AVX2/AArch64: Add native `poly_pointwise_montgomery` #346

mkannwischer · 2025-07-02T03:58:19Z

Resolves AArch64: Add poly_pointwise_montgomery assembly #336
Resolves AVX2: Add poly_pointwise_montgomery assembly #337

This hoists out the native pointwise multiplication from #334 and also adds similar code for AVX2.
Note that it does not yet modify the matrix-vector multiplication - that one is hence not using this code yet.
That will be done in a follow-up PR which will make it easier to understand the performance impact.

oqs-bot

Graviton4

Benchmark suite	Current: `1fcd573`	Previous: `2c8f312`	Ratio
`ML-DSA-44 keypair`	`77278` cycles	`77359` cycles	`1.00`
`ML-DSA-44 sign`	`238351` cycles	`238702` cycles	`1.00`
`ML-DSA-44 verify`	`84670` cycles	`84808` cycles	`1.00`
`ML-DSA-65 keypair`	`137276` cycles	`137174` cycles	`1.00`
`ML-DSA-65 sign`	`385184` cycles	`386156` cycles	`1.00`
`ML-DSA-65 verify`	`137648` cycles	`137679` cycles	`1.00`
`ML-DSA-87 keypair`	`221254` cycles	`221386` cycles	`1.00`
`ML-DSA-87 sign`	`493976` cycles	`493662` cycles	`1.00`
`ML-DSA-87 verify`	`224032` cycles	`224278` cycles	`1.00`

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

oqs-bot

Graviton4 (no-opt)

Benchmark suite	Current: `1fcd573`	Previous: `2c8f312`	Ratio
`ML-DSA-44 keypair`	`131615` cycles	`131443` cycles	`1.00`
`ML-DSA-44 sign`	`455861` cycles	`455906` cycles	`1.00`
`ML-DSA-44 verify`	`151182` cycles	`142900` cycles	`1.06`
`ML-DSA-65 keypair`	`223983` cycles	`225008` cycles	`1.00`
`ML-DSA-65 sign`	`733346` cycles	`733460` cycles	`1.00`
`ML-DSA-65 verify`	`227150` cycles	`226945` cycles	`1.00`
`ML-DSA-87 keypair`	`370115` cycles	`370197` cycles	`1.00`
`ML-DSA-87 sign`	`938333` cycles	`938660` cycles	`1.00`
`ML-DSA-87 verify`	`376843` cycles	`377254` cycles	`1.00`

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

oqs-bot

Graviton2

Benchmark suite	Current: `1fcd573`	Previous: `2c8f312`	Ratio
`ML-DSA-44 keypair`	`128740` cycles	`129013` cycles	`1.00`
`ML-DSA-44 sign`	`418832` cycles	`420050` cycles	`1.00`
`ML-DSA-44 verify`	`142390` cycles	`142767` cycles	`1.00`
`ML-DSA-65 keypair`	`239992` cycles	`240308` cycles	`1.00`
`ML-DSA-65 sign`	`691430` cycles	`695323` cycles	`0.99`
`ML-DSA-65 verify`	`230845` cycles	`231294` cycles	`1.00`
`ML-DSA-87 keypair`	`362280` cycles	`362446` cycles	`1.00`
`ML-DSA-87 sign`	`874006` cycles	`880689` cycles	`0.99`
`ML-DSA-87 verify`	`375607` cycles	`375553` cycles	`1.00`

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

oqs-bot

Graviton3

Benchmark suite	Current: `1fcd573`	Previous: `2c8f312`	Ratio
`ML-DSA-44 keypair`	`82280` cycles	`82277` cycles	`1.00`
`ML-DSA-44 sign`	`253217` cycles	`254105` cycles	`1.00`
`ML-DSA-44 verify`	`92187` cycles	`92276` cycles	`1.00`
`ML-DSA-65 keypair`	`148982` cycles	`149066` cycles	`1.00`
`ML-DSA-65 sign`	`415976` cycles	`416387` cycles	`1.00`
`ML-DSA-65 verify`	`147655` cycles	`147774` cycles	`1.00`
`ML-DSA-87 keypair`	`232365` cycles	`232399` cycles	`1.00`
`ML-DSA-87 sign`	`526443` cycles	`526369` cycles	`1.00`
`ML-DSA-87 verify`	`239330` cycles	`239305` cycles	`1.00`

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

oqs-bot

Graviton2 (no-opt)

