker_u32`	✅	1s	3s	-67%
`mld_value_barrier_i64`	✅	1s	2s	-50%
`mld_value_barrier_u8`	✅	1s	3s	-67%
`poly_chknorm`	✅	1s	5s	-80%
`poly_invntt_tomont`	✅	1s	3s	-67%
`poly_reduce`	✅	1s	3s	-67%
`polyeta_pack`	✅	1s	3s	-67%
`shake128_finalize`	✅	1s	2s	-50%
`shake256_release`	✅	1s	2s	-50%
`shake256_squeeze`	✅	1s	2s	-50%
`shake256x4_absorb_once`	✅	1s	2s	-50%

oqs-bot · 2026-02-03T08:00:03Z

CBMC Results (ML-DSA-65)

Full Results (179 proofs)

Proof	Status	Current	Previous	Change
`TOTAL`	✅	2467s	2424s	+1.8%
`sign_verify_internal`	✅	334s	330s	+1%
`mld_attempt_signature_generation`	✅	284s	277s	+3%
`polyvecl_pointwise_acc_montgomery_c`	✅	187s	188s	-1%
`poly_pointwise_montgomery_c`	✅	168s	157s	+7%
`rej_uniform_native`	✅	149s	140s	+6%
`polyvec_matrix_expand`	✅	121s	123s	-2%
`mld_invntt_layer`	✅	98s	94s	+4%
`mld_ct_memcmp`	✅	82s	73s	+12%
`polyvec_matrix_expand_serial`	✅	68s	66s	+3%
`mld_ntt_layer`	✅	56s	53s	+6%
`polymat_permute_bitrev_to_custom`	✅	30s	29s	+3%
`sign_signature_internal`	✅	28s	30s	-7%
`mld_compute_t0_t1_tr_from_sk_components`	✅	27s	26s	+4%
`fqmul`	✅	24s	19s	+26%
`poly_chknorm_c`	✅	20s	19s	+5%
`rej_uniform`	✅	19s	21s	-10%
`rej_uniform_c`	✅	17s	16s	+6%
`poly_uniform_eta_4x`	✅	16s	19s	-16%
`poly_uniform_4x`	✅	15s	13s	+15%
`keccakf1600x4_permute_native`	✅	13s	12s	+8%
`poly_add`	✅	13s	11s	+18%
`polyt0_unpack`	✅	13s	14s	-7%
`polyveck_decompose`	✅	13s	12s	+8%
`mld_check_pct`	✅	12s	11s	+9%
`mld_ntt_butterfly_block`	✅	12s	12s	+0%
`polyvec_matrix_pointwise_montgomery`	✅	12s	10s	+20%
`polyvecl_chknorm`	✅	12s	11s	+9%
`polyveck_sub`	✅	11s	13s	-15%
`keccak_absorb_once_x4`	✅	10s	10s	+0%
`keccakf1600_permute_native`	✅	10s	8s	+25%
`polyveck_add`	✅	10s	11s	-9%
`polyveck_caddq`	✅	10s	8s	+25%
`polyveck_power2round`	✅	10s	9s	+11%
`keccakf1600_permute`	✅	9s	8s	+12%
`mld_compute_pack_z`	✅	9s	6s	+50%
`poly_decompose_native`	✅	9s	6s	+50%
`polyveck_ntt`	✅	9s	9s	+0%
`poly_invntt_tomont_c`	✅	8s	6s	+33%
`polyveck_pointwise_poly_montgomery`	✅	8s	10s	-20%
`polyveck_shiftl`	✅	8s	7s	+14%
`polyveck_use_hint`	✅	8s	8s	+0%
`polyvecl_ntt`	✅	8s	7s	+14%
`unpack_sk`	✅	8s	7s	+14%
`keccak_absorb`	✅	7s	8s	-12%
`keccak_squeezeblocks_x4`	✅	7s	8s	-12%
`poly_challenge`	✅	7s	4s	+75%
`poly_decompose_c`	✅	7s	7s	+0%
`polyeta_unpack`	✅	7s	8s	-12%
