AArch64: Support unaligned inputs for top-level APIs by mkannwischer · Pull Request #992 · pq-code-package/mldsa-native

mkannwischer · 2026-03-09T08:47:57Z

polyz_unpack_{17,19}_asm are the only AArch64 assembly routines that load from potentially unaligned user-provided buffers (the signature passed to verify). All other assembly (NTT, rej_uniform, etc.) operates on aligned internal buffers.

The ld1 with .4s/.2d element sizes and the ldr s/d instructions used here require 4/8-byte alignment on Device memory (bare-metal AArch64 without MMU). Replace with .16b element sizes and ld1 {v.8b}, which do not require alignment.

With this fixed, remove the MLD_TEST_NO_UNALIGNED workaround from the aarch64-virt baremetal platform so the unaligned-buffer functional test runs on baremetal as well.

github-actions

Arm Cortex-A76 (Raspberry Pi 5) benchmarks (opt)

Details

Benchmark suite	Current: `730be94`	Previous: `339e496`	Ratio
`ML-DSA-44 keypair`	`113103` cycles	`113147` cycles	`1.00`
`ML-DSA-44 sign`	`355458` cycles	`355180` cycles	`1.00`
`ML-DSA-44 verify`	`117778` cycles	`117748` cycles	`1.00`
`ML-DSA-65 keypair`	`196374` cycles	`196464` cycles	`1.00`
`ML-DSA-65 sign`	`588499` cycles	`588331` cycles	`1.00`
`ML-DSA-65 verify`	`194525` cycles	`194432` cycles	`1.00`
`ML-DSA-87 keypair`	`322283` cycles	`322129` cycles	`1.00`
`ML-DSA-87 sign`	`751478` cycles	`752572` cycles	`1.00`
`ML-DSA-87 verify`	`319567` cycles	`319915` cycles	`1.00`

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

github-actions

Arm Cortex-A76 (Raspberry Pi 5) benchmarks (no-opt)

Details

Benchmark suite	Current: `730be94`	Previous: `339e496`	Ratio
`ML-DSA-44 keypair`	`212369` cycles	`212671` cycles	`1.00`
`ML-DSA-44 sign`	`759604` cycles	`759391` cycles	`1.00`
`ML-DSA-44 verify`	`228684` cycles	`228993` cycles	`1.00`
`ML-DSA-65 keypair`	`379979` cycles	`380383` cycles	`1.00`
`ML-DSA-65 sign`	`1252009` cycles	`1251394` cycles	`1.00`
`ML-DSA-65 verify`	`371528` cycles	`372186` cycles	`1.00`
`ML-DSA-87 keypair`	`604426` cycles	`605169` cycles	`1.00`
`ML-DSA-87 sign`	`1593286` cycles	`1591399` cycles	`1.00`
`ML-DSA-87 verify`	`618286` cycles	`617312` cycles	`1.00`

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

oqs-bot

AMD EPYC 3rd gen (c6a)

Details

Benchmark suite	Current: `730be94`	Previous: `339e496`	Ratio
`ML-DSA-44 keypair`	`68994` cycles	`69132` cycles	`1.00`
`ML-DSA-44 sign`	`187502` cycles	`188226` cycles	`1.00`
`ML-DSA-44 verify`	`68839` cycles	`69219` cycles	`0.99`
`ML-DSA-65 keypair`	`119505` cycles	`119330` cycles	`1.00`
`ML-DSA-65 sign`	`300007` cycles	`300188` cycles	`1.00`
`ML-DSA-65 verify`	`115357` cycles	`115292` cycles	`1.00`
`ML-DSA-87 keypair`	`203395` cycles	`203812` cycles	`1.00`
`ML-DSA-87 sign`	`394047` cycles	`395329` cycles	`1.00`
`ML-DSA-87 verify`	`195377` cycles	`195762` 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: `730be94`	Previous: `339e496`	Ratio
`ML-DSA-44 keypair`	`56712` cycles	`56534` cycles	`1.00`
`ML-DSA-44 sign`	`181126` cycles	`181554` cycles	`1.00`
`ML-DSA-44 verify`	`60963` cycles	`61421` cycles	`0.99`
`ML-DSA-65 keypair`	`98739` cycles	`99035` cycles	`1.00`
`ML-DSA-65 sign`	`298327` cycles	`298572` cycles	`1.00`
`ML-DSA-65 verify`	`100362` cycles	`100520` cycles	`1.00`
`ML-DSA-87 keypair`	`152557` cycles	`152827` cycles	`1.00`
`ML-DSA-87 sign`	`354373` cycles	`355251` cycles	`1.00`
`ML-DSA-87 verify`	`153134` cycles	`153945` cycles	`0.99`

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

oqs-bot

Graviton4

Details

Benchmark suite	Current: `730be94`	Previous: `339e496`	Ratio
`ML-DSA-44 keypair`	`68316` cycles	`68196` cycles	`1.00`
`ML-DSA-44 sign`	`201904` cycles	`201973` cycles	`1.00`
`ML-DSA-44 verify`	`70857` cycles	`70544` cycles	`1.00`
`ML-DSA-65 keypair`	`121055` cycles	`121006` cycles	`1.00`
`ML-DSA-65 sign`	`330827` cycles	`331382` cycles	`1.00`
`ML-DSA-65 verify`	`117616` cycles	`118067` cycles	`1.00`
`ML-DSA-87 keypair`	`198261` cycles	`198246` cycles	`1.00`
`ML-DSA-87 sign`	`426845` cycles	`426898` cycles	`1.00`
`ML-DSA-87 verify`	`194474` cycles	`194357` 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: `730be94`	Previous: `339e496`	Ratio
`ML-DSA-44 keypair`	`134547` cycles	`134725` cycles	`1.00`
`ML-DSA-44 sign`	`524933` cycles	`524539` cycles	`1.00`
`ML-DSA-44 verify`	`147503` cycles	`148068` cycles	`1.00`
`ML-DSA-65 keypair`	`226615` cycles	`227491` cycles	`1.00`
`ML-DSA-65 sign`	`860232` cycles	`864775` cycles	`0.99`
`ML-DSA-65 verify`	`234772` cycles	`235581` cycles	`1.00`
`ML-DSA-87 keypair`	`372609` cycles	`373600` cycles	`1.00`
`ML-DSA-87 sign`	`1082433` cycles	`1084589` cycles	`1.00`
`ML-DSA-87 verify`	`384569` cycles	`385427` 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: `730be94`	Previous: `339e496`	Ratio
`ML-DSA-44 keypair`	`41364` cycles	`40899` cycles	`1.01`
`ML-DSA-44 sign`	`132691` cycles	`133167` cycles	`1.00`
`ML-DSA-44 verify`	`43485` cycles	`43772` cycles	`0.99`
`ML-DSA-65 keypair`	`71976` cycles	`71918` cycles	`1.00`
`ML-DSA-65 sign`	`214079` cycles	`214333` cycles	`1.00`
`ML-DSA-65 verify`	`72701` cycles	`72735` cycles	`1.00`
`ML-DSA-87 keypair`	`107842` cycles	`108000` cycles	`1.00`
`ML-DSA-87 sign`	`251560` cycles	`251840` cycles	`1.00`
`ML-DSA-87 verify`	`111427` cycles	`110205` cycles	`1.01`

