Speed up C-reference NTT/invNTT with twisted zetas + 2-layer merging by hanno-becker · Pull Request #1118 · pq-code-package/mldsa-native

hanno-becker · 2026-05-12T18:20:17Z

Replace the single-layer C-reference forward and inverse NTT in
mldsa/src/poly.c with one that merges two layers each.

Also, store each twiddle alongside its precomputed twist, letting
mld_fqmul(a, b, b_twisted) drop the multiply with MLDSA_Q^{-1}
that was previously hidden inside mld_montgomery_reduce.

Mirrors pq-code-package/mlkem-native/463 (@rod-chapman) and pq-code-package/mlkem-native/683

github-actions

Mac Mini (M1, 2020) benchmarks (opt)

Details

Benchmark suite	Current: `38f3f91`	Previous: `a71b5d2`	Ratio
`ML-DSA-44 keypair`	`46538` cycles	`46536` cycles	`1.00`
`ML-DSA-44 sign`	`131062` cycles	`131058` cycles	`1.00`
`ML-DSA-44 verify`	`47344` cycles	`47346` cycles	`1.00`
`ML-DSA-65 keypair`	`81686` cycles	`81682` cycles	`1.00`
`ML-DSA-65 sign`	`215381` cycles	`215367` cycles	`1.00`
`ML-DSA-65 verify`	`79305` cycles	`79306` cycles	`1.00`
`ML-DSA-87 keypair`	`132409` cycles	`132411` cycles	`1.00`
`ML-DSA-87 sign`	`277469` cycles	`277415` cycles	`1.00`
`ML-DSA-87 verify`	`134241` cycles	`134234` 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: `38f3f91`	Previous: `a71b5d2`	Ratio
`ML-DSA-44 keypair`	`113288` cycles	`112746` cycles	`1.00`
`ML-DSA-44 sign`	`403510` cycles	`400854` cycles	`1.01`
`ML-DSA-44 verify`	`121569` cycles	`120116` cycles	`1.01`
`ML-DSA-65 keypair`	`193992` cycles	`192886` cycles	`1.01`
`ML-DSA-65 sign`	`651150` cycles	`649888` cycles	`1.00`
`ML-DSA-65 verify`	`194758` cycles	`192947` cycles	`1.01`
`ML-DSA-87 keypair`	`319376` cycles	`318753` cycles	`1.00`
`ML-DSA-87 sign`	`831047` cycles	`828832` cycles	`1.00`
`ML-DSA-87 verify`	`329038` cycles	`326641` cycles	`1.01`

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

oqs-bot

Intel Xeon 4th gen (c7i)

Details

Benchmark suite	Current: `38f3f91`	Previous: `a71b5d2`	Ratio
`ML-DSA-44 keypair`	`45392` cycles	`45452` cycles	`1.00`
`ML-DSA-44 sign`	`136338` cycles	`136127` cycles	`1.00`
`ML-DSA-44 verify`	`47463` cycles	`47248` cycles	`1.00`
`ML-DSA-65 keypair`	`78478` cycles	`78548` cycles	`1.00`
`ML-DSA-65 sign`	`221924` cycles	`222310` cycles	`1.00`
`ML-DSA-65 verify`	`77951` cycles	`77415` cycles	`1.01`
`ML-DSA-87 keypair`	`126284` cycles	`124515` cycles	`1.01`
`ML-DSA-87 sign`	`279614` cycles	`275775` cycles	`1.01`
`ML-DSA-87 verify`	`123991` cycles	`122738` cycles	`1.01`

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

github-actions

SpacemiT K1 8 (Banana Pi F3) benchmarks (no-opt)

Details

Benchmark suite	Current: `38f3f91`	Previous: `a71b5d2`	Ratio
`ML-DSA-44 keypair`	`719602` cycles	`820317` cycles	`0.88`
`ML-DSA-44 sign`	`2246699` cycles	`3224057` cycles	`0.70`
`ML-DSA-44 verify`	`762835` cycles	`917185` cycles	`0.83`
`ML-DSA-65 keypair`	`1251304` cycles	`1391201` cycles	`0.90`
`ML-DSA-65 sign`	`3627167` cycles	`5232394` cycles	`0.69`
`ML-DSA-65 verify`	`1252102` cycles	`1464903` cycles	`0.85`
`ML-DSA-87 keypair`	`2106818` cycles	`2299598` cycles	`0.92`
`ML-DSA-87 sign`	`4794070` cycles	`6620374` cycles	`0.72`
`ML-DSA-87 verify`	`2120585` cycles	`2408309` cycles	`0.88`

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

oqs-bot

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

Details

Benchmark suite	Current: `38f3f91`	Previous: `a71b5d2`	Ratio
`ML-DSA-44 keypair`	`94177` cycles	`94270` cycles	`1.00`
`ML-DSA-44 sign`	`310203` cycles	`329827` cycles	`0.94`
`ML-DSA-44 verify`	`97031` cycles	`98781` cycles	`0.98`
`ML-DSA-65 keypair`	`158476` cycles	`161555` cycles	`0.98`
`ML-DSA-65 sign`	`500936` cycles	`538788` cycles	`0.93`
`ML-DSA-65 verify`	`157551` cycles	`160081` cycles	`0.98`
`ML-DSA-87 keypair`	`261642` cycles	`264477` cycles	`0.99`
`ML-DSA-87 sign`	`650338` cycles	`695417` cycles	`0.94`
`ML-DSA-87 verify`	`260657` cycles	`266020` cycles	`0.98`

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

oqs-bot

AMD EPYC 3rd gen (c6a)

Details

Benchmark suite	Current: `38f3f91`	Previous: `a71b5d2`	Ratio
`ML-DSA-44 keypair`	`56284` cycles	`57221` cycles	`0.98`
`ML-DSA-44 sign`	`167843` cycles	`166930` cycles	`1.01`
`ML-DSA-44 verify`	`60027` cycles	`58283` cycles	`1.03`
`ML-DSA-65 keypair`	`99358` cycles	`96734` cycles	`1.03`
`ML-DSA-65 sign`	`272455` cycles	`270287` cycles	`1.01`
`ML-DSA-65 verify`	`99857` cycles	`97285` cycles	`1.03`
`ML-DSA-87 keypair`	`158208` cycles	`161661` cycles	`0.98`
`ML-DSA-87 sign`	`334417` cycles	`335089` cycles	`1.00`
`ML-DSA-87 verify`	`157879` cycles	`153800` cycles	`1.03`

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

oqs-bot

⚠️ Performance Alert ⚠️

Possible performance regression was detected for benchmark 'AMD EPYC 3rd gen (c6a)'.
Benchmark result of this commit is worse than the previous benchmark result exceeding threshold 1.03.

Benchmark suite	Current: `dfda657`	Previous: `a71b5d2`	Ratio
`ML-DSA-87 verify`	`158518` cycles	`153800` cycles	`1.03`

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

oqs-bot

AMD EPYC 4th gen (c7a)

Details

Benchmark suite	Current: `38f3f91`	Previous: `a71b5d2`	Ratio
`ML-DSA-44 keypair`	`46964` cycles	`47162` cycles	`1.00`
`ML-DSA-44 sign`	`144418` cycles	`144655` cycles	`1.00`
`ML-DSA-44 verify`	`49951` cycles	`50104` cycles	`1.00`
`ML-DSA-65 keypair`	`84031` cycles	`83041` cycles	`1.01`
`ML-DSA-65 sign`	`232816` cycles	`229850` cycles	`1.01`
`ML-DSA-65 verify`	`83951` cycles	`83119` cycles	`1.01`
`ML-DSA-87 keypair`	`131766` cycles	`131179` cycles	`1.00`
`ML-DSA-87 sign`	`281804` cycles	`281956` cycles	`1.00`
`ML-DSA-87 verify`	`129740` cycles	`129801` 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: `38f3f91`	Previous: `a71b5d2`	Ratio
`ML-DSA-44 keypair`	`130860` cycles	`133677` cycles	`0.98`
`ML-DSA-44 sign`	`492691` cycles	`522396` cycles	`0.94`
`ML-DSA-44 verify`	`142552` cycles	`146685` cycles	`0.97`
`ML-DSA-65 keypair`	`219946` cycles	`223803` cycles	`0.98`
`ML-DSA-65 sign`	`797000` cycles	`850834` cycles	`0.94`
`ML-DSA-65 verify`	`227832` cycles	`233807` cycles	`0.97`
`ML-DSA-87 keypair`	`366825` cycles	`375278` cycles	`0.98`
`ML-DSA-87 sign`	`1017924` cycles	`1083775` cycles	`0.94`
`ML-DSA-87 verify`	`377803` cycles	`387875` cycles	`0.97`

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

oqs-bot

Intel Xeon 3rd gen (c6i)

Details

Benchmark suite	Current: `38f3f91`	Previous: `a71b5d2`	Ratio
`ML-DSA-44 keypair`	`61870` cycles	`61851` cycles	`1.00`
`ML-DSA-44 sign`	`190809` cycles	`191604` cycles	`1.00`
`ML-DSA-44 verify`	`66296` cycles	`66346` cycles	`1.00`
`ML-DSA-65 keypair`	`111571` cycles	`116244` cycles	`0.96`
`ML-DSA-65 sign`	`320974` cycles	`322314` cycles	`1.00`
`ML-DSA-65 verify`	`111271` cycles	`113021` cycles	`0.98`
`ML-DSA-87 keypair`	`172676` cycles	`172777` cycles	`1.00`
`ML-DSA-87 sign`	`380745` cycles	`384407` cycles	`0.99`
`ML-DSA-87 verify`	`172559` cycles	`174711` cycles	`0.99`

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

oqs-bot

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

Details

Benchmark suite	Current: `38f3f91`	Previous: `a71b5d2`	Ratio
`ML-DSA-44 keypair`	`115084` cycles	`118552` cycles	`0.97`
`ML-DSA-44 sign`	`407771` cycles	`446050` cycles	`0.91`
`ML-DSA-44 verify`	`123860` cycles	`128938` cycles	`0.96`
`ML-DSA-65 keypair`	`195940` cycles	`202120` cycles	`0.97`
`ML-DSA-65 sign`	`650499` cycles	`718282` cycles	`0.91`
`ML-DSA-65 verify`	`200657` cycles	`207260` cycles	`0.97`
`ML-DSA-87 keypair`	`325260` cycles	`334395` cycles	`0.97`
`ML-DSA-87 sign`	`836275` cycles	`919394` cycles	`0.91`
`ML-DSA-87 verify`	`331657` cycles	`342499` cycles	`0.97`

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

oqs-bot

Graviton4

Details

Benchmark suite	Current: `38f3f91`	Previous: `a71b5d2`	Ratio
`ML-DSA-44 keypair`	`67289` cycles	`67284` cycles	`1.00`
`ML-DSA-44 sign`	`201443` cycles	`201465` cycles	`1.00`
`ML-DSA-44 verify`	`70197` cycles	`70236` cycles	`1.00`
`ML-DSA-65 keypair`	`119340` cycles	`119592` cycles	`1.00`
`ML-DSA-65 sign`	`327977` cycles	`328455` cycles	`1.00`
`ML-DSA-65 verify`	`116780` cycles	`116975` cycles	`1.00`
`ML-DSA-87 keypair`	`196703` cycles	`196660` cycles	`1.00`
`ML-DSA-87 sign`	`425032` cycles	`424673` cycles	`1.00`
`ML-DSA-87 verify`	`193211` cycles	`193003` cycles	`1.00`

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

oqs-bot

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

Details

Benchmark suite	Current: `38f3f91`	Previous: `a71b5d2`	Ratio
`ML-DSA-44 keypair`	`144207` cycles	`150243` cycles	`0.96`
`ML-DSA-44 sign`	`482947` cycles	`543993` cycles	`0.89`
`ML-DSA-44 verify`	`152571` cycles	`162793` cycles	`0.94`
`ML-DSA-65 keypair`	`248619` cycles	`253828` cycles	`0.98`
`ML-DSA-65 sign`	`797728` cycles	`879250` cycles	`0.91`
`ML-DSA-65 verify`	`250855` cycles	`261051` cycles	`0.96`
`ML-DSA-87 keypair`	`414648` cycles	`428028` cycles	`0.97`
`ML-DSA-87 sign`	`1021025` cycles	`1133779` cycles	`0.90`
`ML-DSA-87 verify`	`417342` cycles	`438707` cycles	`0.95`

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

github-actions

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

Details

Benchmark suite	Current: `38f3f91`	Previous: `a71b5d2`	Ratio
`ML-DSA-44 keypair`	`112457` cycles	`112463` cycles	`1.00`
`ML-DSA-44 sign`	`354616` cycles	`354285` cycles	`1.00`
`ML-DSA-44 verify`	`117054` cycles	`117088` cycles	`1.00`
`ML-DSA-65 keypair`	`194541` cycles	`194650` cycles	`1.00`
`ML-DSA-65 sign`	`584282` cycles	`584287` cycles	`1.00`
`ML-DSA-65 verify`	`193240` cycles	`192995` cycles	`1.00`
`ML-DSA-87 keypair`	`320612` cycles	`321252` cycles	`1.00`
`ML-DSA-87 sign`	`748693` cycles	`749933` cycles	`1.00`
`ML-DSA-87 verify`	`317879` cycles	`318651` cycles	`1.00`