`polyveck_invntt_tomont`	✅	7s	6s	+17%
`sign_pk_from_sk`	✅	7s	10s	-30%
`sign_verify_pre_hash_shake256`	✅	7s	5s	+40%
`mld_polyvecl_permute_bitrev_to_custom_native`	✅	6s	8s	-25%
`polyveck_pack_t0`	✅	6s	2s	+200%
`polyveck_pointwise_poly_montgomery_s2`	✅	6s	9s	-33%
`polyveck_pointwise_poly_montgomery_t0`	✅	6s	5s	+20%
`polyveck_reduce`	✅	6s	7s	-14%
`polyz_unpack`	✅	6s	2s	+200%
`mld_h`	✅	5s	5s	+0%
`mld_sample_s1_s2_serial`	✅	5s	4s	+25%
`poly_caddq_c`	✅	5s	4s	+25%
`poly_pointwise_montgomery_native`	✅	5s	6s	-17%
`poly_power2round`	✅	5s	4s	+25%
`poly_uniform_gamma1_4x`	✅	5s	5s	+0%
`poly_use_hint_c`	✅	5s	3s	+67%
`polyveck_chknorm`	✅	5s	6s	-17%
`polyveck_make_hint`	✅	5s	4s	+25%
`polyveck_unpack_t0`	✅	5s	3s	+67%
`polyvecl_pointwise_acc_montgomery_native`	✅	5s	6s	-17%
`rej_eta_c`	✅	5s	3s	+67%
`sign`	✅	5s	7s	-29%
`sign_signature`	✅	5s	2s	+150%
`sign_signature_pre_hash_shake256`	✅	5s	6s	-17%
`keccak_squeeze`	✅	4s	4s	+0%
`keccakf1600_extract_bytes (big endian)`	✅	4s	2s	+100%
`keccakf1600x4_xor_bytes`	✅	4s	2s	+100%
`mld_ct_abs_i32`	✅	4s	2s	+100%
`ntt_native_x86_64`	✅	4s	4s	+0%
`pack_sig_c_h`	✅	4s	2s	+100%
`poly_chknorm`	✅	4s	3s	+33%
`poly_invntt_tomont`	✅	4s	6s	-33%
`poly_invntt_tomont_native`	✅	4s	4s	+0%
`poly_pointwise_montgomery`	✅	4s	2s	+100%
`poly_uniform_eta`	✅	4s	6s	-33%
`polyeta_pack`	✅	4s	3s	+33%
`polyt0_pack`	✅	4s	5s	-20%
`polyt1_pack`	✅	4s	3s	+33%
`polyveck_pack_w1`	✅	4s	2s	+100%
`polyz_pack`	✅	4s	3s	+33%
`polyz_unpack_c`	✅	4s	4s	+0%
`rej_eta`	✅	4s	4s	+0%
`rej_eta_native`	✅	4s	2s	+100%
`shake256_absorb`	✅	4s	2s	+100%
`sign_keypair_internal`	✅	4s	7s	-43%
`sign_open`	✅	4s	4s	+0%
`sign_signature_pre_hash_internal`	✅	4s	4s	+0%
`sign_verify_extmu`	✅	4s	3s	+33%
`unpack_hints`	✅	4s	6s	-33%
`unpack_pk`	✅	4s	4s	+0%
`decompose`	✅	3s	3s	+0%
`intt_native_x86_64`	✅	3s	3s	+0%
`keccak_init`	✅	3s	2s	+50%
`keccakf1600_xor_bytes (big endian)`	✅	3s	3s	+0%
`keccakf1600x4_extract_bytes`	✅	3s	1s	+200%
`keccakf1600x4_permute`	✅	3s	5s	-40%
`mld_ct_cmask_nonzero_u32`	✅	3s	1s	+200%
`mld_ct_cmask_nonzero_u8`	✅	3s	5s	-40%
`mld_ct_get_optblocker_i64`	✅	3s	3s	+0%
`mld_prepare_domain_separation_prefix`	✅	3s	4s	-25%
`mld_sample_s1_s2`	✅	3s	8s	-62%
`mld_value_barrier_i64`	✅	3s	1s	+200%
`mld_value_barrier_u32`	✅	3s	1s	+200%