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

oqs-bot

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

Details

Benchmark suite	Current: `730be94`	Previous: `339e496`	Ratio
`ML-DSA-44 keypair`	`157459` cycles	`157349` cycles	`1.00`
`ML-DSA-44 sign`	`549434` cycles	`550739` cycles	`1.00`
`ML-DSA-44 verify`	`169228` cycles	`169371` cycles	`1.00`
`ML-DSA-65 keypair`	`267721` cycles	`268104` cycles	`1.00`
`ML-DSA-65 sign`	`903267` cycles	`903355` cycles	`1.00`
`ML-DSA-65 verify`	`274193` cycles	`274822` cycles	`1.00`
`ML-DSA-87 keypair`	`448386` cycles	`448166` cycles	`1.00`
`ML-DSA-87 sign`	`1158776` cycles	`1157655` cycles	`1.00`
`ML-DSA-87 verify`	`458169` cycles	`458080` cycles	`1.00`

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

oqs-bot

Graviton4 (no-opt)

Details

Benchmark suite	Current: `730be94`	Previous: `339e496`	Ratio
`ML-DSA-44 keypair`	`128521` cycles	`128200` cycles	`1.00`
`ML-DSA-44 sign`	`447738` cycles	`447698` cycles	`1.00`
`ML-DSA-44 verify`	`138449` cycles	`144650` cycles	`0.96`
`ML-DSA-65 keypair`	`221172` cycles	`220629` cycles	`1.00`
`ML-DSA-65 sign`	`727356` cycles	`727485` cycles	`1.00`
`ML-DSA-65 verify`	`223271` cycles	`223203` cycles	`1.00`
`ML-DSA-87 keypair`	`365594` cycles	`365009` cycles	`1.00`
`ML-DSA-87 sign`	`927658` cycles	`925883` cycles	`1.00`
`ML-DSA-87 verify`	`373613` cycles	`372797` 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: `730be94`	Previous: `339e496`	Ratio
`ML-DSA-44 keypair`	`120234` cycles	`120079` cycles	`1.00`
`ML-DSA-44 sign`	`447266` cycles	`445606` cycles	`1.00`
`ML-DSA-44 verify`	`130859` cycles	`130275` cycles	`1.00`
`ML-DSA-65 keypair`	`204156` cycles	`204060` cycles	`1.00`
`ML-DSA-65 sign`	`729810` cycles	`727326` cycles	`1.00`
`ML-DSA-65 verify`	`210434` cycles	`209046` cycles	`1.01`
`ML-DSA-87 keypair`	`339526` cycles	`337824` cycles	`1.01`
`ML-DSA-87 sign`	`923483` cycles	`922255` cycles	`1.00`
`ML-DSA-87 verify`	`346563` cycles	`346248` cycles	`1.00`

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

oqs-bot

Graviton3

Details

Benchmark suite	Current: `730be94`	Previous: `339e496`	Ratio
`ML-DSA-44 keypair`	`72358` cycles	`72238` cycles	`1.00`
`ML-DSA-44 sign`	`212138` cycles	`211968` cycles	`1.00`
`ML-DSA-44 verify`	`75655` cycles	`75586` cycles	`1.00`
`ML-DSA-65 keypair`	`127554` cycles	`127524` cycles	`1.00`
`ML-DSA-65 sign`	`350157` cycles	`350022` cycles	`1.00`
`ML-DSA-65 verify`	`125339` cycles	`125427` cycles	`1.00`
`ML-DSA-87 keypair`	`205615` cycles	`208263` cycles	`0.99`
`ML-DSA-87 sign`	`443492` cycles	`448961` cycles	`0.99`
`ML-DSA-87 verify`	`205139` cycles	`205274` cycles	`1.00`

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

oqs-bot

Graviton3 (no-opt)

Details

Benchmark suite	Current: `730be94`	Previous: `339e496`	Ratio
`ML-DSA-44 keypair`	`138682` cycles	`138472` cycles	`1.00`
`ML-DSA-44 sign`	`484073` cycles	`483937` cycles	`1.00`
`ML-DSA-44 verify`	`148468` cycles	`162299` cycles	`0.91`
`ML-DSA-65 keypair`	`241326` cycles	`241394` cycles	`1.00`
`ML-DSA-65 sign`	`792952` cycles	`792376` cycles	`1.00`
`ML-DSA-65 verify`	`240749` cycles	`241234` cycles	`1.00`
`ML-DSA-87 keypair`	`395469` cycles	`396580` cycles	`1.00`
`ML-DSA-87 sign`	`1013175` cycles	`1012710` cycles	`1.00`
`ML-DSA-87 verify`	`402892` cycles	`402628` cycles	`1.00`

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

oqs-bot

Graviton2

Details

Benchmark suite	Current: `730be94`	Previous: `339e496`	Ratio
`ML-DSA-44 keypair`	`113371` cycles	`113500` cycles	`1.00`
`ML-DSA-44 sign`	`355488` cycles	`355773` cycles	`1.00`
`ML-DSA-44 verify`	`118001` cycles	`118295` cycles	`1.00`
`ML-DSA-65 keypair`	`196603` cycles	`196483` cycles	`1.00`
`ML-DSA-65 sign`	`588574` cycles	`588268` cycles	`1.00`
`ML-DSA-65 verify`	`194607` cycles	`194737` cycles	`1.00`
`ML-DSA-87 keypair`	`322577` cycles	`323067` cycles	`1.00`
`ML-DSA-87 sign`	`753420` cycles	`753282` cycles	`1.00`
`ML-DSA-87 verify`	`320148` cycles	`320275` cycles	`1.00`

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

oqs-bot

Graviton2 (no-opt)