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

oqs-bot

Graviton4 (no-opt)

Details

Benchmark suite	Current: `38f3f91`	Previous: `a71b5d2`	Ratio
`ML-DSA-44 keypair`	`125226` cycles	`128439` cycles	`0.97`
`ML-DSA-44 sign`	`421018` cycles	`444902` cycles	`0.95`
`ML-DSA-44 verify`	`134116` cycles	`136577` cycles	`0.98`
`ML-DSA-65 keypair`	`216986` cycles	`220139` cycles	`0.99`
`ML-DSA-65 sign`	`681415` cycles	`718637` cycles	`0.95`
`ML-DSA-65 verify`	`218343` cycles	`221218` cycles	`0.99`
`ML-DSA-87 keypair`	`361874` cycles	`365464` cycles	`0.99`
`ML-DSA-87 sign`	`886549` cycles	`917775` cycles	`0.97`
`ML-DSA-87 verify`	`368411` cycles	`371436` cycles	`0.99`

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

oqs-bot

Graviton3

Details

Benchmark suite	Current: `38f3f91`	Previous: `a71b5d2`	Ratio
`ML-DSA-44 keypair`	`71566` cycles	`71503` cycles	`1.00`
`ML-DSA-44 sign`	`211564` cycles	`211366` cycles	`1.00`
`ML-DSA-44 verify`	`74848` cycles	`74967` cycles	`1.00`
`ML-DSA-65 keypair`	`125946` cycles	`125922` cycles	`1.00`
`ML-DSA-65 sign`	`347535` cycles	`348013` cycles	`1.00`
`ML-DSA-65 verify`	`123867` cycles	`124042` cycles	`1.00`
`ML-DSA-87 keypair`	`206188` cycles	`206707` cycles	`1.00`
`ML-DSA-87 sign`	`443030` cycles	`447437` cycles	`0.99`
`ML-DSA-87 verify`	`204440` cycles	`204174` cycles	`1.00`

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

oqs-bot

Graviton3 (no-opt)

Details

Benchmark suite	Current: `38f3f91`	Previous: `a71b5d2`	Ratio
`ML-DSA-44 keypair`	`135363` cycles	`137989` cycles	`0.98`
`ML-DSA-44 sign`	`457401` cycles	`481848` cycles	`0.95`
`ML-DSA-44 verify`	`145386` cycles	`148733` cycles	`0.98`
`ML-DSA-65 keypair`	`237653` cycles	`240592` cycles	`0.99`
`ML-DSA-65 sign`	`742121` cycles	`785306` cycles	`0.95`
`ML-DSA-65 verify`	`236161` cycles	`241073` cycles	`0.98`
`ML-DSA-87 keypair`	`390893` cycles	`395138` cycles	`0.99`
`ML-DSA-87 sign`	`958819` cycles	`1005113` cycles	`0.95`
`ML-DSA-87 verify`	`396238` cycles	`403185` cycles	`0.98`

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: `38f3f91`	Previous: `a71b5d2`	Ratio
`ML-DSA-44 keypair`	`204952` cycles	`212493` cycles	`0.96`
`ML-DSA-44 sign`	`690273` cycles	`756435` cycles	`0.91`
`ML-DSA-44 verify`	`218086` cycles	`229158` cycles	`0.95`
`ML-DSA-65 keypair`	`369004` cycles	`378664` cycles	`0.97`
`ML-DSA-65 sign`	`1129113` cycles	`1240500` cycles	`0.91`
`ML-DSA-65 verify`	`356558` cycles	`372168` cycles	`0.96`
`ML-DSA-87 keypair`	`589071` cycles	`602034` cycles	`0.98`
`ML-DSA-87 sign`	`1454044` cycles	`1579603` cycles	`0.92`
`ML-DSA-87 verify`	`596654` cycles	`618336` cycles	`0.96`

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

oqs-bot

Graviton2

Details

Benchmark suite	Current: `38f3f91`	Previous: `a71b5d2`	Ratio
`ML-DSA-44 keypair`	`112686` cycles	`112405` cycles	`1.00`
`ML-DSA-44 sign`	`356052` cycles	`354779` cycles	`1.00`
`ML-DSA-44 verify`	`117667` cycles	`117271` cycles	`1.00`
`ML-DSA-65 keypair`	`194353` cycles	`194498` cycles	`1.00`
`ML-DSA-65 sign`	`585143` cycles	`584927` cycles	`1.00`
`ML-DSA-65 verify`	`193113` cycles	`193003` cycles	`1.00`
`ML-DSA-87 keypair`	`321039` cycles	`321197` cycles	`1.00`
`ML-DSA-87 sign`	`749458` cycles	`749906` cycles	`1.00`
`ML-DSA-87 verify`	`318256` cycles	`318296` 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: `38f3f91`	Previous: `a71b5d2`	Ratio
`ML-DSA-44 keypair`	`270334` cycles	`270813` cycles	`1.00`
`ML-DSA-44 sign`	`814667` cycles	`814217` cycles	`1.00`
`ML-DSA-44 verify`	`274970` cycles	`273907` cycles	`1.00`
`ML-DSA-65 keypair`	`467712` cycles	`467318` cycles	`1.00`
`ML-DSA-65 sign`	`1367463` cycles	`1320861` cycles	`1.04`
`ML-DSA-65 verify`	`456340` cycles	`451480` cycles	`1.01`
`ML-DSA-87 keypair`	`805783` cycles	`802075` cycles	`1.00`
`ML-DSA-87 sign`	`1881318` cycles	`1880613` cycles	`1.00`
`ML-DSA-87 verify`	`787853` cycles	`779252` cycles	`1.01`

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

oqs-bot

Graviton2 (no-opt)

Details

Benchmark suite	Current: `38f3f91`	Previous: `a71b5d2`	Ratio
`ML-DSA-44 keypair`	`206083` cycles	`212221` cycles	`0.97`
`ML-DSA-44 sign`	`691174` cycles	`758032` cycles	`0.91`
`ML-DSA-44 verify`	`218535` cycles	`229778` cycles	`0.95`
`ML-DSA-65 keypair`	`369344` cycles	`378417` cycles	`0.98`
`ML-DSA-65 sign`	`1129681` cycles	`1241106` cycles	`0.91`
`ML-DSA-65 verify`	`356857` cycles	`372482` cycles	`0.96`
`ML-DSA-87 keypair`	`589415` cycles	`603782` cycles	`0.98`
`ML-DSA-87 sign`	`1455802` cycles	`1581844` cycles	`0.92`
`ML-DSA-87 verify`	`596976` cycles	`618440` cycles	`0.97`

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

github-actions

Arm Cortex-A55 (Snapdragon 888) benchmarks (no-opt)

Details

Benchmark suite	Current: `38f3f91`	Previous: `a71b5d2`	Ratio
`ML-DSA-44 keypair`	`391571` cycles	`458494` cycles	`0.85`
`ML-DSA-44 sign`	`1483786` cycles	`2126863` cycles	`0.70`
`ML-DSA-44 verify`	`443220` cycles	`552683` cycles	`0.80`
`ML-DSA-65 keypair`	`676713` cycles	`770631` cycles	`0.88`
`ML-DSA-65 sign`	`2440668` cycles	`3460057` cycles	`0.71`
`ML-DSA-65 verify`	`707703` cycles	`857490` cycles	`0.83`
`ML-DSA-87 keypair`	`1128071` cycles	`1249666` cycles	`0.90`
`ML-DSA-87 sign`	`3173423` cycles	`4303345` cycles	`0.74`
`ML-DSA-87 verify`	`1174442` cycles	`1370001` cycles	`0.86`

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

github-actions

Arm Cortex-A72 (Raspberry Pi 4) benchmarks (opt)

Details

Benchmark suite	Current: `38f3f91`	Previous: `a71b5d2`	Ratio
`ML-DSA-44 keypair`	`223259` cycles	`222102` cycles	`1.01`
`ML-DSA-44 sign`	`616370` cycles	`622325` cycles	`0.99`
`ML-DSA-44 verify`	`223427` cycles	`227406` cycles	`0.98`
`ML-DSA-65 keypair`	`396333` cycles	`385188` cycles	`1.03`
`ML-DSA-65 sign`	`1033678` cycles	`1017117` cycles	`1.02`
`ML-DSA-65 verify`	`378398` cycles	`371026` cycles	`1.02`
`ML-DSA-87 keypair`	`656252` cycles	`657858` cycles	`1.00`
`ML-DSA-87 sign`	`1362649` cycles	`1413224` cycles	`0.96`
`ML-DSA-87 verify`	`638462` cycles	`647577` cycles	`0.99`

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

github-actions

Arm Cortex-A72 (Raspberry Pi 4) benchmarks (no-opt)

Details

Benchmark suite	Current: `38f3f91`	Previous: `a71b5d2`	Ratio
`ML-DSA-44 keypair`	`290502` cycles	`317617` cycles	`0.91`
`ML-DSA-44 sign`	`980030` cycles	`1205867` cycles	`0.81`
`ML-DSA-44 verify`	`311188` cycles	`362564` cycles	`0.86`
`ML-DSA-65 keypair`	`549669` cycles	`577543` cycles	`0.95`
`ML-DSA-65 sign`	`1648606` cycles	`1961272` cycles	`0.84`
`ML-DSA-65 verify`	`505767` cycles	`556667` cycles	`0.91`
`ML-DSA-87 keypair`	`835538` cycles	`912005` cycles	`0.92`
`ML-DSA-87 sign`	`2090280` cycles	`2489549` cycles	`0.84`
`ML-DSA-87 verify`	`852209` cycles	`953121` cycles	`0.89`

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

oqs-bot · 2026-05-12T18:49:47Z

CBMC Results (ML-DSA-87, REDUCE-RAM)

⚠️ Attention Required

Proof	Status	Current	Previous	Change
`TOTAL`	⚠️	2924s	1460s	+100.3%
`fqmul`	⚠️	219s	29s	+655%

Full Results (196 proofs)