`mld_value_barrier_u8`	✅	3s	2s	+50%
`montgomery_reduce`	✅	3s	3s	+0%
`ntt_native_aarch64`	✅	3s	3s	+0%
`pack_pk`	✅	3s	4s	-25%
`pack_sig_z`	✅	3s	5s	-40%
`poly_caddq_native_aarch64`	✅	3s	5s	-40%
`poly_chknorm_native`	✅	3s	2s	+50%
`poly_chknorm_native_aarch64`	✅	3s	4s	-25%
`poly_decompose`	✅	3s	1s	+200%
`poly_make_hint`	✅	3s	4s	-25%
`poly_ntt_native`	✅	3s	3s	+0%
`poly_reduce`	✅	3s	2s	+50%
`poly_shiftl`	✅	3s	2s	+50%
`poly_sub`	✅	3s	2s	+50%
`poly_uniform`	✅	3s	6s	-50%
`poly_uniform_gamma1`	✅	3s	3s	+0%
`poly_use_hint`	✅	3s	3s	+0%
`poly_use_hint_native`	✅	3s	5s	-40%
`polyveck_pack_eta`	✅	3s	3s	+0%
`polyveck_unpack_eta`	✅	3s	2s	+50%
`polyvecl_pack_eta`	✅	3s	3s	+0%
`polyvecl_permute_bitrev_to_custom`	✅	3s	4s	-25%
`polyvecl_pointwise_acc_montgomery`	✅	3s	2s	+50%
`polyvecl_uniform_gamma1_serial`	✅	3s	3s	+0%
`polyvecl_unpack_eta`	✅	3s	2s	+50%
`power2round`	✅	3s	3s	+0%
`shake128_release`	✅	3s	1s	+200%
`shake128x4_absorb_once`	✅	3s	2s	+50%
`shake128x4_squeezeblocks`	✅	3s	4s	-25%
`shake256`	✅	3s	2s	+50%
`shake256x4_absorb_once`	✅	3s	8s	-62%
`shake256x4_squeezeblocks`	✅	3s	1s	+200%
`sign_signature_extmu`	✅	3s	6s	-50%
`sign_verify`	✅	3s	2s	+50%
`sign_verify_pre_hash_internal`	✅	3s	6s	-50%
`sys_check_capability`	✅	3s	3s	+0%
`unpack_sig`	✅	3s	2s	+50%
`caddq`	✅	2s	5s	-60%
`fqscale`	✅	2s	3s	-33%
`keccak_finalize`	✅	2s	3s	-33%
`keccakf1600_xor_bytes`	✅	2s	1s	+100%
`make_hint`	✅	2s	3s	-33%
`mld_ct_cmask_neg_i32`	✅	2s	6s	-67%
`mld_ct_get_optblocker_u32`	✅	2s	2s	+0%
`mld_ct_sel_int32`	✅	2s	3s	-33%
`mld_keccakf1600_extract_bytes`	✅	2s	2s	+0%
`pack_sk`	✅	2s	3s	-33%
`poly_caddq`	✅	2s	3s	-33%
`poly_caddq_native`	✅	2s	5s	-60%
`poly_ntt`	✅	2s	4s	-50%
`poly_ntt_c`	✅	2s	5s	-60%
`polyt1_unpack`	✅	2s	4s	-50%
`polyvecl_uniform_gamma1`	✅	2s	3s	-33%
`polyvecl_unpack_z`	✅	2s	3s	-33%
`polyw1_pack`	✅	2s	5s	-60%
`polyz_unpack_native`	✅	2s	3s	-33%
`reduce32`	✅	2s	3s	-33%
`shake128_finalize`	✅	2s	3s	-33%
`shake128_init`	✅	2s	2s	+0%
`shake128_squeeze`	✅	2s	2s	+0%
`shake256_init`	✅	2s	5s	-60%
`shake256_release`	✅	2s	1s	+100%
`sign_keypair`	✅	2s	5s	-60%
`use_hint`	✅	2s	3s	-33%
`mld_ct_get_optblocker_u8`	✅	1s	1s	+0%
`shake128_absorb`	✅	1s	2s	-50%
`shake256_finalize`	✅	1s	2s	-50%
`shake256_squeeze`	✅	1s	2s	-50%