Details

Benchmark suite	Current: `730be94`	Previous: `339e496`	Ratio
`ML-DSA-44 keypair`	`213057` cycles	`212962` cycles	`1.00`
`ML-DSA-44 sign`	`760326` cycles	`761109` cycles	`1.00`
`ML-DSA-44 verify`	`241281` cycles	`234656` cycles	`1.03`
`ML-DSA-65 keypair`	`380798` cycles	`380376` cycles	`1.00`
`ML-DSA-65 sign`	`1252337` cycles	`1253515` cycles	`1.00`
`ML-DSA-65 verify`	`372454` cycles	`371858` cycles	`1.00`
`ML-DSA-87 keypair`	`606382` cycles	`604334` cycles	`1.00`
`ML-DSA-87 sign`	`1593353` cycles	`1594506` cycles	`1.00`
`ML-DSA-87 verify`	`618144` cycles	`618425` 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: `730be94`	Previous: `339e496`	Ratio
`ML-DSA-44 keypair`	`34518` cycles	`35118` cycles	`0.98`
`ML-DSA-44 sign`	`119680` cycles	`120041` cycles	`1.00`
`ML-DSA-44 verify`	`38022` cycles	`38197` cycles	`1.00`
`ML-DSA-65 keypair`	`61187` cycles	`60705` cycles	`1.01`
`ML-DSA-65 sign`	`201763` cycles	`200568` cycles	`1.01`
`ML-DSA-65 verify`	`62728` cycles	`62638` cycles	`1.00`
`ML-DSA-87 keypair`	`94268` cycles	`93448` cycles	`1.01`
`ML-DSA-87 sign`	`238850` cycles	`237996` cycles	`1.00`
`ML-DSA-87 verify`	`96329` cycles	`95574` 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: `730be94`	Previous: `339e496`	Ratio
`ML-DSA-44 keypair`	`93618` cycles	`93421` cycles	`1.00`
`ML-DSA-44 sign`	`332433` cycles	`332489` cycles	`1.00`
`ML-DSA-44 verify`	`99590` cycles	`99620` cycles	`1.00`
`ML-DSA-65 keypair`	`159749` cycles	`159711` cycles	`1.00`
`ML-DSA-65 sign`	`544197` cycles	`544428` cycles	`1.00`
`ML-DSA-65 verify`	`160722` cycles	`160796` cycles	`1.00`
`ML-DSA-87 keypair`	`267030` cycles	`266842` cycles	`1.00`
`ML-DSA-87 sign`	`706066` cycles	`706051` cycles	`1.00`
`ML-DSA-87 verify`	`270154` cycles	`269870` cycles	`1.00`

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

polyz_unpack_{17,19}_asm are the only AArch64 assembly routines that load from potentially unaligned user-provided buffers (the signature passed to verify). All other assembly (NTT, rej_uniform, etc.) operates on aligned internal buffers. The ld1 with .4s/.2d element sizes and the ldr s/d instructions used here require 4/8-byte alignment on Device memory (bare-metal AArch64 without MMU). Replace with .16b element sizes and ld1 {v.8b}, which do not require alignment. With this fixed, remove the MLD_TEST_NO_UNALIGNED workaround from the aarch64-virt baremetal platform so the unaligned-buffer functional test runs on baremetal as well. Signed-off-by: Matthias J. Kannwischer <matthias@kannwischer.eu>

oqs-bot · 2026-03-09T09:07:17Z

CBMC Results (ML-DSA-65)

Full Results (175 proofs)