Proof	Status	Current	Previous	Change
`TOTAL`	⚠️	2924s	1460s	+100.3%
`mld_invntt_2_layers_block`	✅	799s	-	new
`mld_ntt_2_layers_block`	✅	508s	-	new
`fqmul`	⚠️	219s	29s	+655%
`polyvec_matrix_pointwise_montgomery_yvec`	✅	129s	122s	+6%
`rej_uniform_native`	✅	110s	107s	+3%
`poly_pointwise_montgomery_c`	✅	108s	157s	-31%
`mld_ntt_2_layers`	✅	81s	-	new
`mld_invntt_2_layers`	✅	74s	-	new
`mld_ct_memcmp`	✅	73s	70s	+4%
`sign_verify_internal`	✅	39s	40s	-3%
`mld_attempt_signature_generation`	✅	25s	24s	+4%
`keccakf1600x4_permute_native`	✅	20s	24s	-17%
`polyt0_unpack`	✅	18s	11s	+64%
`poly_chknorm_c`	✅	17s	14s	+21%
`polyeta_unpack`	✅	17s	15s	+13%
`poly_invntt_tomont_c`	✅	14s	9s	+56%
`polyveck_decompose`	✅	14s	16s	-12%
`mld_check_pct`	✅	12s	11s	+9%
`poly_add`	✅	12s	11s	+9%
`poly_uniform_eta_4x`	✅	11s	13s	-15%
`rej_uniform_c`	✅	11s	20s	-45%
`keccak_absorb_once_x4`	✅	9s	10s	-10%
`keccakf1600_permute_native`	✅	9s	5s	+80%
`keccakf1600_permute`	✅	8s	6s	+33%
`polyvec_matrix_pointwise_montgomery_row`	✅	8s	10s	-20%
`polyz_unpack_c`	✅	8s	5s	+60%
`rej_uniform`	✅	8s	18s	-56%
`compute_pack_t0_t1`	✅	7s	7s	+0%
`mld_sample_s1_s2_serial`	✅	7s	7s	+0%
`poly_caddq_c`	✅	7s	6s	+17%
`polyveck_chknorm`	✅	7s	6s	+17%
`polyveck_reduce`	✅	7s	8s	-12%
`polyvecl_ntt`	✅	7s	9s	-22%
`sign`	✅	7s	10s	-30%
`sign_signature_internal`	✅	7s	6s	+17%
`keccak_absorb`	✅	6s	6s	+0%
`pointwise_acc_native_aarch64`	✅	6s	6s	+0%
`pointwise_acc_native_x86_64`	✅	6s	6s	+0%
`poly_power2round`	✅	6s	7s	-14%
`polyt0_pack`	✅	6s	5s	+20%
`polyveck_invntt_tomont`	✅	6s	6s	+0%
`sign_verify_pre_hash_internal`	✅	6s	5s	+20%
`sign_verify_pre_hash_shake256`	✅	6s	4s	+50%
`caddq`	✅	5s	2s	+150%
`mld_prepare_domain_separation_prefix`	✅	5s	3s	+67%
`mld_sample_s1_s2`	✅	5s	4s	+25%
`ntt_native_x86_64`	✅	5s	3s	+67%
`pack_sig_z`	✅	5s	2s	+150%
`poly_caddq`	✅	5s	4s	+25%
`poly_decompose_native`	✅	5s	3s	+67%
`poly_ntt_native`	✅	5s	2s	+150%
`poly_uniform_eta`	✅	5s	3s	+67%
`polyt1_unpack`	✅	5s	2s	+150%
`polyveck_caddq`	✅	5s	4s	+25%
`polyveck_ntt`	✅	5s	2s	+150%
`polyveck_unpack_eta`	✅	5s	3s	+67%
`polyvecl_unpack_z`	✅	5s	6s	-17%
`polyz_unpack_19_native_aarch64`	✅	5s	4s	+25%
`shake128_squeeze`	✅	5s	3s	+67%
`sign_open`	✅	5s	6s	-17%
`sign_pk_from_sk`	✅	5s	6s	-17%
`sign_signature`	✅	5s	5s	+0%
`unpack_sk_t0hat`	✅	5s	5s	+0%
`keccak_f1600_x4_native_avx2`	✅	4s	4s	+0%
`poly_chknorm_native_aarch64`	✅	4s	3s	+33%
`poly_decompose_32_native_aarch64`	✅	4s	3s	+33%
`poly_invntt_tomont`	✅	4s	3s	+33%
`poly_invntt_tomont_native`	✅	4s	2s	+100%
`poly_permute_bitrev_to_custom_optional_native`	✅	4s	2s	+100%
`poly_pointwise_montgomery`	✅	4s	2s	+100%
`poly_reduce`	✅	4s	4s	+0%
`poly_shiftl`	✅	4s	4s	+0%
`poly_sub`	✅	4s	5s	-20%
`poly_uniform`	✅	4s	6s	-33%
`poly_use_hint`	✅	4s	3s	+33%
`polyvecl_chknorm`	✅	4s	5s	-20%
`polyvecl_uniform_gamma1`	✅	4s	3s	+33%
`polyw1_pack`	✅	4s	6s	-33%
`polyz_pack`	✅	4s	3s	+33%
`rej_eta_c`	✅	4s	5s	-20%
`shake256x4_absorb_once`	✅	4s	2s	+100%
`sign_signature_pre_hash_internal`	✅	4s	4s	+0%
`sign_signature_pre_hash_shake256`	✅	4s	2s	+100%
`sign_verify`	✅	4s	3s	+33%
`sign_verify_extmu`	✅	4s	5s	-20%
`sys_check_capability`	✅	4s	2s	+100%
`decompose`	✅	3s	2s	+50%
`fqscale`	✅	3s	2s	+50%
`intt_native_aarch64`	✅	3s	5s	-40%
`keccak_f1600_x1_native_aarch64_v84a`	✅	3s	2s	+50%
`keccak_f1600_x4_native_aarch64_v8a_scalar_hybrid`	✅	3s	2s	+50%
`keccak_f1600_x4_native_aarch64_v8a_v84a_scalar_hybrid`	✅	3s	3s	+0%
`keccak_squeezeblocks_x4`	✅	3s	3s	+0%
`keccakf1600_extract_bytes (big endian)`	✅	3s	1s	+200%
`keccakf1600x4_extract_bytes`	✅	3s	3s	+0%
`keccakf1600x4_permute`	✅	3s	3s	+0%
`keccakf1600x4_xor_bytes`	✅	3s	5s	-40%
`keccakf1600x4_xor_bytes_native`	✅	3s	3s	+0%
`make_hint`	✅	3s	2s	+50%
`mld_compute_pack_z`	✅	3s	5s	-40%
`mld_ct_abs_i32`	✅	3s	3s	+0%
`mld_ct_cmask_neg_i32`	✅	3s	3s	+0%
`mld_ct_get_optblocker_i64`	✅	3s	2s	+50%
`mld_ct_get_optblocker_u32`	✅	3s	2s	+50%
`mld_value_barrier_u32`	✅	3s	2s	+50%
`ntt_native_aarch64`	✅	3s	4s	-25%
`pack_sig_h`	✅	3s	3s	+0%
`pointwise_native_aarch64`	✅	3s	3s	+0%
`pointwise_native_x86_64`	✅	3s	2s	+50%
`poly_challenge`	✅	3s	5s	-40%
`poly_chknorm`	✅	3s	3s	+0%
`poly_chknorm_native`	✅	3s	5s	-40%
`poly_decompose`	✅	3s	3s	+0%
`poly_decompose_88_native_aarch64`	✅	3s	3s	+0%
`poly_decompose_c`	✅	3s	2s	+50%
`poly_ntt`	✅	3s	2s	+50%
`poly_pointwise_montgomery_native`	✅	3s	3s	+0%
`poly_use_hint_c`	✅	3s	4s	-25%
`poly_use_hint_native`	✅	3s	3s	+0%
`polyeta_pack`	✅	3s	2s	+50%
`polyt1_pack`	✅	3s	4s	-25%
`polyvec_matrix_expand`	✅	3s	2s	+50%
`polyveck_pack_w1`	✅	3s	2s	+50%
`polyvecl_pack_eta`	✅	3s	3s	+0%
`polyvecl_pointwise_acc_montgomery`	✅	3s	3s	+0%
`polyvecl_pointwise_acc_montgomery_native`	✅	3s	1s	+200%
`polyvecl_uniform_gamma1_serial`	✅	3s	2s	+50%
`polyvecl_unpack_eta`	✅	3s	6s	-50%
`polyz_unpack_native`	✅	3s	3s	+0%
`power2round`	✅	3s	4s	-25%
`rej_eta`	✅	3s	1s	+200%
`shake128_release`	✅	3s	4s	-25%
`shake128x4_absorb_once`	✅	3s	3s	+0%
`shake256_finalize`	✅	3s	3s	+0%
`shake256_release`	✅	3s	2s	+50%
`shake256_squeeze`	✅	3s	2s	+50%
`sign_keypair`	✅	3s	4s	-25%
`sign_keypair_internal`	✅	3s	4s	-25%
`sign_signature_extmu`	✅	3s	2s	+50%
`sk_s1hat_get_poly`	✅	3s	2s	+50%
`sk_s2hat_get_poly`	✅	3s	4s	-25%
`sk_t0hat_get_poly`	✅	3s	3s	+0%
`unpack_sk_s1hat`	✅	3s	1s	+200%
`unpack_sk_s2hat`	✅	3s	3s	+0%
`yvec_init`	✅	3s	2s	+50%
`intt_native_x86_64`	✅	2s	5s	-60%
`keccak_f1600_x4_native_aarch64_v84a`	✅	2s	3s	-33%
`keccak_init`	✅	2s	3s	-33%
`keccak_squeeze`	✅	2s	4s	-50%
`keccakf1600_xor_bytes (big endian)`	✅	2s	2s	+0%
`keccakf1600x4_extract_bytes_native`	✅	2s	2s	+0%
`mld_ct_cmask_nonzero_u32`	✅	2s	2s	+0%
`mld_ct_cmask_nonzero_u8`	✅	2s	2s	+0%
`mld_ct_get_optblocker_u8`	✅	2s	2s	+0%
`mld_ct_sel_int32`	✅	2s	1s	+100%
`mld_h`	✅	2s	3s	-33%
`mld_value_barrier_u8`	✅	2s	2s	+0%
`montgomery_reduce`	✅	2s	5s	-60%
`nttunpack_native_x86_64`	✅	2s	2s	+0%
`pack_sk_rho_key_tr_s2`	✅	2s	3s	-33%
`pack_sk_s1`	✅	2s	1s	+100%
`poly_caddq_native`	✅	2s	5s	-60%
`poly_ntt_c`	✅	2s	3s	-33%
`poly_permute_bitrev_to_custom_optional`	✅	2s	2s	+0%
`poly_uniform_4x`	✅	2s	3s	-33%
`poly_uniform_gamma1`	✅	2s	4s	-50%
`poly_uniform_gamma1_4x`	✅	2s	4s	-50%
`poly_use_hint_native_aarch64`	✅	2s	4s	-50%
`polyvec_matrix_expand_serial`	✅	2s	3s	-33%
`polyveck_pack_eta`	✅	2s	3s	-33%
`polyvecl_pointwise_acc_montgomery_c`	✅	2s	2s	+0%
`polyz_unpack_17_native_aarch64`	✅	2s	3s	-33%
`reduce32`	✅	2s	2s	+0%
`rej_eta_native`	✅	2s	4s	-50%
`shake128_absorb`	✅	2s	2s	+0%
`shake128_finalize`	✅	2s	2s	+0%
`shake128x4_squeezeblocks`	✅	2s	3s	-33%
`shake256`	✅	2s	3s	-33%
`shake256_absorb`	✅	2s	2s	+0%
`shake256_init`	✅	2s	2s	+0%
`shake256x4_squeezeblocks`	✅	2s	3s	-33%
`unpack_pk_t1`	✅	2s	3s	-33%
`use_hint`	✅	2s	3s	-33%
`yvec_get_poly`	✅	2s	4s	-50%
`keccak_f1600_x1_native_aarch64`	✅	1s	1s	+0%
`keccak_finalize`	✅	1s	1s	+0%
`keccakf1600_xor_bytes`	✅	1s	2s	-50%
`mld_keccakf1600_extract_bytes`	✅	1s	1s	+0%
`mld_polymat_expand_entry`	✅	1s	2s	-50%
`mld_value_barrier_i64`	✅	1s	1s	+0%
`pack_sig_c`	✅	1s	3s	-67%
`poly_caddq_native_aarch64`	✅	1s	2s	-50%
`polyz_unpack`	✅	1s	3s	-67%
`shake128_init`	✅	1s	3s	-67%
`sig_unpack_hints`	✅	1s	4s	-75%
`unpack_sk`	✅	1s	2s	-50%

oqs-bot · 2026-05-12T18:50:52Z

CBMC Results (ML-DSA-65, REDUCE-RAM)

⚠️ Attention Required

Proof	Status	Current	Previous	Change
`TOTAL`	⚠️	3004s	1496s	+100.8%
`fqmul`	⚠️	225s	28s	+704%

Full Results (196 proofs)