github-actions

Mac Mini (M1, 2020) benchmarks (opt)

Details

Benchmark suite	Current: `8a19e9a`	Previous: `41da557`	Ratio
`ML-DSA-44 keypair`	`46205` cycles	`46203` cycles	`1.00`
`ML-DSA-44 sign`	`131278` cycles	`131278` cycles	`1`
`ML-DSA-44 verify`	`47765` cycles	`47768` cycles	`1.00`
`ML-DSA-65 keypair`	`81014` cycles	`81024` cycles	`1.00`
`ML-DSA-65 sign`	`215785` cycles	`215787` cycles	`1.00`
`ML-DSA-65 verify`	`80057` cycles	`80052` cycles	`1.00`
`ML-DSA-87 keypair`	`132158` cycles	`132151` cycles	`1.00`
`ML-DSA-87 sign`	`276862` cycles	`276816` cycles	`1.00`
`ML-DSA-87 verify`	`130418` cycles	`130384` cycles	`1.00`

This comment was automatically generated by workflow using github-action-benchmark.

github-actions

Mac Mini (M1, 2020) benchmarks (no-opt)

Details

Benchmark suite	Current: `8a19e9a`	Previous: `41da557`	Ratio
`ML-DSA-44 keypair`	`114213` cycles	`114155` cycles	`1.00`
`ML-DSA-44 sign`	`418158` cycles	`417994` cycles	`1.00`
`ML-DSA-44 verify`	`122319` cycles	`122262` cycles	`1.00`
`ML-DSA-65 keypair`	`195508` cycles	`195499` cycles	`1.00`
`ML-DSA-65 sign`	`682497` cycles	`682470` cycles	`1.00`
`ML-DSA-65 verify`	`197760` cycles	`197741` cycles	`1.00`
`ML-DSA-87 keypair`	`322642` cycles	`322656` cycles	`1.00`
`ML-DSA-87 sign`	`864585` cycles	`864584` cycles	`1.00`
`ML-DSA-87 verify`	`328628` cycles	`328653` cycles	`1.00`

This comment was automatically generated by workflow using github-action-benchmark.

oqs-bot

Intel Xeon 4th gen (c7i)

Details

Benchmark suite	Current: `1c31329`	Previous: `0f8b8e0`	Ratio
`ML-DSA-44 keypair`	`34909` cycles	`34607` cycles	`1.01`
`ML-DSA-44 sign`	`120375` cycles	`120704` cycles	`1.00`
`ML-DSA-44 verify`	`38205` cycles	`38101` cycles	`1.00`
`ML-DSA-65 keypair`	`60968` cycles	`61787` cycles	`0.99`
`ML-DSA-65 sign`	`202493` cycles	`204750` cycles	`0.99`
`ML-DSA-65 verify`	`62726` cycles	`62947` cycles	`1.00`
`ML-DSA-87 keypair`	`94450` cycles	`94143` cycles	`1.00`
`ML-DSA-87 sign`	`241633` cycles	`240274` cycles	`1.01`
`ML-DSA-87 verify`	`96451` cycles	`95109` cycles	`1.01`

This comment was automatically generated by workflow using github-action-benchmark.

oqs-bot

Intel Xeon 4th gen (c7i) (no-opt)

Details

Benchmark suite	Current: `1c31329`	Previous: `0f8b8e0`	Ratio
`ML-DSA-44 keypair`	`94555` cycles	`94592` cycles	`1.00`
`ML-DSA-44 sign`	`333735` cycles	`333857` cycles	`1.00`
`ML-DSA-44 verify`	`99826` cycles	`99864` cycles	`1.00`
`ML-DSA-65 keypair`	`159716` cycles	`159928` cycles	`1.00`
`ML-DSA-65 sign`	`544638` cycles	`544846` cycles	`1.00`
`ML-DSA-65 verify`	`160752` cycles	`160968` cycles	`1.00`
`ML-DSA-87 keypair`	`267459` cycles	`267912` cycles	`1.00`
`ML-DSA-87 sign`	`709420` cycles	`709152` cycles	`1.00`
`ML-DSA-87 verify`	`270024` cycles	`270923` cycles	`1.00`

This comment was automatically generated by workflow using github-action-benchmark.

github-actions

Arm Cortex-A55 (Snapdragon 888) benchmarks (opt)