Proof	Status	Current	Previous	Change
`TOTAL`	✅	2328s	2385s	-2.4%
`polyvecl_pointwise_acc_montgomery_c`	✅	236s	258s	-9%
`mld_attempt_signature_generation`	✅	201s	202s	-0%
`sign_verify_internal`	✅	181s	181s	+0%
`polyvec_matrix_expand`	✅	149s	150s	-1%
`rej_uniform_native`	✅	147s	145s	+1%
`poly_pointwise_montgomery_c`	✅	142s	153s	-7%
`mld_invntt_layer`	✅	118s	124s	-5%
`mld_ct_memcmp`	✅	79s	86s	-8%
`polyvec_matrix_expand_serial`	✅	66s	70s	-6%
`sign_signature_internal`	✅	51s	52s	-2%
`keccak_squeezeblocks_x4`	✅	45s	42s	+7%
`mld_ntt_layer`	✅	43s	45s	-4%
`mld_compute_t0_t1_tr_from_sk_components`	✅	26s	26s	+0%
`fqmul`	✅	21s	20s	+5%
`polyveck_decompose`	✅	20s	18s	+11%
`rej_uniform`	✅	20s	21s	-5%
`rej_uniform_c`	✅	20s	20s	+0%
`polymat_permute_bitrev_to_custom`	✅	19s	18s	+6%
`poly_uniform_eta_4x`	✅	18s	17s	+6%
`poly_chknorm_c`	✅	16s	18s	-11%
`poly_uniform_4x`	✅	16s	17s	-6%
`polyt0_unpack`	✅	15s	19s	-21%
`polyvec_matrix_pointwise_montgomery`	✅	14s	13s	+8%
`polyveck_use_hint`	✅	14s	12s	+17%
`keccakf1600x4_permute_native`	✅	13s	15s	-13%
`mld_polyvecl_permute_bitrev_to_custom_native`	✅	12s	14s	-14%
`polyveck_invntt_tomont`	✅	12s	8s	+50%
`mld_check_pct`	✅	11s	13s	-15%
`mld_ntt_butterfly_block`	✅	11s	14s	-21%
`polyveck_add`	✅	10s	10s	+0%
`keccak_absorb_once_x4`	✅	9s	9s	+0%
`poly_caddq_c`	✅	9s	8s	+12%
`poly_invntt_tomont_c`	✅	9s	15s	-40%
`sign`	✅	9s	12s	-25%
`keccakf1600_permute`	✅	8s	7s	+14%
`poly_use_hint_c`	✅	8s	5s	+60%
`polyveck_ntt`	✅	8s	8s	+0%
`polyveck_reduce`	✅	8s	9s	-11%
`mld_sample_s1_s2_serial`	✅	7s	5s	+40%
`poly_decompose_c`	✅	7s	9s	-22%
`polyeta_unpack`	✅	7s	7s	+0%
`polyveck_caddq`	✅	7s	10s	-30%
`polyveck_make_hint`	✅	7s	7s	+0%
`polyveck_power2round`	✅	7s	12s	-42%
`polyveck_sub`	✅	7s	7s	+0%
`sign_pk_from_sk`	✅	7s	8s	-12%
`keccakf1600_permute_native`	✅	6s	8s	-25%
`mld_h`	✅	6s	5s	+20%
`mld_sample_s1_s2`	✅	6s	4s	+50%
`polyveck_shiftl`	✅	6s	9s	-33%
`polyvecl_ntt`	✅	6s	8s	-25%
`polyvecl_uniform_gamma1_serial`	✅	6s	5s	+20%
`rej_eta_c`	✅	6s	3s	+100%
`sign_signature`	✅	6s	6s	+0%
`sign_verify_pre_hash_internal`	✅	6s	6s	+0%
`sign_verify_pre_hash_shake256`	✅	6s	4s	+50%
`caddq`	✅	5s	2s	+150%
`keccak_absorb`	✅	5s	6s	-17%
`mld_compute_pack_z`	✅	5s	6s	-17%
`mld_ct_get_optblocker_i64`	✅	5s	3s	+67%
`montgomery_reduce`	✅	5s	1s	+400%
`pack_pk`	✅	5s	2s	+150%
`poly_caddq_native`	✅	5s	2s	+150%
`poly_decompose_native`	✅	5s	3s	+67%
`poly_power2round`	✅	5s	3s	+67%
`poly_shiftl`	✅	5s	3s	+67%
`poly_uniform`	✅	5s	4s	+25%
`polyveck_pack_t0`	✅	5s	2s	+150%
`polyveck_pointwise_poly_montgomery`	✅	5s	6s	-17%
`polyz_unpack_c`	✅	5s	6s	-17%
`shake256_squeeze`	✅	5s	2s	+150%
`sign_open`	✅	5s	5s	+0%
`sign_signature_extmu`	✅	5s	3s	+67%
`sign_signature_pre_hash_internal`	✅	5s	5s	+0%
`unpack_hints`	✅	5s	5s	+0%
`unpack_pk`	✅	5s	4s	+25%
`unpack_sig`	✅	5s	3s	+67%
`unpack_sk`	✅	5s	5s	+0%
`make_hint`	✅	4s	3s	+33%
`poly_add`	✅	4s	4s	+0%
`poly_caddq`	✅	4s	4s	+0%
`poly_caddq_native_aarch64`	✅	4s	2s	+100%
`poly_challenge`	✅	4s	5s	-20%
`poly_chknorm`	✅	4s	4s	+0%
`poly_ntt`	✅	4s	3s	+33%
`poly_ntt_c`	✅	4s	3s	+33%
`poly_ntt_native`	✅	4s	4s	+0%
`poly_pointwise_montgomery`	✅	4s	3s	+33%
`poly_reduce`	✅	4s	2s	+100%
`poly_sub`	✅	4s	3s	+33%
`poly_uniform_eta`	✅	4s	8s	-50%
`polyeta_pack`	✅	4s	3s	+33%
`polyt0_pack`	✅	4s	6s	-33%
`polyt1_pack`	✅	4s	1s	+300%
`polyveck_chknorm`	✅	4s	5s	-20%
`polyveck_unpack_eta`	✅	4s	3s	+33%
`polyvecl_chknorm`	✅	4s	5s	-20%
`polyvecl_pack_eta`	✅	4s	5s	-20%
`polyvecl_uniform_gamma1`	✅	4s	2s	+100%
`polyvecl_unpack_eta`	✅	4s	4s	+0%
`polyw1_pack`	✅	4s	2s	+100%
`polyz_unpack`	✅	4s	3s	+33%
`power2round`	✅	4s	5s	-20%
`rej_eta_native`	✅	4s	3s	+33%
`shake128_absorb`	✅	4s	3s	+33%
`shake128_squeeze`	✅	4s	2s	+100%
`sign_keypair`	✅	4s	3s	+33%
`sign_keypair_internal`	✅	4s	3s	+33%
`sign_signature_pre_hash_shake256`	✅	4s	4s	+0%
`decompose`	✅	3s	2s	+50%
`keccak_finalize`	✅	3s	2s	+50%
`keccak_init`	✅	3s	3s	+0%
`keccakf1600_xor_bytes (big endian)`	✅	3s	3s	+0%
`keccakf1600x4_permute`	✅	3s	2s	+50%
`keccakf1600x4_xor_bytes`	✅	3s	2s	+50%
`mld_ct_cmask_nonzero_u32`	✅	3s	5s	-40%
`mld_ct_cmask_nonzero_u8`	✅	3s	3s	+0%
`mld_ct_get_optblocker_u8`	✅	3s	1s	+200%
`mld_prepare_domain_separation_prefix`	✅	3s	2s	+50%
`mld_value_barrier_u8`	✅	3s	1s	+200%
`ntt_native_x86_64`	✅	3s	5s	-40%
`pack_sk`	✅	3s	2s	+50%
`poly_chknorm_native`	✅	3s	3s	+0%
`poly_invntt_tomont`	✅	3s	4s	-25%
`poly_uniform_gamma1`	✅	3s	3s	+0%
`poly_uniform_gamma1_4x`	✅	3s	7s	-57%
`poly_use_hint`	✅	3s	3s	+0%
`poly_use_hint_native`	✅	3s	3s	+0%
`polyt1_unpack`	✅	3s	2s	+50%
`polyveck_pack_eta`	✅	3s	3s	+0%
`polyveck_pack_w1`	✅	3s	4s	-25%
`polyveck_unpack_t0`	✅	3s	6s	-50%
`polyvecl_permute_bitrev_to_custom`	✅	3s	3s	+0%
`polyvecl_pointwise_acc_montgomery`	✅	3s	4s	-25%
`polyvecl_pointwise_acc_montgomery_native`	✅	3s	5s	-40%
`polyvecl_unpack_z`	✅	3s	3s	+0%
`polyz_pack`	✅	3s	4s	-25%
`polyz_unpack_native`	✅	3s	3s	+0%
`shake128_init`	✅	3s	2s	+50%
`shake128_release`	✅	3s	4s	-25%
`shake128x4_squeezeblocks`	✅	3s	2s	+50%
`shake256`	✅	3s	2s	+50%
`shake256_init`	✅	3s	3s	+0%
`shake256_release`	✅	3s	3s	+0%
`sign_verify`	✅	3s	4s	-25%
`sys_check_capability`	✅	3s	5s	-40%
`fqscale`	✅	2s	3s	-33%
`intt_native_x86_64`	✅	2s	3s	-33%
`keccakf1600_extract_bytes (big endian)`	✅	2s	5s	-60%
`keccakf1600_xor_bytes`	✅	2s	2s	+0%
`keccakf1600x4_extract_bytes`	✅	2s	2s	+0%
`mld_ct_cmask_neg_i32`	✅	2s	2s	+0%
`mld_ct_get_optblocker_u32`	✅	2s	4s	-50%
`mld_ct_sel_int32`	✅	2s	3s	-33%
`mld_keccakf1600_extract_bytes`	✅	2s	3s	-33%
`mld_value_barrier_i64`	✅	2s	3s	-33%
`mld_value_barrier_u32`	✅	2s	2s	+0%
`pack_sig_c_h`	✅	2s	2s	+0%
`pack_sig_z`	✅	2s	2s	+0%
`poly_decompose`	✅	2s	5s	-60%
`poly_invntt_tomont_native`	✅	2s	3s	-33%
`poly_make_hint`	✅	2s	4s	-50%
`poly_pointwise_montgomery_native`	✅	2s	2s	+0%
`rej_eta`	✅	2s	2s	+0%
`shake128x4_absorb_once`	✅	2s	2s	+0%
`shake256_absorb`	✅	2s	1s	+100%
`shake256_finalize`	✅	2s	3s	-33%
`shake256x4_absorb_once`	✅	2s	1s	+100%
`sign_verify_extmu`	✅	2s	5s	-60%
`use_hint`	✅	2s	3s	-33%
`keccak_squeeze`	✅	1s	2s	-50%
`mld_ct_abs_i32`	✅	1s	2s	-50%
`reduce32`	✅	1s	2s	-50%
`shake128_finalize`	✅	1s	2s	-50%
`shake256x4_squeezeblocks`	✅	1s	4s	-75%

oqs-bot · 2026-03-09T09:08:08Z

CBMC Results (ML-DSA-44)