Proof	Status	Current	Previous	Change
`TOTAL`	⚠️	3004s	1496s	+100.8%
`mld_invntt_2_layers_block`	✅	819s	-	new
`mld_ntt_2_layers_block`	✅	510s	-	new
`fqmul`	⚠️	225s	28s	+704%
`polyvec_matrix_pointwise_montgomery_yvec`	✅	163s	149s	+9%
`poly_pointwise_montgomery_c`	✅	112s	167s	-33%
`rej_uniform_native`	✅	112s	106s	+6%
`mld_ntt_2_layers`	✅	85s	-	new
`mld_ct_memcmp`	✅	77s	75s	+3%
`mld_invntt_2_layers`	✅	77s	-	new
`mld_attempt_signature_generation`	✅	30s	27s	+11%
`keccakf1600x4_permute_native`	✅	23s	23s	+0%
`poly_chknorm_c`	✅	19s	14s	+36%
`polyvecl_chknorm`	✅	19s	19s	+0%
`sign_verify_internal`	✅	19s	18s	+6%
`poly_invntt_tomont_c`	✅	16s	7s	+129%
`polyt0_unpack`	✅	16s	10s	+60%
`mld_check_pct`	✅	14s	14s	+0%
`polyveck_decompose`	✅	14s	14s	+0%
`poly_add`	✅	13s	13s	+0%
`rej_uniform_c`	✅	13s	19s	-32%
`poly_uniform_eta_4x`	✅	11s	11s	+0%
`polyvec_matrix_pointwise_montgomery_row`	✅	10s	10s	+0%
`polyveck_caddq`	✅	10s	8s	+25%
`compute_pack_t0_t1`	✅	9s	9s	+0%
`keccak_absorb_once_x4`	✅	9s	9s	+0%
`keccakf1600_permute`	✅	9s	6s	+50%
`polyvecl_ntt`	✅	9s	6s	+50%
`rej_uniform`	✅	9s	18s	-50%
`keccak_absorb`	✅	8s	7s	+14%
`polyveck_reduce`	✅	8s	8s	+0%
`keccakf1600_permute_native`	✅	7s	7s	+0%
`mld_compute_pack_z`	✅	7s	5s	+40%
`poly_caddq_c`	✅	7s	7s	+0%
`poly_decompose_native`	✅	7s	5s	+40%
`polyveck_invntt_tomont`	✅	7s	5s	+40%
`polyveck_pack_w1`	✅	7s	3s	+133%
`polyz_unpack_c`	✅	7s	6s	+17%
`shake256_init`	✅	7s	3s	+133%
`sign`	✅	7s	7s	+0%
`sign_verify_pre_hash_internal`	✅	7s	3s	+133%
`keccak_f1600_x1_native_aarch64`	✅	6s	2s	+200%
`keccak_f1600_x4_native_avx2`	✅	6s	4s	+50%
`mld_h`	✅	6s	2s	+200%
`pointwise_acc_native_aarch64`	✅	6s	3s	+100%
`poly_shiftl`	✅	6s	8s	-25%
`intt_native_aarch64`	✅	5s	2s	+150%
`keccak_squeezeblocks_x4`	✅	5s	4s	+25%
`pack_sig_h`	✅	5s	2s	+150%
`pointwise_acc_native_x86_64`	✅	5s	6s	-17%
`poly_uniform_gamma1`	✅	5s	4s	+25%
`poly_use_hint_native`	✅	5s	1s	+400%
`shake128_init`	✅	5s	3s	+67%
`shake256_squeeze`	✅	5s	4s	+25%
`sign_keypair_internal`	✅	5s	3s	+67%
`sign_pk_from_sk`	✅	5s	5s	+0%
`sign_signature`	✅	5s	3s	+67%
`sign_verify_extmu`	✅	5s	4s	+25%
`sys_check_capability`	✅	5s	4s	+25%
`intt_native_x86_64`	✅	4s	3s	+33%
`mld_ct_cmask_neg_i32`	✅	4s	1s	+300%
`mld_keccakf1600_extract_bytes`	✅	4s	3s	+33%
`mld_sample_s1_s2`	✅	4s	5s	-20%
`nttunpack_native_x86_64`	✅	4s	4s	+0%
`pack_sig_z`	✅	4s	2s	+100%
`pointwise_native_x86_64`	✅	4s	4s	+0%
`poly_challenge`	✅	4s	4s	+0%
`poly_chknorm`	✅	4s	4s	+0%
`poly_chknorm_native_aarch64`	✅	4s	2s	+100%
`poly_decompose_32_native_aarch64`	✅	4s	4s	+0%
`poly_decompose_c`	✅	4s	4s	+0%
`poly_invntt_tomont_native`	✅	4s	3s	+33%
`poly_permute_bitrev_to_custom_optional_native`	✅	4s	2s	+100%
`poly_uniform`	✅	4s	4s	+0%
`poly_use_hint_c`	✅	4s	5s	-20%
`polyt0_pack`	✅	4s	3s	+33%
`polyt1_pack`	✅	4s	4s	+0%
`polyveck_unpack_eta`	✅	4s	3s	+33%
`polyw1_pack`	✅	4s	2s	+100%
`polyz_unpack_native`	✅	4s	2s	+100%
`sign_signature_pre_hash_internal`	✅	4s	3s	+33%
`sign_verify`	✅	4s	4s	+0%
`sk_s1hat_get_poly`	✅	4s	3s	+33%
`unpack_sk_t0hat`	✅	4s	1s	+300%
`yvec_init`	✅	4s	2s	+100%
`keccak_f1600_x1_native_aarch64_v84a`	✅	3s	3s	+0%
`keccak_f1600_x4_native_aarch64_v8a_v84a_scalar_hybrid`	✅	3s	1s	+200%
`keccak_init`	✅	3s	4s	-25%
`keccak_squeeze`	✅	3s	3s	+0%
`keccakf1600_extract_bytes (big endian)`	✅	3s	3s	+0%
`keccakf1600x4_extract_bytes_native`	✅	3s	4s	-25%
`keccakf1600x4_xor_bytes_native`	✅	3s	5s	-40%
`make_hint`	✅	3s	2s	+50%
`mld_ct_cmask_nonzero_u32`	✅	3s	2s	+50%
`mld_ct_cmask_nonzero_u8`	✅	3s	2s	+50%
`mld_ct_get_optblocker_i64`	✅	3s	2s	+50%
`mld_polymat_expand_entry`	✅	3s	2s	+50%
`mld_sample_s1_s2_serial`	✅	3s	5s	-40%
`mld_value_barrier_i64`	✅	3s	2s	+50%
`mld_value_barrier_u32`	✅	3s	2s	+50%
`mld_value_barrier_u8`	✅	3s	1s	+200%
`ntt_native_x86_64`	✅	3s	3s	+0%
`poly_caddq_native_aarch64`	✅	3s	3s	+0%
`poly_chknorm_native`	✅	3s	4s	-25%
`poly_decompose_88_native_aarch64`	✅	3s	3s	+0%
`poly_invntt_tomont`	✅	3s	1s	+200%
`poly_ntt`	✅	3s	4s	-25%
`poly_ntt_native`	✅	3s	4s	-25%
`poly_permute_bitrev_to_custom_optional`	✅	3s	3s	+0%
`poly_pointwise_montgomery_native`	✅	3s	2s	+50%
`poly_reduce`	✅	3s	5s	-40%
`poly_uniform_eta`	✅	3s	8s	-62%
`poly_uniform_gamma1_4x`	✅	3s	6s	-50%
`polyeta_unpack`	✅	3s	2s	+50%
`polyveck_chknorm`	✅	3s	3s	+0%
`polyveck_ntt`	✅	3s	2s	+50%
`polyvecl_pointwise_acc_montgomery`	✅	3s	3s	+0%
`polyvecl_pointwise_acc_montgomery_c`	✅	3s	2s	+50%
`polyvecl_uniform_gamma1`	✅	3s	3s	+0%
`polyvecl_uniform_gamma1_serial`	✅	3s	3s	+0%
`polyvecl_unpack_z`	✅	3s	3s	+0%
`polyz_pack`	✅	3s	5s	-40%
`polyz_unpack`	✅	3s	3s	+0%
`polyz_unpack_19_native_aarch64`	✅	3s	3s	+0%
`rej_eta`	✅	3s	2s	+50%
`rej_eta_c`	✅	3s	3s	+0%
`shake128_absorb`	✅	3s	2s	+50%
`shake128x4_squeezeblocks`	✅	3s	4s	-25%
`shake256_release`	✅	3s	2s	+50%
`shake256x4_absorb_once`	✅	3s	3s	+0%
`sign_keypair`	✅	3s	3s	+0%
`sign_open`	✅	3s	4s	-25%
`sign_signature_extmu`	✅	3s	4s	-25%
`sign_signature_internal`	✅	3s	7s	-57%
`sign_signature_pre_hash_shake256`	✅	3s	6s	-50%
`sign_verify_pre_hash_shake256`	✅	3s	4s	-25%
`unpack_pk_t1`	✅	3s	2s	+50%
`unpack_sk`	✅	3s	5s	-40%
`unpack_sk_s2hat`	✅	3s	3s	+0%
`caddq`	✅	2s	3s	-33%
`decompose`	✅	2s	4s	-50%
`fqscale`	✅	2s	5s	-60%
`keccak_f1600_x4_native_aarch64_v8a_scalar_hybrid`	✅	2s	4s	-50%
`keccak_finalize`	✅	2s	3s	-33%
`keccakf1600_xor_bytes`	✅	2s	4s	-50%
`keccakf1600_xor_bytes (big endian)`	✅	2s	2s	+0%
`keccakf1600x4_permute`	✅	2s	2s	+0%
`mld_ct_get_optblocker_u32`	✅	2s	2s	+0%
`mld_ct_get_optblocker_u8`	✅	2s	1s	+100%
`mld_prepare_domain_separation_prefix`	✅	2s	4s	-50%
`montgomery_reduce`	✅	2s	4s	-50%
`ntt_native_aarch64`	✅	2s	1s	+100%
`pack_sk_rho_key_tr_s2`	✅	2s	2s	+0%
`pack_sk_s1`	✅	2s	3s	-33%
`pointwise_native_aarch64`	✅	2s	4s	-50%
`poly_caddq`	✅	2s	2s	+0%
`poly_caddq_native`	✅	2s	5s	-60%
`poly_decompose`	✅	2s	2s	+0%
`poly_ntt_c`	✅	2s	4s	-50%
`poly_pointwise_montgomery`	✅	2s	3s	-33%
`poly_power2round`	✅	2s	7s	-71%
`poly_uniform_4x`	✅	2s	3s	-33%
`poly_use_hint`	✅	2s	4s	-50%
`poly_use_hint_native_aarch64`	✅	2s	2s	+0%
`polyeta_pack`	✅	2s	3s	-33%
`polyt1_unpack`	✅	2s	2s	+0%
`polyvec_matrix_expand`	✅	2s	4s	-50%
`polyveck_pack_eta`	✅	2s	2s	+0%
`polyvecl_pack_eta`	✅	2s	3s	-33%
`polyvecl_unpack_eta`	✅	2s	2s	+0%
`polyz_unpack_17_native_aarch64`	✅	2s	4s	-50%
`power2round`	✅	2s	3s	-33%
`reduce32`	✅	2s	3s	-33%
`rej_eta_native`	✅	2s	4s	-50%
`shake128_finalize`	✅	2s	1s	+100%
`shake128_release`	✅	2s	2s	+0%
`shake128_squeeze`	✅	2s	1s	+100%
`shake256`	✅	2s	1s	+100%
`shake256_absorb`	✅	2s	2s	+0%
`shake256_finalize`	✅	2s	2s	+0%
`shake256x4_squeezeblocks`	✅	2s	6s	-67%
`sig_unpack_hints`	✅	2s	2s	+0%
`sk_s2hat_get_poly`	✅	2s	3s	-33%
`sk_t0hat_get_poly`	✅	2s	2s	+0%
`use_hint`	✅	2s	4s	-50%
`yvec_get_poly`	✅	2s	2s	+0%
`keccak_f1600_x4_native_aarch64_v84a`	✅	1s	4s	-75%
`keccakf1600x4_extract_bytes`	✅	1s	3s	-67%
`keccakf1600x4_xor_bytes`	✅	1s	2s	-50%
`mld_ct_abs_i32`	✅	1s	4s	-75%
`mld_ct_sel_int32`	✅	1s	2s	-50%
`pack_sig_c`	✅	1s	2s	-50%
`poly_sub`	✅	1s	4s	-75%
`polyvec_matrix_expand_serial`	✅	1s	2s	-50%
`polyvecl_pointwise_acc_montgomery_native`	✅	1s	3s	-67%
`shake128x4_absorb_once`	✅	1s	2s	-50%
`unpack_sk_s1hat`	✅	1s	2s	-50%

oqs-bot · 2026-05-12T18:50:53Z

CBMC Results (ML-DSA-44, REDUCE-RAM)

⚠️ Attention Required

Proof	Status	Current	Previous	Change
`TOTAL`	⚠️	3130s	1443s	+116.9%
`fqmul`	⚠️	240s	28s	+757%
`poly_chknorm_c`	⚠️	21s	14s	+50%

Full Results (196 proofs)