Details

Benchmark suite	Current: `8a19e9a`	Previous: `41da557`	Ratio
`ML-DSA-44 keypair`	`276468` cycles	`277102` cycles	`1.00`
`ML-DSA-44 sign`	`818650` cycles	`810656` cycles	`1.01`
`ML-DSA-44 verify`	`276672` cycles	`278882` cycles	`0.99`
`ML-DSA-65 keypair`	`475323` cycles	`478906` cycles	`0.99`
`ML-DSA-65 sign`	`1367640` cycles	`1360800` cycles	`1.01`
`ML-DSA-65 verify`	`459822` cycles	`466415` cycles	`0.99`
`ML-DSA-87 keypair`	`825623` cycles	`818822` cycles	`1.01`
`ML-DSA-87 sign`	`1873209` cycles	`1878770` cycles	`1.00`
`ML-DSA-87 verify`	`800938` cycles	`794467` cycles	`1.01`

This comment was automatically generated by workflow using github-action-benchmark.

oqs-bot

AMD EPYC 3rd gen (c6a)

Details

Benchmark suite	Current: `1c31329`	Previous: `0f8b8e0`	Ratio
`ML-DSA-44 keypair`	`69188` cycles	`69341` cycles	`1.00`
`ML-DSA-44 sign`	`188711` cycles	`188628` cycles	`1.00`
`ML-DSA-44 verify`	`69609` cycles	`69167` cycles	`1.01`
`ML-DSA-65 keypair`	`119110` cycles	`119048` cycles	`1.00`
`ML-DSA-65 sign`	`301012` cycles	`300972` cycles	`1.00`
`ML-DSA-65 verify`	`115433` cycles	`115129` cycles	`1.00`
`ML-DSA-87 keypair`	`202783` cycles	`202705` cycles	`1.00`
`ML-DSA-87 sign`	`393591` cycles	`393401` cycles	`1.00`
`ML-DSA-87 verify`	`194240` cycles	`194477` cycles	`1.00`

This comment was automatically generated by workflow using github-action-benchmark.

oqs-bot

Intel Xeon 3rd gen (c6i)

Details

Benchmark suite	Current: `1c31329`	Previous: `0f8b8e0`	Ratio
`ML-DSA-44 keypair`	`56699` cycles	`57040` cycles	`0.99`
`ML-DSA-44 sign`	`182012` cycles	`183077` cycles	`0.99`
`ML-DSA-44 verify`	`61080` cycles	`61515` cycles	`0.99`
`ML-DSA-65 keypair`	`99136` cycles	`98855` cycles	`1.00`
`ML-DSA-65 sign`	`302786` cycles	`300890` cycles	`1.01`
`ML-DSA-65 verify`	`101006` cycles	`100170` cycles	`1.01`
`ML-DSA-87 keypair`	`154691` cycles	`153387` cycles	`1.01`
`ML-DSA-87 sign`	`357607` cycles	`356600` cycles	`1.00`
`ML-DSA-87 verify`	`155516` cycles	`153458` cycles	`1.01`

This comment was automatically generated by workflow using github-action-benchmark.

oqs-bot

Graviton4

Details

Benchmark suite	Current: `1c31329`	Previous: `0f8b8e0`	Ratio
`ML-DSA-44 keypair`	`68195` cycles	`68176` cycles	`1.00`
`ML-DSA-44 sign`	`203778` cycles	`203661` cycles	`1.00`
`ML-DSA-44 verify`	`70887` cycles	`70749` cycles	`1.00`
`ML-DSA-65 keypair`	`120728` cycles	`120835` cycles	`1.00`
`ML-DSA-65 sign`	`334605` cycles	`334759` cycles	`1.00`
`ML-DSA-65 verify`	`117912` cycles	`118016` cycles	`1.00`
`ML-DSA-87 keypair`	`198206` cycles	`198256` cycles	`1.00`
`ML-DSA-87 sign`	`431061` cycles	`431078` cycles	`1.00`
`ML-DSA-87 verify`	`194668` cycles	`194587` cycles	`1.00`

This comment was automatically generated by workflow using github-action-benchmark.

oqs-bot

AMD EPYC 3rd gen (c6a) (no-opt)