Full Results (175 proofs)

Proof	Status	Current	Previous	Change
`TOTAL`	✅	2191s	2160s	+1.4%
`sign_verify_internal`	✅	277s	268s	+3%
`mld_attempt_signature_generation`	✅	239s	238s	+0%
`polyvecl_pointwise_acc_montgomery_c`	✅	227s	221s	+3%
`rej_uniform_native`	✅	152s	149s	+2%
`poly_pointwise_montgomery_c`	✅	147s	152s	-3%
`mld_ct_memcmp`	✅	88s	86s	+2%
`mld_invntt_layer`	✅	56s	55s	+2%
`mld_ntt_layer`	✅	47s	48s	-2%
`sign_signature_internal`	✅	47s	49s	-4%
`poly_invntt_tomont_c`	✅	46s	42s	+10%
`keccak_squeezeblocks_x4`	✅	42s	43s	-2%
`rej_uniform`	✅	21s	23s	-9%
`poly_uniform_eta_4x`	✅	20s	16s	+25%
`fqmul`	✅	19s	19s	+0%
`rej_uniform_c`	✅	18s	20s	-10%
`poly_uniform_4x`	✅	17s	17s	+0%
`polyt0_unpack`	✅	17s	16s	+6%
`polyvec_matrix_expand`	✅	17s	15s	+13%
`polymat_permute_bitrev_to_custom`	✅	16s	16s	+0%
`keccakf1600x4_permute_native`	✅	14s	13s	+8%
`mld_compute_t0_t1_tr_from_sk_components`	✅	14s	15s	-7%
`mld_polyvecl_permute_bitrev_to_custom_native`	✅	14s	16s	-12%
`poly_chknorm_c`	✅	14s	15s	-7%
`mld_ntt_butterfly_block`	✅	13s	15s	-13%
`polyz_unpack_c`	✅	13s	12s	+8%
`polyeta_unpack`	✅	12s	14s	-14%
`keccak_absorb_once_x4`	✅	10s	12s	-17%
`keccakf1600_permute`	✅	9s	7s	+29%
`sign`	✅	9s	7s	+29%
`keccakf1600_permute_native`	✅	8s	8s	+0%
`mld_check_pct`	✅	8s	6s	+33%
`polyveck_add`	✅	8s	7s	+14%
`polyveck_pointwise_poly_montgomery`	✅	8s	6s	+33%
`keccak_absorb`	✅	7s	6s	+17%
`poly_caddq_c`	✅	7s	9s	-22%
`polyvec_matrix_expand_serial`	✅	7s	8s	-12%
`polyveck_caddq`	✅	7s	4s	+75%
`sign_pk_from_sk`	✅	7s	6s	+17%
`mld_compute_pack_z`	✅	6s	7s	-14%
`mld_h`	✅	6s	5s	+20%
`polyvec_matrix_pointwise_montgomery`	✅	6s	5s	+20%
`polyveck_decompose`	✅	6s	6s	+0%
`polyveck_power2round`	✅	6s	4s	+50%
`polyveck_sub`	✅	6s	7s	-14%
`polyvecl_chknorm`	✅	6s	5s	+20%
`rej_eta_c`	✅	6s	4s	+50%
`sign_keypair`	✅	6s	2s	+200%
`sign_verify_pre_hash_shake256`	✅	6s	5s	+20%
`mld_prepare_domain_separation_prefix`	✅	5s	2s	+150%
`mld_sample_s1_s2`	✅	5s	5s	+0%
`pack_sig_z`	✅	5s	2s	+150%
`poly_caddq`	✅	5s	2s	+150%
`poly_caddq_native`	✅	5s	5s	+0%
`poly_challenge`	✅	5s	5s	+0%
`poly_decompose_c`	✅	5s	7s	-29%
`poly_invntt_tomont`	✅	5s	3s	+67%
`poly_ntt_native`	✅	5s	5s	+0%
`poly_uniform`	✅	5s	5s	+0%
`polyt0_pack`	✅	5s	4s	+25%
`polyveck_make_hint`	✅	5s	3s	+67%
`polyveck_ntt`	✅	5s	6s	-17%
`polyveck_pack_eta`	✅	5s	2s	+150%
`polyveck_reduce`	✅	5s	5s	+0%
`polyveck_use_hint`	✅	5s	7s	-29%
`polyvecl_pack_eta`	✅	5s	4s	+25%
`rej_eta_native`	✅	5s	3s	+67%
`shake128_squeeze`	✅	5s	3s	+67%
`sign_verify`	✅	5s	3s	+67%
`unpack_hints`	✅	5s	4s	+25%
`caddq`	✅	4s	2s	+100%
`keccak_finalize`	✅	4s	2s	+100%
`keccakf1600_xor_bytes (big endian)`	✅	4s	3s	+33%
`mld_ct_cmask_neg_i32`	✅	4s	2s	+100%
`mld_ct_cmask_nonzero_u32`	✅	4s	4s	+0%
`mld_ct_cmask_nonzero_u8`	✅	4s	4s	+0%
`mld_ct_get_optblocker_i64`	✅	4s	1s	+300%
`mld_ct_get_optblocker_u32`	✅	4s	3s	+33%
`mld_sample_s1_s2_serial`	✅	4s	5s	-20%
`mld_value_barrier_u8`	✅	4s	2s	+100%
`poly_caddq_native_aarch64`	✅	4s	3s	+33%
`poly_chknorm`	✅	4s	4s	+0%
`poly_decompose_native`	✅	4s	2s	+100%
`poly_make_hint`	✅	4s	5s	-20%
`poly_pointwise_montgomery_native`	✅	4s	3s	+33%
`poly_sub`	✅	4s	3s	+33%
`polyeta_pack`	✅	4s	4s	+0%
`polyt1_pack`	✅	4s	6s	-33%
`polyveck_chknorm`	✅	4s	5s	-20%
`polyveck_invntt_tomont`	✅	4s	4s	+0%
`polyveck_unpack_t0`	✅	4s	2s	+100%
`polyvecl_ntt`	✅	4s	5s	-20%
`polyvecl_unpack_eta`	✅	4s	4s	+0%
`polyz_unpack_native`	✅	4s	3s	+33%
`shake256`	✅	4s	2s	+100%
`shake256_init`	✅	4s	4s	+0%
`shake256_squeeze`	✅	4s	2s	+100%
`shake256x4_squeezeblocks`	✅	4s	3s	+33%
`sign_open`	✅	4s	4s	+0%
`sign_signature`	✅	4s	4s	+0%
`sign_signature_extmu`	✅	4s	5s	-20%
`sign_verify_extmu`	✅	4s	5s	-20%
`unpack_pk`	✅	4s	3s	+33%
`unpack_sig`	✅	4s	3s	+33%
`unpack_sk`	✅	4s	4s	+0%
`use_hint`	✅	4s	2s	+100%
`decompose`	✅	3s	2s	+50%
`intt_native_x86_64`	✅	3s	5s	-40%
`keccak_init`	✅	3s	3s	+0%
`keccakf1600x4_extract_bytes`	✅	3s	3s	+0%
`keccakf1600x4_xor_bytes`	✅	3s	2s	+50%
`make_hint`	✅	3s	1s	+200%
`mld_ct_get_optblocker_u8`	✅	3s	2s	+50%
`mld_ct_sel_int32`	✅	3s	4s	-25%
`mld_keccakf1600_extract_bytes`	✅	3s	1s	+200%
`mld_value_barrier_i64`	✅	3s	2s	+50%
`mld_value_barrier_u32`	✅	3s	4s	-25%
`montgomery_reduce`	✅	3s	4s	-25%
`ntt_native_x86_64`	✅	3s	4s	-25%
`pack_pk`	✅	3s	3s	+0%
`pack_sig_c_h`	✅	3s	3s	+0%
`poly_decompose`	✅	3s	6s	-50%
`poly_power2round`	✅	3s	1s	+200%
`poly_reduce`	✅	3s	4s	-25%
`poly_shiftl`	✅	3s	3s	+0%
`poly_uniform_eta`	✅	3s	5s	-40%
`poly_uniform_gamma1`	✅	3s	4s	-25%
`poly_uniform_gamma1_4x`	✅	3s	4s	-25%
`poly_use_hint`	✅	3s	4s	-25%
`poly_use_hint_c`	✅	3s	5s	-40%
`polyt1_unpack`	✅	3s	3s	+0%
`polyveck_pack_t0`	✅	3s	4s	-25%
`polyveck_shiftl`	✅	3s	5s	-40%
`polyvecl_permute_bitrev_to_custom`	✅	3s	4s	-25%
`polyvecl_pointwise_acc_montgomery`	✅	3s	3s	+0%
`polyvecl_pointwise_acc_montgomery_native`	✅	3s	7s	-57%
`polyvecl_uniform_gamma1`	✅	3s	2s	+50%
`polyvecl_uniform_gamma1_serial`	✅	3s	3s	+0%
`polyz_pack`	✅	3s	2s	+50%
`rej_eta`	✅	3s	3s	+0%
`shake128_finalize`	✅	3s	2s	+50%
`shake128_init`	✅	3s	2s	+50%
`shake128_release`	✅	3s	3s	+0%
`shake256_absorb`	✅	3s	2s	+50%
`sign_keypair_internal`	✅	3s	6s	-50%
`sign_signature_pre_hash_internal`	✅	3s	5s	-40%
`sign_signature_pre_hash_shake256`	✅	3s	4s	-25%
`sys_check_capability`	✅	3s	2s	+50%
`fqscale`	✅	2s	2s	+0%
`keccak_squeeze`	✅	2s	4s	-50%
`keccakf1600_xor_bytes`	✅	2s	3s	-33%
`keccakf1600x4_permute`	✅	2s	2s	+0%
`mld_ct_abs_i32`	✅	2s	3s	-33%
`pack_sk`	✅	2s	2s	+0%
`poly_add`	✅	2s	3s	-33%
`poly_chknorm_native`	✅	2s	3s	-33%
`poly_invntt_tomont_native`	✅	2s	4s	-50%
`poly_ntt`	✅	2s	4s	-50%
`poly_ntt_c`	✅	2s	3s	-33%
`poly_pointwise_montgomery`	✅	2s	2s	+0%
`poly_use_hint_native`	✅	2s	3s	-33%
`polyveck_pack_w1`	✅	2s	6s	-67%
`polyveck_unpack_eta`	✅	2s	3s	-33%
`polyvecl_unpack_z`	✅	2s	4s	-50%
`polyz_unpack`	✅	2s	1s	+100%
`power2round`	✅	2s	2s	+0%
`reduce32`	✅	2s	2s	+0%
`shake128_absorb`	✅	2s	1s	+100%
`shake128x4_absorb_once`	✅	2s	3s	-33%
`shake128x4_squeezeblocks`	✅	2s	2s	+0%
`shake256_finalize`	✅	2s	3s	-33%
`shake256_release`	✅	2s	5s	-60%
`shake256x4_absorb_once`	✅	2s	3s	-33%
`sign_verify_pre_hash_internal`	✅	2s	3s	-33%
`keccakf1600_extract_bytes (big endian)`	✅	1s	1s	+0%
`polyw1_pack`	✅	1s	6s	-83%