Proof	Status	Current	Previous	Change
`TOTAL`	⚠️	3130s	1443s	+116.9%
`mld_invntt_2_layers_block`	✅	889s	-	new
`mld_ntt_2_layers_block`	✅	565s	-	new
`fqmul`	⚠️	240s	28s	+757%
`poly_pointwise_montgomery_c`	✅	119s	176s	-32%
`rej_uniform_native`	✅	112s	108s	+4%
`polyvec_matrix_pointwise_montgomery_yvec`	✅	94s	88s	+7%
`mld_ntt_2_layers`	✅	87s	-	new
`mld_ct_memcmp`	✅	79s	76s	+4%
`mld_invntt_2_layers`	✅	76s	-	new
`mld_attempt_signature_generation`	✅	26s	26s	+0%
`keccakf1600x4_permute_native`	✅	22s	23s	-4%
`poly_chknorm_c`	⚠️	21s	14s	+50%
`sign_verify_internal`	✅	19s	21s	-10%
`poly_invntt_tomont_c`	✅	17s	8s	+112%
`polyt0_unpack`	✅	17s	10s	+70%
`polyeta_unpack`	✅	16s	16s	+0%
`mld_check_pct`	✅	14s	15s	-7%
`rej_uniform_c`	✅	14s	18s	-22%
`polyveck_chknorm`	✅	13s	12s	+8%
`polyz_unpack_c`	✅	13s	14s	-7%
`poly_uniform_eta_4x`	✅	12s	11s	+9%
`keccak_absorb_once_x4`	✅	11s	11s	+0%
`poly_decompose_c`	✅	11s	9s	+22%
`rej_uniform`	✅	11s	20s	-45%
`compute_pack_t0_t1`	✅	10s	7s	+43%
`poly_add`	✅	10s	10s	+0%
`keccakf1600_permute_native`	✅	9s	7s	+29%
`poly_caddq_c`	✅	9s	7s	+29%
`polyvec_matrix_pointwise_montgomery_row`	✅	9s	8s	+12%
`sign`	✅	9s	6s	+50%
`poly_challenge`	✅	7s	4s	+75%
`poly_pointwise_montgomery_native`	✅	7s	5s	+40%
`polyveck_reduce`	✅	7s	6s	+17%
`sign_signature_pre_hash_shake256`	✅	7s	5s	+40%
`keccak_squeezeblocks_x4`	✅	6s	5s	+20%
`keccakf1600_permute`	✅	6s	8s	-25%
`keccakf1600x4_extract_bytes_native`	✅	6s	2s	+200%
`mld_compute_pack_z`	✅	6s	6s	+0%
`mld_ct_cmask_nonzero_u32`	✅	6s	2s	+200%
`pointwise_acc_native_aarch64`	✅	6s	5s	+20%
`pointwise_acc_native_x86_64`	✅	6s	5s	+20%
`poly_invntt_tomont_native`	✅	6s	3s	+100%
`poly_ntt_native`	✅	6s	3s	+100%
`polyveck_unpack_eta`	✅	6s	3s	+100%
`polyvecl_ntt`	✅	6s	3s	+100%
`sign_keypair`	✅	6s	2s	+200%
`unpack_sk_s2hat`	✅	6s	2s	+200%
`keccak_absorb`	✅	5s	9s	-44%
`make_hint`	✅	5s	2s	+150%
`mld_h`	✅	5s	3s	+67%
`mld_polymat_expand_entry`	✅	5s	3s	+67%
`mld_value_barrier_i64`	✅	5s	3s	+67%
`nttunpack_native_x86_64`	✅	5s	2s	+150%
`pack_sk_s1`	✅	5s	3s	+67%
`poly_caddq_native`	✅	5s	3s	+67%
`poly_chknorm`	✅	5s	3s	+67%
`poly_chknorm_native`	✅	5s	6s	-17%
`poly_shiftl`	✅	5s	4s	+25%
`poly_uniform`	✅	5s	4s	+25%
`polyvec_matrix_expand`	✅	5s	5s	+0%
`polyvec_matrix_expand_serial`	✅	5s	3s	+67%
`polyvecl_uniform_gamma1_serial`	✅	5s	3s	+67%
`polyz_pack`	✅	5s	3s	+67%
`polyz_unpack_native`	✅	5s	4s	+25%
`rej_eta_c`	✅	5s	3s	+67%
`rej_eta_native`	✅	5s	2s	+150%
`sign_verify`	✅	5s	3s	+67%
`sign_verify_pre_hash_shake256`	✅	5s	4s	+25%
`sk_s2hat_get_poly`	✅	5s	4s	+25%
`sk_t0hat_get_poly`	✅	5s	2s	+150%
`unpack_pk_t1`	✅	5s	1s	+400%
`intt_native_x86_64`	✅	4s	4s	+0%
`keccakf1600_xor_bytes`	✅	4s	2s	+100%
`keccakf1600_xor_bytes (big endian)`	✅	4s	2s	+100%
`mld_sample_s1_s2_serial`	✅	4s	4s	+0%
`poly_caddq_native_aarch64`	✅	4s	3s	+33%
`poly_invntt_tomont`	✅	4s	4s	+0%
`poly_ntt_c`	✅	4s	3s	+33%
`poly_use_hint`	✅	4s	4s	+0%
`poly_use_hint_native_aarch64`	✅	4s	1s	+300%
`polyt1_unpack`	✅	4s	3s	+33%
`polyveck_caddq`	✅	4s	4s	+0%
`polyveck_decompose`	✅	4s	5s	-20%
`polyveck_invntt_tomont`	✅	4s	5s	-20%
`polyveck_pack_eta`	✅	4s	2s	+100%
`polyvecl_pointwise_acc_montgomery_native`	✅	4s	5s	-20%
`polyvecl_unpack_eta`	✅	4s	1s	+300%
`polyw1_pack`	✅	4s	3s	+33%
`shake128_finalize`	✅	4s	3s	+33%
`shake128x4_squeezeblocks`	✅	4s	3s	+33%
`shake256_squeeze`	✅	4s	2s	+100%
`shake256x4_absorb_once`	✅	4s	4s	+0%
`sig_unpack_hints`	✅	4s	2s	+100%
`sign_keypair_internal`	✅	4s	2s	+100%
`sign_signature`	✅	4s	6s	-33%
`sign_signature_internal`	✅	4s	4s	+0%
`sign_signature_pre_hash_internal`	✅	4s	3s	+33%
`sign_verify_pre_hash_internal`	✅	4s	4s	+0%
`unpack_sk`	✅	4s	2s	+100%
`use_hint`	✅	4s	4s	+0%
`yvec_init`	✅	4s	5s	-20%
`caddq`	✅	3s	2s	+50%
`fqscale`	✅	3s	1s	+200%
`keccak_f1600_x1_native_aarch64_v84a`	✅	3s	3s	+0%
`keccak_f1600_x4_native_aarch64_v84a`	✅	3s	2s	+50%
`keccak_f1600_x4_native_aarch64_v8a_scalar_hybrid`	✅	3s	3s	+0%
`keccak_finalize`	✅	3s	2s	+50%
`keccak_init`	✅	3s	2s	+50%
`keccakf1600_extract_bytes (big endian)`	✅	3s	4s	-25%
`keccakf1600x4_permute`	✅	3s	2s	+50%
`mld_ct_get_optblocker_i64`	✅	3s	3s	+0%
`mld_ct_get_optblocker_u8`	✅	3s	1s	+200%
`mld_sample_s1_s2`	✅	3s	3s	+0%
`montgomery_reduce`	✅	3s	4s	-25%
`ntt_native_x86_64`	✅	3s	5s	-40%
`pack_sig_c`	✅	3s	4s	-25%
`pack_sk_rho_key_tr_s2`	✅	3s	4s	-25%
`pointwise_native_aarch64`	✅	3s	3s	+0%
`poly_chknorm_native_aarch64`	✅	3s	4s	-25%
`poly_decompose`	✅	3s	1s	+200%
`poly_decompose_native`	✅	3s	2s	+50%
`poly_permute_bitrev_to_custom_optional_native`	✅	3s	3s	+0%
`poly_reduce`	✅	3s	2s	+50%
`poly_uniform_eta`	✅	3s	4s	-25%
`poly_uniform_gamma1_4x`	✅	3s	3s	+0%
`polyeta_pack`	✅	3s	2s	+50%
`polyt1_pack`	✅	3s	7s	-57%
`polyvecl_chknorm`	✅	3s	5s	-40%
`polyvecl_pack_eta`	✅	3s	2s	+50%
`polyvecl_pointwise_acc_montgomery`	✅	3s	2s	+50%
`polyz_unpack`	✅	3s	3s	+0%
`polyz_unpack_17_native_aarch64`	✅	3s	4s	-25%
`power2round`	✅	3s	3s	+0%
`reduce32`	✅	3s	3s	+0%
`rej_eta`	✅	3s	4s	-25%
`shake128_absorb`	✅	3s	3s	+0%
`shake128_init`	✅	3s	3s	+0%
`shake128_release`	✅	3s	3s	+0%
`shake128_squeeze`	✅	3s	2s	+50%
`sign_open`	✅	3s	4s	-25%
`sign_pk_from_sk`	✅	3s	6s	-50%
`sign_signature_extmu`	✅	3s	3s	+0%
`sign_verify_extmu`	✅	3s	3s	+0%
`sk_s1hat_get_poly`	✅	3s	2s	+50%
`unpack_sk_s1hat`	✅	3s	4s	-25%
`yvec_get_poly`	✅	3s	3s	+0%
`decompose`	✅	2s	5s	-60%
`intt_native_aarch64`	✅	2s	3s	-33%
`keccak_f1600_x4_native_aarch64_v8a_v84a_scalar_hybrid`	✅	2s	3s	-33%
`keccak_f1600_x4_native_avx2`	✅	2s	1s	+100%
`keccak_squeeze`	✅	2s	1s	+100%
`keccakf1600x4_extract_bytes`	✅	2s	4s	-50%
`keccakf1600x4_xor_bytes`	✅	2s	3s	-33%
`keccakf1600x4_xor_bytes_native`	✅	2s	4s	-50%
`mld_ct_abs_i32`	✅	2s	2s	+0%
`mld_ct_cmask_nonzero_u8`	✅	2s	3s	-33%
`mld_ct_get_optblocker_u32`	✅	2s	1s	+100%
`mld_ct_sel_int32`	✅	2s	2s	+0%
`mld_keccakf1600_extract_bytes`	✅	2s	1s	+100%
`mld_prepare_domain_separation_prefix`	✅	2s	3s	-33%
`mld_value_barrier_u32`	✅	2s	3s	-33%
`ntt_native_aarch64`	✅	2s	3s	-33%
`pack_sig_h`	✅	2s	5s	-60%
`pack_sig_z`	✅	2s	3s	-33%
`pointwise_native_x86_64`	✅	2s	2s	+0%
`poly_caddq`	✅	2s	3s	-33%
`poly_decompose_32_native_aarch64`	✅	2s	3s	-33%
`poly_decompose_88_native_aarch64`	✅	2s	4s	-50%
`poly_permute_bitrev_to_custom_optional`	✅	2s	1s	+100%
`poly_power2round`	✅	2s	6s	-67%
`poly_sub`	✅	2s	2s	+0%
`poly_uniform_4x`	✅	2s	3s	-33%
`poly_uniform_gamma1`	✅	2s	2s	+0%
`poly_use_hint_c`	✅	2s	3s	-33%
`poly_use_hint_native`	✅	2s	2s	+0%
`polyt0_pack`	✅	2s	4s	-50%
`polyveck_ntt`	✅	2s	5s	-60%
`polyveck_pack_w1`	✅	2s	5s	-60%
`polyvecl_pointwise_acc_montgomery_c`	✅	2s	2s	+0%
`polyvecl_uniform_gamma1`	✅	2s	4s	-50%
`polyvecl_unpack_z`	✅	2s	2s	+0%
`polyz_unpack_19_native_aarch64`	✅	2s	2s	+0%
`shake128x4_absorb_once`	✅	2s	2s	+0%
`shake256`	✅	2s	2s	+0%
`shake256_absorb`	✅	2s	2s	+0%
`shake256_finalize`	✅	2s	4s	-50%
`shake256_init`	✅	2s	2s	+0%
`shake256_release`	✅	2s	3s	-33%
`shake256x4_squeezeblocks`	✅	2s	2s	+0%
`sys_check_capability`	✅	2s	2s	+0%
`unpack_sk_t0hat`	✅	2s	4s	-50%
`keccak_f1600_x1_native_aarch64`	✅	1s	1s	+0%
`mld_ct_cmask_neg_i32`	✅	1s	2s	-50%
`mld_value_barrier_u8`	✅	1s	2s	-50%
`poly_ntt`	✅	1s	2s	-50%
`poly_pointwise_montgomery`	✅	1s	1s	+0%

oqs-bot · 2026-05-12T18:51:24Z

CBMC Results (ML-DSA-87)

⚠️ Attention Required

Proof	Status	Current	Previous	Change
`TOTAL`	⚠️	3813s	1958s	+94.7%
`fqmul`	⚠️	244s	27s	+804%
`poly_pointwise_montgomery_c`	⚠️	158s	93s	+70%

Full Results (196 proofs)