Details

Benchmark suite	Current: `1c31329`	Previous: `0f8b8e0`	Ratio
`ML-DSA-44 keypair`	`135143` cycles	`136158` cycles	`0.99`
`ML-DSA-44 sign`	`526833` cycles	`531116` cycles	`0.99`
`ML-DSA-44 verify`	`147425` cycles	`148648` cycles	`0.99`
`ML-DSA-65 keypair`	`226627` cycles	`226842` cycles	`1.00`
`ML-DSA-65 sign`	`859933` cycles	`861270` cycles	`1.00`
`ML-DSA-65 verify`	`234683` cycles	`235270` cycles	`1.00`
`ML-DSA-87 keypair`	`370822` cycles	`370874` cycles	`1.00`
`ML-DSA-87 sign`	`1078880` cycles	`1077097` cycles	`1.00`
`ML-DSA-87 verify`	`383211` cycles	`382857` cycles	`1.00`

This comment was automatically generated by workflow using github-action-benchmark.

oqs-bot

AMD EPYC 4th gen (c7a)

Details

Benchmark suite	Current: `1c31329`	Previous: `0f8b8e0`	Ratio
`ML-DSA-44 keypair`	`40627` cycles	`40587` cycles	`1.00`
`ML-DSA-44 sign`	`133192` cycles	`136713` cycles	`0.97`
`ML-DSA-44 verify`	`43499` cycles	`43374` cycles	`1.00`
`ML-DSA-65 keypair`	`72359` cycles	`71982` cycles	`1.01`
`ML-DSA-65 sign`	`214367` cycles	`214626` cycles	`1.00`
`ML-DSA-65 verify`	`73011` cycles	`73104` cycles	`1.00`
`ML-DSA-87 keypair`	`108917` cycles	`108890` cycles	`1.00`
`ML-DSA-87 sign`	`254496` cycles	`253022` cycles	`1.01`
`ML-DSA-87 verify`	`114127` cycles	`110459` cycles	`1.03`

This comment was automatically generated by workflow using github-action-benchmark.

oqs-bot

Intel Xeon 3rd gen (c6i) (no-opt)

Details

Benchmark suite	Current: `1c31329`	Previous: `0f8b8e0`	Ratio
`ML-DSA-44 keypair`	`157441` cycles	`157064` cycles	`1.00`
`ML-DSA-44 sign`	`552409` cycles	`549451` cycles	`1.01`
`ML-DSA-44 verify`	`169081` cycles	`168897` cycles	`1.00`
`ML-DSA-65 keypair`	`269113` cycles	`268697` cycles	`1.00`
`ML-DSA-65 sign`	`905233` cycles	`905890` cycles	`1.00`
`ML-DSA-65 verify`	`274920` cycles	`274888` cycles	`1.00`
`ML-DSA-87 keypair`	`448473` cycles	`448496` cycles	`1.00`
`ML-DSA-87 sign`	`1160497` cycles	`1159979` cycles	`1.00`
`ML-DSA-87 verify`	`458580` cycles	`458091` cycles	`1.00`

This comment was automatically generated by workflow using github-action-benchmark.

oqs-bot

Graviton3

Details

Benchmark suite	Current: `1c31329`	Previous: `0f8b8e0`	Ratio
`ML-DSA-44 keypair`	`72272` cycles	`72243` cycles	`1.00`
`ML-DSA-44 sign`	`213477` cycles	`213451` cycles	`1.00`
`ML-DSA-44 verify`	`75713` cycles	`75744` cycles	`1.00`
`ML-DSA-65 keypair`	`127604` cycles	`127603` cycles	`1.00`
`ML-DSA-65 sign`	`353407` cycles	`353426` cycles	`1.00`
`ML-DSA-65 verify`	`125750` cycles	`125745` cycles	`1.00`
`ML-DSA-87 keypair`	`208441` cycles	`208481` cycles	`1.00`
`ML-DSA-87 sign`	`452287` cycles	`452641` cycles	`1.00`
`ML-DSA-87 verify`	`205856` cycles	`205909` cycles	`1.00`

This comment was automatically generated by workflow using github-action-benchmark.

oqs-bot

Graviton4 (no-opt)