oqs-bot · 2026-03-09T09:10:45Z

CBMC Results (ML-DSA-87)

Full Results (175 proofs)

Proof	Status	Current	Previous	Change
`TOTAL`	✅	2434s	2459s	-1.0%
`sign_verify_internal`	✅	356s	358s	-1%
`mld_attempt_signature_generation`	✅	234s	230s	+2%
`polyvecl_pointwise_acc_montgomery_c`	✅	164s	174s	-6%
`polyvec_matrix_expand`	✅	155s	154s	+1%
`poly_pointwise_montgomery_c`	✅	139s	127s	+9%
`rej_uniform_native`	✅	137s	143s	-4%
`mld_invntt_layer`	✅	119s	117s	+2%
`polyvec_matrix_expand_serial`	✅	105s	108s	-3%
`mld_ct_memcmp`	✅	80s	78s	+3%
`sign_signature_internal`	✅	45s	46s	-2%
`keccak_squeezeblocks_x4`	✅	44s	42s	+5%
`mld_ntt_layer`	✅	43s	45s	-4%
`mld_compute_t0_t1_tr_from_sk_components`	✅	24s	27s	-11%
`polymat_permute_bitrev_to_custom`	✅	24s	24s	+0%
`rej_uniform`	✅	20s	22s	-9%
`fqmul`	✅	18s	23s	-22%
`poly_chknorm_c`	✅	18s	18s	+0%
`poly_uniform_eta_4x`	✅	17s	15s	+13%
`poly_uniform_4x`	✅	16s	15s	+7%
`rej_uniform_c`	✅	16s	15s	+7%
`polyt0_unpack`	✅	14s	15s	-7%
`polyvec_matrix_pointwise_montgomery`	✅	14s	12s	+17%
`polyeta_unpack`	✅	13s	14s	-7%
`keccakf1600x4_permute_native`	✅	12s	15s	-20%
`polyveck_power2round`	✅	12s	12s	+0%
`mld_ntt_butterfly_block`	✅	11s	12s	-8%
`poly_invntt_tomont_c`	✅	11s	11s	+0%
`keccak_absorb_once_x4`	✅	10s	9s	+11%
`keccakf1600_permute_native`	✅	10s	8s	+25%
`polyveck_reduce`	✅	10s	9s	+11%
`poly_decompose_c`	✅	9s	10s	-10%
`keccakf1600_permute`	✅	8s	7s	+14%
`mld_polyvecl_permute_bitrev_to_custom_native`	✅	8s	8s	+0%
`mld_sample_s1_s2_serial`	✅	8s	9s	-11%
`polyveck_decompose`	✅	8s	8s	+0%
`polyveck_use_hint`	✅	8s	9s	-11%
`sign`	✅	8s	8s	+0%
`mld_check_pct`	✅	7s	6s	+17%
`mld_sample_s1_s2`	✅	7s	5s	+40%
`poly_caddq_c`	✅	7s	6s	+17%
`polyveck_add`	✅	7s	5s	+40%
`polyveck_caddq`	✅	7s	6s	+17%
`polyveck_chknorm`	✅	7s	7s	+0%
`polyveck_invntt_tomont`	✅	7s	6s	+17%
`polyveck_make_hint`	✅	7s	4s	+75%
`polyveck_pointwise_poly_montgomery`	✅	7s	7s	+0%
`polyveck_shiftl`	✅	7s	7s	+0%
`polyveck_sub`	✅	7s	4s	+75%
`polyvecl_ntt`	✅	7s	8s	-12%
`sign_keypair_internal`	✅	7s	3s	+133%
`sign_pk_from_sk`	✅	7s	10s	-30%
`keccak_absorb`	✅	6s	6s	+0%
`keccak_squeeze`	✅	6s	4s	+50%
`mld_compute_pack_z`	✅	6s	5s	+20%
`sign_verify_pre_hash_internal`	✅	6s	2s	+200%
`poly_challenge`	✅	5s	3s	+67%
`poly_invntt_tomont_native`	✅	5s	3s	+67%
`poly_uniform`	✅	5s	5s	+0%
`poly_uniform_gamma1_4x`	✅	5s	6s	-17%
`poly_use_hint_c`	✅	5s	3s	+67%
`polyt0_pack`	✅	5s	4s	+25%
`polyveck_ntt`	✅	5s	6s	-17%
`polyvecl_chknorm`	✅	5s	5s	+0%
`polyvecl_uniform_gamma1`	✅	5s	4s	+25%
`sign_verify`	✅	5s	5s	+0%
`sign_verify_extmu`	✅	5s	4s	+25%
`unpack_hints`	✅	5s	4s	+25%
`caddq`	✅	4s	4s	+0%
`fqscale`	✅	4s	3s	+33%
`keccakf1600_xor_bytes`	✅	4s	1s	+300%
`keccakf1600_xor_bytes (big endian)`	✅	4s	3s	+33%
`mld_ct_cmask_nonzero_u8`	✅	4s	1s	+300%
`ntt_native_x86_64`	✅	4s	6s	-33%
`pack_pk`	✅	4s	4s	+0%
`pack_sig_c_h`	✅	4s	5s	-20%
`pack_sig_z`	✅	4s	2s	+100%
`pack_sk`	✅	4s	3s	+33%
`poly_caddq`	✅	4s	5s	-20%
`poly_make_hint`	✅	4s	4s	+0%
`poly_power2round`	✅	4s	4s	+0%
`poly_reduce`	✅	4s	4s	+0%
`poly_uniform_eta`	✅	4s	5s	-20%
`poly_uniform_gamma1`	✅	4s	4s	+0%
`poly_use_hint`	✅	4s	5s	-20%