Proof	Status	Current	Previous	Change
`TOTAL`	⚠️	3813s	1958s	+94.7%
`mld_invntt_2_layers_block`	✅	854s	-	new
`mld_ntt_2_layers_block`	✅	565s	-	new
`polyvecl_pointwise_acc_montgomery_c`	✅	298s	237s	+26%
`fqmul`	⚠️	244s	27s	+804%
`polyvec_matrix_expand`	✅	180s	165s	+9%
`poly_pointwise_montgomery_c`	⚠️	158s	93s	+70%
`rej_uniform_native`	✅	138s	122s	+13%
`mld_attempt_signature_generation`	✅	104s	102s	+2%
`mld_ntt_2_layers`	✅	87s	-	new
`mld_invntt_2_layers`	✅	78s	-	new
`mld_ct_memcmp`	✅	77s	76s	+1%
`sign_verify_internal`	✅	59s	56s	+5%
`sign_signature_internal`	✅	53s	54s	-2%
`polyvec_matrix_expand_serial`	✅	39s	36s	+8%
`compute_pack_t0_t1`	✅	25s	23s	+9%
`keccakf1600x4_permute_native`	✅	24s	22s	+9%
`rej_uniform`	✅	23s	16s	+44%
`polyt0_unpack`	✅	19s	13s	+46%
`polyvec_matrix_pointwise_montgomery_yvec`	✅	19s	22s	-14%
`mld_check_pct`	✅	17s	14s	+21%
`poly_chknorm_c`	✅	17s	16s	+6%
`rej_uniform_c`	✅	17s	20s	-15%
`poly_uniform_eta_4x`	✅	13s	15s	-13%
`poly_uniform_4x`	✅	12s	13s	-8%
`polyveck_decompose`	✅	12s	11s	+9%
`mld_compute_pack_z`	✅	11s	9s	+22%
`poly_add`	✅	11s	11s	+0%
`polyeta_unpack`	✅	11s	12s	-8%
`keccakf1600_permute`	✅	10s	7s	+43%
`polyveck_ntt`	✅	10s	12s	-17%
`keccak_absorb_once_x4`	✅	9s	10s	-10%
`polyveck_caddq`	✅	9s	8s	+12%
`polyveck_invntt_tomont`	✅	9s	7s	+29%
`keccakf1600_permute_native`	✅	8s	10s	-20%
`pointwise_acc_native_x86_64`	✅	8s	8s	+0%
`poly_invntt_tomont_c`	✅	8s	9s	-11%
`unpack_sk_t0hat`	✅	8s	5s	+60%
`keccak_absorb`	✅	7s	7s	+0%
`pointwise_acc_native_aarch64`	✅	7s	10s	-30%
`polyveck_chknorm`	✅	7s	5s	+40%
`polyvecl_ntt`	✅	7s	6s	+17%
`sign_pk_from_sk`	✅	7s	7s	+0%
`mld_h`	✅	6s	3s	+100%
`mld_sample_s1_s2`	✅	6s	6s	+0%
`mld_sample_s1_s2_serial`	✅	6s	4s	+50%
`poly_uniform_eta`	✅	6s	5s	+20%
`polyvecl_chknorm`	✅	6s	4s	+50%
`polyz_unpack_c`	✅	6s	6s	+0%
`sig_unpack_hints`	✅	6s	3s	+100%
`sign`	✅	6s	9s	-33%
`sign_keypair_internal`	✅	6s	5s	+20%
`sign_verify_extmu`	✅	6s	3s	+100%
`fqscale`	✅	5s	1s	+400%
`keccak_squeezeblocks_x4`	✅	5s	7s	-29%
`keccakf1600_extract_bytes (big endian)`	✅	5s	3s	+67%
`keccakf1600x4_extract_bytes`	✅	5s	2s	+150%
`mld_ct_cmask_nonzero_u32`	✅	5s	4s	+25%
`mld_prepare_domain_separation_prefix`	✅	5s	4s	+25%
`montgomery_reduce`	✅	5s	3s	+67%
`poly_caddq_native`	✅	5s	4s	+25%
`poly_chknorm_native`	✅	5s	3s	+67%
`poly_ntt`	✅	5s	4s	+25%
`poly_ntt_c`	✅	5s	2s	+150%
`poly_use_hint_c`	✅	5s	3s	+67%
`polyt0_pack`	✅	5s	5s	+0%
`polyveck_unpack_eta`	✅	5s	3s	+67%
`polyvecl_pointwise_acc_montgomery_native`	✅	5s	4s	+25%
`polyvecl_unpack_eta`	✅	5s	5s	+0%
`polyvecl_unpack_z`	✅	5s	3s	+67%
`shake128_release`	✅	5s	1s	+400%
`shake256`	✅	5s	2s	+150%
`sign_verify_pre_hash_shake256`	✅	5s	4s	+25%
`caddq`	✅	4s	3s	+33%
`decompose`	✅	4s	5s	-20%
`intt_native_aarch64`	✅	4s	6s	-33%
`keccak_f1600_x4_native_aarch64_v8a_scalar_hybrid`	✅	4s	1s	+300%
`keccak_init`	✅	4s	2s	+100%
`keccakf1600_xor_bytes`	✅	4s	5s	-20%
`keccakf1600x4_extract_bytes_native`	✅	4s	3s	+33%
`keccakf1600x4_xor_bytes`	✅	4s	2s	+100%
`mld_ct_cmask_nonzero_u8`	✅	4s	2s	+100%
`mld_polymat_expand_entry`	✅	4s	2s	+100%
`ntt_native_aarch64`	✅	4s	3s	+33%
`ntt_native_x86_64`	✅	4s	4s	+0%
`nttunpack_native_x86_64`	✅	4s	2s	+100%
`pack_sig_c`	✅	4s	1s	+300%
`pack_sig_z`	✅	4s	2s	+100%
`pack_sk_rho_key_tr_s2`	✅	4s	4s	+0%
`pointwise_native_aarch64`	✅	4s	3s	+33%
`pointwise_native_x86_64`	✅	4s	3s	+33%
`poly_caddq_c`	✅	4s	6s	-33%
`poly_caddq_native_aarch64`	✅	4s	5s	-20%
`poly_decompose_32_native_aarch64`	✅	4s	1s	+300%
`poly_decompose_c`	✅	4s	5s	-20%
`poly_invntt_tomont_native`	✅	4s	4s	+0%
`poly_permute_bitrev_to_custom_optional`	✅	4s	3s	+33%
`poly_sub`	✅	4s	1s	+300%
`poly_uniform`	✅	4s	5s	-20%
`poly_uniform_gamma1`	✅	4s	2s	+100%
`poly_use_hint_native_aarch64`	✅	4s	2s	+100%
`polyveck_pack_w1`	✅	4s	4s	+0%
`polyvecl_pointwise_acc_montgomery`	✅	4s	6s	-33%
`polyvecl_uniform_gamma1`	✅	4s	3s	+33%
`polyvecl_uniform_gamma1_serial`	✅	4s	3s	+33%
`polyw1_pack`	✅	4s	5s	-20%
`polyz_unpack`	✅	4s	3s	+33%
`polyz_unpack_17_native_aarch64`	✅	4s	2s	+100%
`polyz_unpack_native`	✅	4s	2s	+100%
`power2round`	✅	4s	4s	+0%
`rej_eta_c`	✅	4s	4s	+0%
`shake128_init`	✅	4s	1s	+300%
`shake128x4_absorb_once`	✅	4s	3s	+33%
`shake256_init`	✅	4s	3s	+33%
`sign_keypair`	✅	4s	5s	-20%
`sign_open`	✅	4s	6s	-33%
`sign_signature_extmu`	✅	4s	3s	+33%
`sign_signature_pre_hash_internal`	✅	4s	6s	-33%
`sk_s1hat_get_poly`	✅	4s	2s	+100%
`sk_s2hat_get_poly`	✅	4s	3s	+33%
`unpack_pk_t1`	✅	4s	2s	+100%
`unpack_sk`	✅	4s	3s	+33%
`keccak_f1600_x4_native_avx2`	✅	3s	2s	+50%
`keccak_finalize`	✅	3s	3s	+0%
`keccak_squeeze`	✅	3s	3s	+0%
`make_hint`	✅	3s	3s	+0%
`pack_sig_h`	✅	3s	5s	-40%
`pack_sk_s1`	✅	3s	3s	+0%
`poly_caddq`	✅	3s	4s	-25%
`poly_challenge`	✅	3s	5s	-40%
`poly_chknorm_native_aarch64`	✅	3s	4s	-25%
`poly_decompose_native`	✅	3s	3s	+0%
`poly_pointwise_montgomery_native`	✅	3s	3s	+0%
`poly_reduce`	✅	3s	3s	+0%
`poly_shiftl`	✅	3s	4s	-25%
`poly_use_hint`	✅	3s	4s	-25%
`poly_use_hint_native`	✅	3s	3s	+0%
`polyt1_pack`	✅	3s	3s	+0%
`polyt1_unpack`	✅	3s	3s	+0%
`reduce32`	✅	3s	4s	-25%
`shake128_absorb`	✅	3s	3s	+0%
`shake128x4_squeezeblocks`	✅	3s	2s	+50%
`shake256_absorb`	✅	3s	2s	+50%
`shake256_release`	✅	3s	2s	+50%
`shake256x4_absorb_once`	✅	3s	3s	+0%
`shake256x4_squeezeblocks`	✅	3s	3s	+0%
`sign_signature`	✅	3s	3s	+0%
`sign_signature_pre_hash_shake256`	✅	3s	3s	+0%
`sign_verify`	✅	3s	7s	-57%
`sign_verify_pre_hash_internal`	✅	3s	4s	-25%
`sk_t0hat_get_poly`	✅	3s	2s	+50%
`sys_check_capability`	✅	3s	2s	+50%
`unpack_sk_s1hat`	✅	3s	2s	+50%
`unpack_sk_s2hat`	✅	3s	3s	+0%
`intt_native_x86_64`	✅	2s	2s	+0%
`keccak_f1600_x1_native_aarch64`	✅	2s	3s	-33%
`keccak_f1600_x1_native_aarch64_v84a`	✅	2s	3s	-33%
`keccak_f1600_x4_native_aarch64_v84a`	✅	2s	2s	+0%
`keccak_f1600_x4_native_aarch64_v8a_v84a_scalar_hybrid`	✅	2s	2s	+0%
`keccakf1600x4_permute`	✅	2s	4s	-50%
`keccakf1600x4_xor_bytes_native`	✅	2s	5s	-60%
`mld_ct_abs_i32`	✅	2s	4s	-50%
`mld_ct_cmask_neg_i32`	✅	2s	2s	+0%
`mld_ct_get_optblocker_i64`	✅	2s	1s	+100%
`mld_ct_get_optblocker_u32`	✅	2s	1s	+100%
`mld_ct_get_optblocker_u8`	✅	2s	4s	-50%
`mld_ct_sel_int32`	✅	2s	3s	-33%
`mld_value_barrier_i64`	✅	2s	2s	+0%
`mld_value_barrier_u32`	✅	2s	3s	-33%
`mld_value_barrier_u8`	✅	2s	6s	-67%
`poly_chknorm`	✅	2s	3s	-33%
`poly_decompose`	✅	2s	3s	-33%
`poly_decompose_88_native_aarch64`	✅	2s	3s	-33%
`poly_invntt_tomont`	✅	2s	2s	+0%
`poly_ntt_native`	✅	2s	5s	-60%
`poly_permute_bitrev_to_custom_optional_native`	✅	2s	4s	-50%
`poly_pointwise_montgomery`	✅	2s	4s	-50%
`poly_power2round`	✅	2s	8s	-75%
`poly_uniform_gamma1_4x`	✅	2s	6s	-67%
`polyvec_matrix_pointwise_montgomery_row`	✅	2s	2s	+0%
`polyveck_pack_eta`	✅	2s	3s	-33%
`polyveck_reduce`	✅	2s	6s	-67%
`polyvecl_pack_eta`	✅	2s	2s	+0%
`polyz_pack`	✅	2s	2s	+0%
`polyz_unpack_19_native_aarch64`	✅	2s	3s	-33%
`rej_eta`	✅	2s	3s	-33%
`rej_eta_native`	✅	2s	4s	-50%
`shake128_finalize`	✅	2s	1s	+100%
`shake128_squeeze`	✅	2s	2s	+0%
`shake256_finalize`	✅	2s	2s	+0%
`use_hint`	✅	2s	3s	-33%
`yvec_get_poly`	✅	2s	3s	-33%
`yvec_init`	✅	2s	3s	-33%
`keccakf1600_xor_bytes (big endian)`	✅	1s	2s	-50%
`mld_keccakf1600_extract_bytes`	✅	1s	2s	-50%
`polyeta_pack`	✅	1s	4s	-75%
`shake256_squeeze`	✅	1s	4s	-75%

oqs-bot · 2026-05-12T18:51:27Z

CBMC Results (ML-DSA-65)

⚠️ Attention Required

Proof	Status	Current	Previous	Change
`TOTAL`	⚠️	3629s	1912s	+89.8%
`fqmul`	⚠️	237s	28s	+746%
`poly_pointwise_montgomery_c`	⚠️	159s	95s	+67%

Full Results (196 proofs)