Details

Benchmark suite	Current: `1c31329`	Previous: `0f8b8e0`	Ratio
`ML-DSA-44 keypair`	`128174` cycles	`128347` cycles	`1.00`
`ML-DSA-44 sign`	`448073` cycles	`448111` cycles	`1.00`
`ML-DSA-44 verify`	`138265` cycles	`144871` cycles	`0.95`
`ML-DSA-65 keypair`	`220367` cycles	`220834` cycles	`1.00`
`ML-DSA-65 sign`	`729443` cycles	`729991` cycles	`1.00`
`ML-DSA-65 verify`	`223253` cycles	`223754` cycles	`1.00`
`ML-DSA-87 keypair`	`366585` cycles	`367262` cycles	`1.00`
`ML-DSA-87 sign`	`928832` cycles	`929744` cycles	`1.00`
`ML-DSA-87 verify`	`373916` cycles	`374445` cycles	`1.00`

This comment was automatically generated by workflow using github-action-benchmark.

oqs-bot

AMD EPYC 4th gen (c7a) (no-opt)

Details

Benchmark suite	Current: `1c31329`	Previous: `0f8b8e0`	Ratio
`ML-DSA-44 keypair`	`120691` cycles	`120833` cycles	`1.00`
`ML-DSA-44 sign`	`449407` cycles	`449229` cycles	`1.00`
`ML-DSA-44 verify`	`130264` cycles	`130297` cycles	`1.00`
`ML-DSA-65 keypair`	`204729` cycles	`204649` cycles	`1.00`
`ML-DSA-65 sign`	`730192` cycles	`731243` cycles	`1.00`
`ML-DSA-65 verify`	`210457` cycles	`210085` cycles	`1.00`
`ML-DSA-87 keypair`	`338245` cycles	`337488` cycles	`1.00`
`ML-DSA-87 sign`	`925719` cycles	`929314` cycles	`1.00`
`ML-DSA-87 verify`	`346809` cycles	`346839` cycles	`1.00`

This comment was automatically generated by workflow using github-action-benchmark.

mkannwischer · 2026-04-04T06:00:23Z

@willieyz, can you rebase this PR, please?

willieyz · 2026-04-05T03:59:54Z

Hello, @mkannwischer , thank you for reviewing, I had rebased it on top of the main!

oqs-bot

⚠️ Performance Alert ⚠️

Possible performance regression was detected for benchmark 'AMD EPYC 4th gen (c7a)'.
Benchmark result of this commit is worse than the previous benchmark result exceeding threshold 1.03.

Benchmark suite	Current: `1c31329`	Previous: `0f8b8e0`	Ratio
`ML-DSA-87 verify`	`114127` cycles	`110459` cycles	`1.03`

This comment was automatically generated by workflow using github-action-benchmark.

This commit adds poly_use_hint to bench --components for benchmarking the performance impact of the changes to: - poly_use_hint_32 - poly_use_hint_88 Signed-off-by: willieyz <willie.zhao@chelpis.com>

This commit replaces the AVX2 intrinsics implementation of poly_use_hint_32 and poly_use_hint_88 with a x86_64 assembly version, this is part of the effort to enable HOL-Light proofs. Signed-off-by: willieyz <willie.zhao@chelpis.com>

mkannwischer

Thanks @willieyz. I made a couple of small changes to the comments (to align it with the intrinsics). Now I am happy with the changes. The performance degradation is unfortunate, but we can revisit that in a follow-up.

@hanno-becker, @jakemas, could you also take a look?

willieyz force-pushed the eliminate-use_hint_32_88-intrinsics branch 9 times, most recently from 1ea9d5f to 8a19e9a Compare February 5, 2026 06:05

willieyz marked this pull request as ready for review February 5, 2026 06:39

willieyz requested a review from a team as a code owner February 5, 2026 06:39

willieyz added the benchmark label Feb 5, 2026

willieyz marked this pull request as draft February 5, 2026 07:19

github-actions bot reviewed Feb 5, 2026

View reviewed changes

oqs-bot reviewed Feb 5, 2026

View reviewed changes

github-actions bot reviewed Feb 5, 2026

View reviewed changes

oqs-bot reviewed Feb 5, 2026

View reviewed changes

willieyz force-pushed the eliminate-use_hint_32_88-intrinsics branch 6 times, most recently from f820653 to 91cddd6 Compare February 24, 2026 10:16

willieyz marked this pull request as draft February 25, 2026 01:24

willieyz force-pushed the eliminate-use_hint_32_88-intrinsics branch 3 times, most recently from d3f2de3 to e5ad167 Compare February 25, 2026 03:49