`poly_use_hint_native`	✅	4s	2s	+100%
`polyveck_pack_t0`	✅	4s	3s	+33%
`polyveck_unpack_eta`	✅	4s	5s	-20%
`polyveck_unpack_t0`	✅	4s	4s	+0%
`polyvecl_uniform_gamma1_serial`	✅	4s	4s	+0%
`polyz_unpack_c`	✅	4s	4s	+0%
`rej_eta`	✅	4s	3s	+33%
`rej_eta_c`	✅	4s	4s	+0%
`rej_eta_native`	✅	4s	5s	-20%
`shake256_finalize`	✅	4s	4s	+0%
`sign_signature`	✅	4s	6s	-33%
`sys_check_capability`	✅	4s	3s	+33%
`decompose`	✅	3s	3s	+0%
`keccakf1600_extract_bytes (big endian)`	✅	3s	2s	+50%
`keccakf1600x4_extract_bytes`	✅	3s	4s	-25%
`keccakf1600x4_permute`	✅	3s	3s	+0%
`mld_ct_cmask_nonzero_u32`	✅	3s	3s	+0%
`mld_ct_get_optblocker_i64`	✅	3s	3s	+0%
`mld_h`	✅	3s	5s	-40%
`mld_prepare_domain_separation_prefix`	✅	3s	4s	-25%
`mld_value_barrier_u32`	✅	3s	4s	-25%
`poly_caddq_native`	✅	3s	5s	-40%
`poly_chknorm`	✅	3s	2s	+50%
`poly_decompose`	✅	3s	4s	-25%
`poly_ntt_c`	✅	3s	2s	+50%
`poly_pointwise_montgomery_native`	✅	3s	4s	-25%
`poly_shiftl`	✅	3s	1s	+200%
`polyt1_pack`	✅	3s	3s	+0%
`polyveck_pack_eta`	✅	3s	2s	+50%
`polyvecl_permute_bitrev_to_custom`	✅	3s	1s	+200%
`polyvecl_pointwise_acc_montgomery`	✅	3s	6s	-50%
`polyvecl_pointwise_acc_montgomery_native`	✅	3s	2s	+50%
`polyvecl_unpack_eta`	✅	3s	4s	-25%
`polyw1_pack`	✅	3s	3s	+0%
`polyz_pack`	✅	3s	3s	+0%
`polyz_unpack`	✅	3s	2s	+50%
`power2round`	✅	3s	3s	+0%
`shake128_absorb`	✅	3s	3s	+0%
`shake128_finalize`	✅	3s	3s	+0%
`shake128x4_squeezeblocks`	✅	3s	2s	+50%
`shake256_absorb`	✅	3s	1s	+200%
`shake256_init`	✅	3s	2s	+50%
`sign_keypair`	✅	3s	4s	-25%
`sign_open`	✅	3s	5s	-40%
`sign_signature_extmu`	✅	3s	2s	+50%
`sign_signature_pre_hash_shake256`	✅	3s	4s	-25%
`sign_verify_pre_hash_shake256`	✅	3s	4s	-25%
`unpack_sig`	✅	3s	2s	+50%
`unpack_sk`	✅	3s	4s	-25%
`intt_native_x86_64`	✅	2s	2s	+0%
`keccak_finalize`	✅	2s	2s	+0%
`keccak_init`	✅	2s	1s	+100%
`keccakf1600x4_xor_bytes`	✅	2s	1s	+100%
`make_hint`	✅	2s	2s	+0%
`mld_ct_abs_i32`	✅	2s	2s	+0%
`mld_ct_cmask_neg_i32`	✅	2s	3s	-33%
`mld_ct_get_optblocker_u32`	✅	2s	3s	-33%
`mld_ct_get_optblocker_u8`	✅	2s	3s	-33%
`mld_ct_sel_int32`	✅	2s	3s	-33%
`mld_keccakf1600_extract_bytes`	✅	2s	3s	-33%
`mld_value_barrier_i64`	✅	2s	4s	-50%
`mld_value_barrier_u8`	✅	2s	3s	-33%
`montgomery_reduce`	✅	2s	3s	-33%
`poly_add`	✅	2s	5s	-60%
`poly_caddq_native_aarch64`	✅	2s	3s	-33%
`poly_invntt_tomont`	✅	2s	3s	-33%
`poly_ntt_native`	✅	2s	3s	-33%
`poly_pointwise_montgomery`	✅	2s	5s	-60%
`polyeta_pack`	✅	2s	3s	-33%
`polyveck_pack_w1`	✅	2s	3s	-33%
`polyvecl_pack_eta`	✅	2s	2s	+0%
`polyvecl_unpack_z`	✅	2s	3s	-33%
`polyz_unpack_native`	✅	2s	4s	-50%
`reduce32`	✅	2s	3s	-33%
`shake128_init`	✅	2s	2s	+0%
`shake128_release`	✅	2s	4s	-50%
`shake128_squeeze`	✅	2s	2s	+0%
`shake256`	✅	2s	3s	-33%
`shake256_release`	✅	2s	5s	-60%
`shake256_squeeze`	✅	2s	3s	-33%
`shake256x4_squeezeblocks`	✅	2s	3s	-33%
`sign_signature_pre_hash_internal`	✅	2s	2s	+0%
`unpack_pk`	✅	2s	3s	-33%
`use_hint`	✅	2s	3s	-33%
`poly_chknorm_native`	✅	1s	3s	-67%
`poly_decompose_native`	✅	1s	2s	-50%
`poly_ntt`	✅	1s	4s	-75%
`poly_sub`	✅	1s	4s	-75%
`polyt1_unpack`	✅	1s	3s	-67%
`shake128x4_absorb_once`	✅	1s	3s	-67%
`shake256x4_absorb_once`	✅	1s	4s	-75%