Proof	Status	Current	Previous	Change
`TOTAL`	⚠️	3629s	1912s	+89.8%
`mld_invntt_2_layers_block`	✅	841s	-	new
`mld_ntt_2_layers_block`	✅	539s	-	new
`polyvecl_pointwise_acc_montgomery_c`	✅	315s	286s	+10%
`fqmul`	⚠️	237s	28s	+746%
`poly_pointwise_montgomery_c`	⚠️	159s	95s	+67%
`polyvec_matrix_expand`	✅	147s	146s	+1%
`rej_uniform_native`	✅	136s	124s	+10%
`mld_ntt_2_layers`	✅	94s	-	new
`mld_ct_memcmp`	✅	78s	73s	+7%
`mld_invntt_2_layers`	✅	72s	-	new
`mld_attempt_signature_generation`	✅	67s	71s	-6%
`sign_verify_internal`	✅	50s	55s	-9%
`sign_signature_internal`	✅	49s	45s	+9%
`polyvec_matrix_pointwise_montgomery_yvec`	✅	28s	29s	-3%
`polyvec_matrix_expand_serial`	✅	25s	25s	+0%
`keccakf1600x4_permute_native`	✅	22s	23s	-4%
`rej_uniform`	✅	21s	16s	+31%
`polyt0_unpack`	✅	17s	15s	+13%
`compute_pack_t0_t1`	✅	15s	13s	+15%
`poly_uniform_eta_4x`	✅	15s	13s	+15%
`polyveck_decompose`	✅	14s	14s	+0%
`rej_uniform_c`	✅	14s	21s	-33%
`poly_uniform_4x`	✅	13s	10s	+30%
`mld_check_pct`	✅	12s	13s	-8%
`poly_chknorm_c`	✅	12s	17s	-29%
`poly_add`	✅	11s	11s	+0%
`polyvecl_ntt`	✅	11s	8s	+38%
`keccakf1600_permute_native`	✅	10s	9s	+11%
`poly_decompose_c`	✅	10s	6s	+67%
`polyveck_chknorm`	✅	10s	10s	+0%
`polyveck_caddq`	✅	9s	7s	+29%
`polyveck_invntt_tomont`	✅	9s	9s	+0%
`keccak_absorb_once_x4`	✅	8s	9s	-11%
`mld_compute_pack_z`	✅	8s	9s	-11%
`poly_invntt_tomont_c`	✅	8s	10s	-20%
`poly_ntt_native`	✅	7s	2s	+250%
`polyvecl_chknorm`	✅	7s	3s	+133%
`polyvecl_uniform_gamma1_serial`	✅	7s	3s	+133%
`sign`	✅	7s	7s	+0%
`unpack_sk`	✅	7s	3s	+133%
`keccak_absorb`	✅	6s	8s	-25%
`keccak_squeezeblocks_x4`	✅	6s	4s	+50%
`keccakf1600_permute`	✅	6s	9s	-33%
`mld_h`	✅	6s	4s	+50%
`poly_caddq_c`	✅	6s	4s	+50%
`poly_chknorm`	✅	6s	1s	+500%
`poly_pointwise_montgomery_native`	✅	6s	4s	+50%
`polyveck_ntt`	✅	6s	7s	-14%
`sign_pk_from_sk`	✅	6s	5s	+20%
`sign_verify`	✅	6s	4s	+50%
`pointwise_acc_native_x86_64`	✅	5s	7s	-29%
`poly_caddq_native_aarch64`	✅	5s	2s	+150%
`poly_invntt_tomont_native`	✅	5s	2s	+150%
`poly_permute_bitrev_to_custom_optional`	✅	5s	3s	+67%
`poly_uniform_gamma1`	✅	5s	3s	+67%
`polyveck_unpack_eta`	✅	5s	3s	+67%
`sig_unpack_hints`	✅	5s	4s	+25%
`sign_keypair_internal`	✅	5s	5s	+0%
`fqscale`	✅	4s	2s	+100%
`intt_native_aarch64`	✅	4s	4s	+0%
`intt_native_x86_64`	✅	4s	3s	+33%
`keccakf1600x4_xor_bytes_native`	✅	4s	4s	+0%
`mld_ct_get_optblocker_i64`	✅	4s	4s	+0%
`mld_sample_s1_s2`	✅	4s	3s	+33%
`mld_sample_s1_s2_serial`	✅	4s	6s	-33%
`ntt_native_aarch64`	✅	4s	4s	+0%
`nttunpack_native_x86_64`	✅	4s	4s	+0%
`pack_sig_c`	✅	4s	4s	+0%
`pack_sig_h`	✅	4s	3s	+33%
`pack_sk_s1`	✅	4s	3s	+33%
`pointwise_acc_native_aarch64`	✅	4s	6s	-33%
`poly_chknorm_native_aarch64`	✅	4s	2s	+100%
`poly_decompose_88_native_aarch64`	✅	4s	4s	+0%
`poly_decompose_native`	✅	4s	3s	+33%
`poly_invntt_tomont`	✅	4s	4s	+0%
`poly_shiftl`	✅	4s	6s	-33%
`poly_uniform`	✅	4s	2s	+100%
`poly_uniform_gamma1_4x`	✅	4s	3s	+33%
`poly_use_hint_native`	✅	4s	2s	+100%
`poly_use_hint_native_aarch64`	✅	4s	4s	+0%
`polyeta_pack`	✅	4s	3s	+33%
`polyt0_pack`	✅	4s	3s	+33%
`polyt1_pack`	✅	4s	5s	-20%
`polyveck_reduce`	✅	4s	3s	+33%
`polyvecl_pack_eta`	✅	4s	2s	+100%
`polyvecl_pointwise_acc_montgomery_native`	✅	4s	2s	+100%
`polyvecl_uniform_gamma1`	✅	4s	3s	+33%
`polyvecl_unpack_eta`	✅	4s	5s	-20%
`polyz_unpack_c`	✅	4s	7s	-43%
`shake256_absorb`	✅	4s	4s	+0%
`shake256x4_absorb_once`	✅	4s	2s	+100%
`sign_keypair`	✅	4s	5s	-20%
`sign_open`	✅	4s	4s	+0%
`sign_signature_pre_hash_internal`	✅	4s	5s	-20%
`sign_verify_pre_hash_internal`	✅	4s	6s	-33%
`sign_verify_pre_hash_shake256`	✅	4s	3s	+33%
`sys_check_capability`	✅	4s	2s	+100%
`unpack_pk_t1`	✅	4s	3s	+33%
`unpack_sk_s1hat`	✅	4s	3s	+33%
`unpack_sk_t0hat`	✅	4s	5s	-20%
`yvec_init`	✅	4s	3s	+33%
`decompose`	✅	3s	4s	-25%
`keccak_f1600_x4_native_aarch64_v8a_scalar_hybrid`	✅	3s	2s	+50%
`keccak_f1600_x4_native_aarch64_v8a_v84a_scalar_hybrid`	✅	3s	2s	+50%
`keccak_finalize`	✅	3s	3s	+0%
`keccakf1600_extract_bytes (big endian)`	✅	3s	2s	+50%
`keccakf1600x4_extract_bytes`	✅	3s	2s	+50%
`keccakf1600x4_extract_bytes_native`	✅	3s	2s	+50%
`keccakf1600x4_permute`	✅	3s	3s	+0%
`make_hint`	✅	3s	2s	+50%
`mld_ct_abs_i32`	✅	3s	2s	+50%
`mld_ct_cmask_nonzero_u8`	✅	3s	2s	+50%
`mld_ct_get_optblocker_u8`	✅	3s	3s	+0%
`mld_value_barrier_i64`	✅	3s	4s	-25%
`mld_value_barrier_u32`	✅	3s	4s	-25%
`mld_value_barrier_u8`	✅	3s	3s	+0%
`ntt_native_x86_64`	✅	3s	3s	+0%
`pack_sig_z`	✅	3s	2s	+50%
`pointwise_native_x86_64`	✅	3s	4s	-25%
`poly_caddq`	✅	3s	4s	-25%
`poly_caddq_native`	✅	3s	2s	+50%
`poly_challenge`	✅	3s	3s	+0%
`poly_chknorm_native`	✅	3s	4s	-25%
`poly_ntt`	✅	3s	4s	-25%
`poly_ntt_c`	✅	3s	4s	-25%
`poly_pointwise_montgomery`	✅	3s	3s	+0%
`poly_power2round`	✅	3s	8s	-62%
`poly_sub`	✅	3s	3s	+0%
`poly_use_hint`	✅	3s	3s	+0%
`polyeta_unpack`	✅	3s	5s	-40%
`polyvec_matrix_pointwise_montgomery_row`	✅	3s	2s	+50%
`polyvecl_pointwise_acc_montgomery`	✅	3s	3s	+0%
`polyvecl_unpack_z`	✅	3s	4s	-25%
`polyz_pack`	✅	3s	3s	+0%
`polyz_unpack`	✅	3s	2s	+50%
`polyz_unpack_19_native_aarch64`	✅	3s	2s	+50%
`reduce32`	✅	3s	1s	+200%
`rej_eta`	✅	3s	4s	-25%
`rej_eta_native`	✅	3s	3s	+0%
`shake128_finalize`	✅	3s	2s	+50%
`shake128_squeeze`	✅	3s	2s	+50%
`shake256_finalize`	✅	3s	3s	+0%
`shake256_init`	✅	3s	3s	+0%
`shake256_release`	✅	3s	2s	+50%
`sign_signature`	✅	3s	6s	-50%
`sign_signature_extmu`	✅	3s	5s	-40%
`sign_verify_extmu`	✅	3s	6s	-50%
`sk_s1hat_get_poly`	✅	3s	2s	+50%
`sk_s2hat_get_poly`	✅	3s	2s	+50%
`sk_t0hat_get_poly`	✅	3s	4s	-25%
`unpack_sk_s2hat`	✅	3s	3s	+0%
`use_hint`	✅	3s	3s	+0%
`caddq`	✅	2s	6s	-67%
`keccak_f1600_x1_native_aarch64`	✅	2s	3s	-33%
`keccak_f1600_x1_native_aarch64_v84a`	✅	2s	5s	-60%
`keccak_f1600_x4_native_aarch64_v84a`	✅	2s	4s	-50%
`keccak_init`	✅	2s	4s	-50%
`keccak_squeeze`	✅	2s	2s	+0%
`keccakf1600_xor_bytes`	✅	2s	2s	+0%
`keccakf1600_xor_bytes (big endian)`	✅	2s	2s	+0%
`keccakf1600x4_xor_bytes`	✅	2s	3s	-33%
`mld_ct_cmask_neg_i32`	✅	2s	2s	+0%
`mld_ct_cmask_nonzero_u32`	✅	2s	3s	-33%
`mld_ct_get_optblocker_u32`	✅	2s	1s	+100%
`mld_ct_sel_int32`	✅	2s	1s	+100%
`mld_keccakf1600_extract_bytes`	✅	2s	3s	-33%
`mld_prepare_domain_separation_prefix`	✅	2s	3s	-33%
`montgomery_reduce`	✅	2s	3s	-33%
`pack_sk_rho_key_tr_s2`	✅	2s	3s	-33%
`pointwise_native_aarch64`	✅	2s	1s	+100%
`poly_decompose`	✅	2s	3s	-33%
`poly_decompose_32_native_aarch64`	✅	2s	2s	+0%
`poly_permute_bitrev_to_custom_optional_native`	✅	2s	2s	+0%
`poly_uniform_eta`	✅	2s	3s	-33%
`poly_use_hint_c`	✅	2s	5s	-60%
`polyt1_unpack`	✅	2s	2s	+0%
`polyveck_pack_eta`	✅	2s	4s	-50%
`polyveck_pack_w1`	✅	2s	3s	-33%
`polyw1_pack`	✅	2s	3s	-33%
`polyz_unpack_17_native_aarch64`	✅	2s	4s	-50%
`power2round`	✅	2s	3s	-33%
`rej_eta_c`	✅	2s	3s	-33%
`shake128_absorb`	✅	2s	4s	-50%
`shake128_init`	✅	2s	4s	-50%
`shake128_release`	✅	2s	1s	+100%
`shake128x4_absorb_once`	✅	2s	1s	+100%
`shake256`	✅	2s	3s	-33%
`shake256_squeeze`	✅	2s	4s	-50%
`shake256x4_squeezeblocks`	✅	2s	2s	+0%
`sign_signature_pre_hash_shake256`	✅	2s	4s	-50%
`yvec_get_poly`	✅	2s	2s	+0%
`keccak_f1600_x4_native_avx2`	✅	1s	2s	-50%
`mld_polymat_expand_entry`	✅	1s	4s	-75%
`poly_reduce`	✅	1s	4s	-75%
`polyz_unpack_native`	✅	1s	2s	-50%
`shake128x4_squeezeblocks`	✅	1s	2s	-50%

oqs-bot · 2026-05-12T18:51:33Z

CBMC Results (ML-DSA-44)

⚠️ Attention Required

Proof	Status	Current	Previous	Change
`TOTAL`	⚠️	3595s	1601s	+124.5%
`fqmul`	⚠️	251s	27s	+830%
`poly_pointwise_montgomery_c`	⚠️	172s	86s	+100%

Full Results (196 proofs)