willieyz marked this pull request as ready for review February 25, 2026 03:56

willieyz requested a review from mkannwischer February 25, 2026 05:53

willieyz force-pushed the eliminate-use_hint_32_88-intrinsics branch from e5ad167 to d278675 Compare March 24, 2026 07:56

mkannwischer self-assigned this Apr 1, 2026

willieyz force-pushed the eliminate-use_hint_32_88-intrinsics branch from d278675 to 726fc9d Compare April 4, 2026 18:31

mkannwischer force-pushed the eliminate-use_hint_32_88-intrinsics branch from 726fc9d to 1c31329 Compare April 6, 2026 01:51

mkannwischer added benchmark and removed benchmark labels Apr 6, 2026

oqs-bot reviewed Apr 6, 2026

View reviewed changes

mkannwischer force-pushed the eliminate-use_hint_32_88-intrinsics branch from 1c31329 to 5437035 Compare April 6, 2026 02:27

willieyz added 2 commits April 6, 2026 10:27

Add components benchmark for poly_use_hint_32 and poly_use_hint_88

e5e58cf

This commit adds poly_use_hint to bench --components for benchmarking the performance impact of the changes to: - poly_use_hint_32 - poly_use_hint_88 Signed-off-by: willieyz <willie.zhao@chelpis.com>

mkannwischer force-pushed the eliminate-use_hint_32_88-intrinsics branch from 5437035 to c53c97f Compare April 6, 2026 02:27

mkannwischer approved these changes Apr 6, 2026

View reviewed changes

mkannwischer assigned hanno-becker and unassigned mkannwischer Apr 6, 2026

mkannwischer changed the title ~~Eliminate use_hint 32/88 intrinsics~~ x86_64: Eliminate use_hint 32/88 intrinsics Apr 6, 2026

mkannwischer added the x86_64 label Apr 6, 2026

Conversation

willieyz commented Feb 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

oqs-bot commented Feb 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

CBMC Results (ML-DSA-87)

Uh oh!

oqs-bot commented Feb 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

CBMC Results (ML-DSA-44)

Uh oh!

oqs-bot commented Feb 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

CBMC Results (ML-DSA-65)

Uh oh!

github-actions bot left a comment

Choose a reason for hiding this comment

Mac Mini (M1, 2020) benchmarks (opt)

Uh oh!

github-actions bot left a comment

Choose a reason for hiding this comment

Mac Mini (M1, 2020) benchmarks (no-opt)

Uh oh!

oqs-bot left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Intel Xeon 4th gen (c7i)

Uh oh!

oqs-bot left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Intel Xeon 4th gen (c7i) (no-opt)

Uh oh!

github-actions bot left a comment

Choose a reason for hiding this comment

Arm Cortex-A55 (Snapdragon 888) benchmarks (opt)

Uh oh!

oqs-bot left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

AMD EPYC 3rd gen (c6a)

Uh oh!

oqs-bot left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Intel Xeon 3rd gen (c6i)

Uh oh!

oqs-bot left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Graviton4

Uh oh!

oqs-bot left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

AMD EPYC 3rd gen (c6a) (no-opt)

Uh oh!

oqs-bot left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

AMD EPYC 4th gen (c7a)

Uh oh!

oqs-bot left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Intel Xeon 3rd gen (c6i) (no-opt)

Uh oh!

oqs-bot left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Graviton3

Uh oh!

oqs-bot left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Graviton4 (no-opt)

Uh oh!

oqs-bot left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

willieyz commented Feb 3, 2026 •

edited

Loading

oqs-bot commented Feb 3, 2026 •

edited

Loading

oqs-bot commented Feb 3, 2026 •

edited

Loading

oqs-bot commented Feb 3, 2026 •

edited

Loading

oqs-bot left a comment •

edited

Loading

oqs-bot left a comment •

edited

Loading

oqs-bot left a comment •

edited

Loading

oqs-bot left a comment •

edited

Loading

oqs-bot left a comment •

edited

Loading

oqs-bot left a comment •

edited

Loading

oqs-bot left a comment •

edited

Loading

oqs-bot left a comment •

edited

Loading

oqs-bot left a comment •

edited

Loading

oqs-bot left a comment •

edited

Loading

oqs-bot left a comment •

edited

Loading