hanno-becker

I agree with this change. I checked that the API documentation did not mention the AArch64 alignment requirements, so we don't need any change there.

mkannwischer added the benchmark label Mar 9, 2026

github-actions Bot reviewed Mar 9, 2026

View reviewed changes

oqs-bot reviewed Mar 9, 2026

View reviewed changes

mkannwischer force-pushed the unaligned-inputs branch from 730be94 to 18abe46 Compare March 9, 2026 09:03

mkannwischer mentioned this pull request Mar 9, 2026

aarch64: Fix verify with unaligned signature #991

Closed

mkannwischer marked this pull request as ready for review March 9, 2026 09:10

mkannwischer requested a review from a team as a code owner March 9, 2026 09:10

hanno-becker approved these changes Mar 9, 2026

View reviewed changes

hanno-becker merged commit 521f3bb into main Mar 9, 2026
371 checks passed

hanno-becker deleted the unaligned-inputs branch March 9, 2026 22:40

Conversation

mkannwischer commented Mar 9, 2026

Uh oh!

github-actions Bot left a comment

Choose a reason for hiding this comment

Arm Cortex-A76 (Raspberry Pi 5) benchmarks (opt)

Uh oh!

github-actions Bot left a comment

Choose a reason for hiding this comment

Arm Cortex-A76 (Raspberry Pi 5) benchmarks (no-opt)

Uh oh!

oqs-bot left a comment

Choose a reason for hiding this comment

AMD EPYC 3rd gen (c6a)

Uh oh!

oqs-bot left a comment

Choose a reason for hiding this comment

Intel Xeon 3rd gen (c6i)

Uh oh!

oqs-bot left a comment

Choose a reason for hiding this comment

Graviton4

Uh oh!

oqs-bot left a comment

Choose a reason for hiding this comment

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

Uh oh!

oqs-bot left a comment

Choose a reason for hiding this comment

AMD EPYC 4th gen (c7a)

Uh oh!

oqs-bot left a comment

Choose a reason for hiding this comment

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

Uh oh!

oqs-bot left a comment

Choose a reason for hiding this comment

Graviton4 (no-opt)

Uh oh!

oqs-bot left a comment

Choose a reason for hiding this comment

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

Uh oh!

oqs-bot left a comment

Choose a reason for hiding this comment

Graviton3

Uh oh!

oqs-bot left a comment

Choose a reason for hiding this comment

Graviton3 (no-opt)

Uh oh!

oqs-bot left a comment

Choose a reason for hiding this comment

Graviton2

Uh oh!

oqs-bot left a comment

Choose a reason for hiding this comment

Graviton2 (no-opt)

Uh oh!

oqs-bot left a comment

Choose a reason for hiding this comment

Intel Xeon 4th gen (c7i)

Uh oh!

oqs-bot left a comment

Choose a reason for hiding this comment

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

Uh oh!

oqs-bot commented Mar 9, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

CBMC Results (ML-DSA-65)

Uh oh!

oqs-bot commented Mar 9, 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 Mar 9, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

CBMC Results (ML-DSA-87)

Uh oh!

hanno-becker left a comment

oqs-bot commented Mar 9, 2026 •

edited

Loading

oqs-bot commented Mar 9, 2026 •

edited

Loading

oqs-bot commented Mar 9, 2026 •

edited

Loading