Proof	Status	Current	Previous	Change
`TOTAL`	⚠️	3595s	1601s	+124.5%
`mld_invntt_2_layers_block`	✅	915s	-	new
`mld_ntt_2_layers_block`	✅	577s	-	new
`polyvecl_pointwise_acc_montgomery_c`	✅	284s	237s	+20%
`fqmul`	⚠️	251s	27s	+830%
`poly_pointwise_montgomery_c`	⚠️	172s	86s	+100%
`rej_uniform_native`	✅	141s	115s	+23%
`mld_ntt_2_layers`	✅	93s	-	new
`mld_ct_memcmp`	✅	89s	72s	+24%
`mld_invntt_2_layers`	✅	80s	-	new
`mld_attempt_signature_generation`	✅	58s	54s	+7%
`sign_verify_internal`	✅	32s	30s	+7%
`sign_signature_internal`	✅	31s	28s	+11%
`polyvec_matrix_expand`	✅	29s	27s	+7%
`keccakf1600x4_permute_native`	✅	24s	23s	+4%
`rej_uniform`	✅	22s	17s	+29%
`polyt0_unpack`	✅	20s	15s	+33%
`polyvecl_chknorm`	✅	17s	16s	+6%
`rej_uniform_c`	✅	16s	16s	+0%
`compute_pack_t0_t1`	✅	15s	15s	+0%
`poly_chknorm_c`	✅	15s	15s	+0%
`poly_uniform_4x`	✅	15s	11s	+36%
`poly_uniform_eta_4x`	✅	15s	14s	+7%
`polyeta_unpack`	✅	15s	11s	+36%
`polyvec_matrix_pointwise_montgomery_yvec`	✅	14s	14s	+0%
`polyz_unpack_c`	✅	14s	13s	+8%
`polyvec_matrix_expand_serial`	✅	13s	10s	+30%
`poly_add`	✅	12s	13s	-8%
`mld_check_pct`	✅	11s	15s	-27%
`poly_invntt_tomont_c`	✅	11s	6s	+83%
`keccak_absorb_once_x4`	✅	10s	9s	+11%
`mld_compute_pack_z`	✅	8s	8s	+0%
`poly_decompose_c`	✅	8s	6s	+33%
`keccak_absorb`	✅	7s	6s	+17%
`keccakf1600_permute_native`	✅	7s	7s	+0%
`polyveck_caddq`	✅	7s	5s	+40%
`sign`	✅	7s	7s	+0%
`sign_keypair_internal`	✅	7s	3s	+133%
`keccakf1600x4_xor_bytes_native`	✅	6s	4s	+50%
`poly_decompose_32_native_aarch64`	✅	6s	3s	+100%
`poly_ntt_native`	✅	6s	2s	+200%
`poly_uniform_eta`	✅	6s	4s	+50%
`poly_use_hint_c`	✅	6s	4s	+50%
`polyveck_decompose`	✅	6s	7s	-14%
`shake256_squeeze`	✅	6s	2s	+200%
`sign_pk_from_sk`	✅	6s	6s	+0%
`sign_signature_pre_hash_shake256`	✅	6s	6s	+0%
`intt_native_x86_64`	✅	5s	2s	+150%
`keccak_squeezeblocks_x4`	✅	5s	5s	+0%
`keccakf1600_permute`	✅	5s	7s	-29%
`mld_ct_get_optblocker_i64`	✅	5s	4s	+25%
`mld_ct_get_optblocker_u8`	✅	5s	3s	+67%
`mld_h`	✅	5s	3s	+67%
`mld_value_barrier_i64`	✅	5s	2s	+150%
`montgomery_reduce`	✅	5s	2s	+150%
`ntt_native_aarch64`	✅	5s	5s	+0%
`pointwise_acc_native_aarch64`	✅	5s	5s	+0%
`pointwise_acc_native_x86_64`	✅	5s	6s	-17%
`poly_caddq_c`	✅	5s	4s	+25%
`poly_chknorm_native`	✅	5s	3s	+67%
`poly_invntt_tomont_native`	✅	5s	3s	+67%
`polyt0_pack`	✅	5s	4s	+25%
`polyveck_ntt`	✅	5s	5s	+0%
`polyvecl_ntt`	✅	5s	6s	-17%
`shake128x4_squeezeblocks`	✅	5s	3s	+67%
`shake256x4_squeezeblocks`	✅	5s	4s	+25%
`sign_open`	✅	5s	3s	+67%
`sign_signature_extmu`	✅	5s	4s	+25%
`unpack_pk_t1`	✅	5s	2s	+150%
`caddq`	✅	4s	3s	+33%
`decompose`	✅	4s	1s	+300%
`keccak_init`	✅	4s	3s	+33%
`keccakf1600x4_extract_bytes_native`	✅	4s	1s	+300%
`mld_keccakf1600_extract_bytes`	✅	4s	3s	+33%
`mld_prepare_domain_separation_prefix`	✅	4s	3s	+33%
`mld_sample_s1_s2`	✅	4s	2s	+100%
`pack_sig_h`	✅	4s	2s	+100%
`pack_sig_z`	✅	4s	3s	+33%
`pack_sk_s1`	✅	4s	3s	+33%
`pointwise_native_aarch64`	✅	4s	3s	+33%
`poly_caddq`	✅	4s	3s	+33%
`poly_caddq_native_aarch64`	✅	4s	4s	+0%
`poly_challenge`	✅	4s	7s	-43%
`poly_decompose_native`	✅	4s	5s	-20%
`poly_invntt_tomont`	✅	4s	2s	+100%
`poly_ntt_c`	✅	4s	3s	+33%
`poly_pointwise_montgomery_native`	✅	4s	3s	+33%
`poly_power2round`	✅	4s	6s	-33%
`poly_shiftl`	✅	4s	9s	-56%
`poly_uniform`	✅	4s	4s	+0%
`poly_use_hint`	✅	4s	4s	+0%
`polyveck_invntt_tomont`	✅	4s	4s	+0%
`polyveck_pack_w1`	✅	4s	3s	+33%
`polyvecl_pointwise_acc_montgomery`	✅	4s	3s	+33%
`polyvecl_pointwise_acc_montgomery_native`	✅	4s	2s	+100%
`polyw1_pack`	✅	4s	4s	+0%
`polyz_pack`	✅	4s	3s	+33%
`polyz_unpack_17_native_aarch64`	✅	4s	4s	+0%
`polyz_unpack_native`	✅	4s	2s	+100%
`reduce32`	✅	4s	3s	+33%
`rej_eta_native`	✅	4s	6s	-33%
`shake128_squeeze`	✅	4s	2s	+100%
`shake256_init`	✅	4s	2s	+100%
`sign_signature`	✅	4s	4s	+0%
`sign_signature_pre_hash_internal`	✅	4s	6s	-33%
`sign_verify_extmu`	✅	4s	3s	+33%
`sign_verify_pre_hash_internal`	✅	4s	4s	+0%
`use_hint`	✅	4s	2s	+100%
`fqscale`	✅	3s	2s	+50%
`keccakf1600_extract_bytes (big endian)`	✅	3s	3s	+0%
`keccakf1600x4_extract_bytes`	✅	3s	5s	-40%
`make_hint`	✅	3s	3s	+0%
`mld_ct_cmask_neg_i32`	✅	3s	1s	+200%
`mld_ct_get_optblocker_u32`	✅	3s	1s	+200%
`mld_ct_sel_int32`	✅	3s	4s	-25%
`mld_polymat_expand_entry`	✅	3s	1s	+200%
`pack_sig_c`	✅	3s	3s	+0%
`pack_sk_rho_key_tr_s2`	✅	3s	5s	-40%
`pointwise_native_x86_64`	✅	3s	3s	+0%
`poly_caddq_native`	✅	3s	3s	+0%
`poly_decompose`	✅	3s	3s	+0%
`poly_ntt`	✅	3s	3s	+0%
`poly_permute_bitrev_to_custom_optional`	✅	3s	1s	+200%
`poly_permute_bitrev_to_custom_optional_native`	✅	3s	2s	+50%
`poly_pointwise_montgomery`	✅	3s	4s	-25%
`poly_sub`	✅	3s	4s	-25%
`poly_uniform_gamma1_4x`	✅	3s	3s	+0%
`poly_use_hint_native_aarch64`	✅	3s	2s	+50%
`polyeta_pack`	✅	3s	5s	-40%
`polyvec_matrix_pointwise_montgomery_row`	✅	3s	2s	+50%
`polyveck_reduce`	✅	3s	2s	+50%
`polyveck_unpack_eta`	✅	3s	2s	+50%
`polyvecl_uniform_gamma1`	✅	3s	4s	-25%
`polyvecl_unpack_eta`	✅	3s	4s	-25%
`polyz_unpack`	✅	3s	4s	-25%
`power2round`	✅	3s	4s	-25%
`rej_eta_c`	✅	3s	3s	+0%
`shake128_absorb`	✅	3s	1s	+200%
`shake256`	✅	3s	2s	+50%
`shake256_absorb`	✅	3s	1s	+200%
`shake256_finalize`	✅	3s	1s	+200%
`shake256_release`	✅	3s	4s	-25%
`shake256x4_absorb_once`	✅	3s	3s	+0%
`sign_keypair`	✅	3s	4s	-25%
`sign_verify`	✅	3s	2s	+50%
`sign_verify_pre_hash_shake256`	✅	3s	3s	+0%
`sk_s1hat_get_poly`	✅	3s	3s	+0%
`sk_s2hat_get_poly`	✅	3s	3s	+0%
`unpack_sk`	✅	3s	3s	+0%
`unpack_sk_s1hat`	✅	3s	2s	+50%
`unpack_sk_t0hat`	✅	3s	3s	+0%
`yvec_init`	✅	3s	2s	+50%
`intt_native_aarch64`	✅	2s	2s	+0%
`keccak_f1600_x1_native_aarch64`	✅	2s	3s	-33%
`keccak_f1600_x1_native_aarch64_v84a`	✅	2s	3s	-33%
`keccak_f1600_x4_native_aarch64_v84a`	✅	2s	2s	+0%
`keccak_f1600_x4_native_avx2`	✅	2s	4s	-50%
`keccak_finalize`	✅	2s	1s	+100%
`keccak_squeeze`	✅	2s	2s	+0%
`keccakf1600x4_permute`	✅	2s	3s	-33%
`keccakf1600x4_xor_bytes`	✅	2s	3s	-33%
`mld_ct_abs_i32`	✅	2s	4s	-50%
`mld_ct_cmask_nonzero_u32`	✅	2s	3s	-33%
`mld_ct_cmask_nonzero_u8`	✅	2s	3s	-33%
`mld_sample_s1_s2_serial`	✅	2s	5s	-60%
`mld_value_barrier_u8`	✅	2s	2s	+0%
`ntt_native_x86_64`	✅	2s	5s	-60%
`nttunpack_native_x86_64`	✅	2s	4s	-50%
`poly_chknorm`	✅	2s	3s	-33%
`poly_chknorm_native_aarch64`	✅	2s	3s	-33%
`poly_decompose_88_native_aarch64`	✅	2s	2s	+0%
`poly_reduce`	✅	2s	3s	-33%
`poly_uniform_gamma1`	✅	2s	1s	+100%
`poly_use_hint_native`	✅	2s	2s	+0%
`polyt1_pack`	✅	2s	4s	-50%
`polyt1_unpack`	✅	2s	2s	+0%
`polyveck_chknorm`	✅	2s	7s	-71%
`polyveck_pack_eta`	✅	2s	1s	+100%
`polyvecl_pack_eta`	✅	2s	4s	-50%
`polyvecl_uniform_gamma1_serial`	✅	2s	2s	+0%
`polyvecl_unpack_z`	✅	2s	1s	+100%
`polyz_unpack_19_native_aarch64`	✅	2s	5s	-60%
`rej_eta`	✅	2s	3s	-33%
`shake128_init`	✅	2s	2s	+0%
`shake128x4_absorb_once`	✅	2s	2s	+0%
`sig_unpack_hints`	✅	2s	2s	+0%
`sys_check_capability`	✅	2s	2s	+0%
`yvec_get_poly`	✅	2s	4s	-50%
`keccak_f1600_x4_native_aarch64_v8a_scalar_hybrid`	✅	1s	2s	-50%
`keccak_f1600_x4_native_aarch64_v8a_v84a_scalar_hybrid`	✅	1s	1s	+0%
`keccakf1600_xor_bytes`	✅	1s	3s	-67%
`keccakf1600_xor_bytes (big endian)`	✅	1s	6s	-83%
`mld_value_barrier_u32`	✅	1s	2s	-50%
`shake128_finalize`	✅	1s	2s	-50%
`shake128_release`	✅	1s	1s	+0%
`sk_t0hat_get_poly`	✅	1s	1s	+0%
`unpack_sk_s2hat`	✅	1s	4s	-75%

github-actions

⚠️ Performance Alert ⚠️

Possible performance regression was detected for benchmark 'Arm Cortex-A55 (Snapdragon 888) benchmarks (opt)'.
Benchmark result of this commit is worse than the previous benchmark result exceeding threshold 1.03.

Benchmark suite	Current: `38f3f91`	Previous: `a71b5d2`	Ratio
`ML-DSA-65 sign`	`1367463` cycles	`1320861` cycles	`1.04`

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

Replace the single-layer C-reference forward and inverse NTT in `mldsa/src/poly.c` with one that merges two layers each. Also, store each twiddle alongside its precomputed twist, letting `mld_fqmul(a, b, b_twisted)` drop the multiply with MLDSA_Q^{-1} that was previously hidden inside `mld_montgomery_reduce`. Mirrors pq-code-package/mlkem-native/#463 and pq-code/package/mlkem-native/#683 Signed-off-by: Hanno Becker <beckphan@amazon.co.uk>

rod-chapman

Look good. 1 suggestion to improve proof times.

rod-chapman · 2026-05-13T09:22:37Z


 # Disable any setting of EXTERNAL_SAT_SOLVER, and choose SMT backend instead
 EXTERNAL_SAT_SOLVER=
 CBMCFLAGS=--bitwuzla


On my laptop, proof of this new implementation takes 131s with bitwuzla, so I tried z3, which compltes the proof in about 26s. Suggest switch to CBMCFLAGS=--smt2

hanno-becker changed the title ~~[WIP] Speed up C-reference NTT/invNTT with twisted zetas + 2-layer me…~~ [WIP] Speed up C-reference NTT/invNTT with twisted zetas + 2-layer merging May 12, 2026

hanno-becker force-pushed the c_ntt_2 branch from 601a861 to 2eaffea Compare May 12, 2026 18:23

hanno-becker added the benchmark label May 12, 2026