Benchmark suite	Current: `1fcd573`	Previous: `2c8f312`	Ratio
`ML-DSA-44 keypair`	`209562` cycles	`210419` cycles	`1.00`
`ML-DSA-44 sign`	`722381` cycles	`722578` cycles	`1.00`
`ML-DSA-44 verify`	`233528` cycles	`238793` cycles	`0.98`
`ML-DSA-65 keypair`	`376334` cycles	`377835` cycles	`1.00`
`ML-DSA-65 sign`	`1186815` cycles	`1187372` cycles	`1.00`
`ML-DSA-65 verify`	`370207` cycles	`371286` cycles	`1.00`
`ML-DSA-87 keypair`	`597405` cycles	`595539` cycles	`1.00`
`ML-DSA-87 sign`	`1516820` cycles	`1517135` cycles	`1.00`
`ML-DSA-87 verify`	`613145` cycles	`613408` cycles	`1.00`

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

oqs-bot

Graviton3 (no-opt)

Benchmark suite	Current: `1fcd573`	Previous: `2c8f312`	Ratio
`ML-DSA-44 keypair`	`136224` cycles	`136206` cycles	`1.00`
`ML-DSA-44 sign`	`451156` cycles	`451058` cycles	`1.00`
`ML-DSA-44 verify`	`147230` cycles	`147189` cycles	`1.00`
`ML-DSA-65 keypair`	`239506` cycles	`239220` cycles	`1.00`
`ML-DSA-65 sign`	`733037` cycles	`732792` cycles	`1.00`
`ML-DSA-65 verify`	`237567` cycles	`237444` cycles	`1.00`
`ML-DSA-87 keypair`	`390260` cycles	`390239` cycles	`1.00`
`ML-DSA-87 sign`	`947572` cycles	`947960` cycles	`1.00`
`ML-DSA-87 verify`	`397074` cycles	`396989` cycles	`1.00`

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

github-actions

Mac Mini (M1, 2020) benchmarks (opt)

Benchmark suite	Current: `1fcd573`	Previous: `2c8f312`	Ratio
`ML-DSA-44 keypair`	`55129` cycles	`55117` cycles	`1.00`
`ML-DSA-44 sign`	`170291` cycles	`181260` cycles	`0.94`
`ML-DSA-44 verify`	`68909` cycles	`70286` cycles	`0.98`
`ML-DSA-65 keypair`	`98085` cycles	`98148` cycles	`1.00`
`ML-DSA-65 sign`	`271242` cycles	`288768` cycles	`0.94`
`ML-DSA-65 verify`	`108785` cycles	`110645` cycles	`0.98`
`ML-DSA-87 keypair`	`153013` cycles	`152964` cycles	`1.00`
`ML-DSA-87 sign`	`336690` cycles	`357460` cycles	`0.94`
`ML-DSA-87 verify`	`170662` cycles	`173366` cycles	`0.98`

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

Signed-off-by: Matthias J. Kannwischer <[email protected]>

oqs-bot

Intel Xeon 4th gen (c7i)

Benchmark suite	Current: `1fcd573`	Previous: `2c8f312`	Ratio
`ML-DSA-44 keypair`	`46639` cycles	`46293` cycles	`1.01`
`ML-DSA-44 sign`	`141354` cycles	`143051` cycles	`0.99`
`ML-DSA-44 verify`	`52616` cycles	`52701` cycles	`1.00`
`ML-DSA-65 keypair`	`82758` cycles	`83834` cycles	`0.99`
`ML-DSA-65 sign`	`234561` cycles	`239453` cycles	`0.98`
`ML-DSA-65 verify`	`82524` cycles	`83668` cycles	`0.99`
`ML-DSA-87 keypair`	`125143` cycles	`125409` cycles	`1.00`
`ML-DSA-87 sign`	`283577` cycles	`287343` cycles	`0.99`
`ML-DSA-87 verify`	`126906` cycles	`127995` cycles	`0.99`

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

oqs-bot

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

Benchmark suite	Current: `1fcd573`	Previous: `2c8f312`	Ratio
`ML-DSA-44 keypair`	`95700` cycles	`95631` cycles	`1.00`
`ML-DSA-44 sign`	`322282` cycles	`323360` cycles	`1.00`
`ML-DSA-44 verify`	`102390` cycles	`102372` cycles	`1.00`
`ML-DSA-65 keypair`	`163967` cycles	`163735` cycles	`1.00`
`ML-DSA-65 sign`	`528300` cycles	`528058` cycles	`1.00`
`ML-DSA-65 verify`	`162999` cycles	`162683` cycles	`1.00`
`ML-DSA-87 keypair`	`266517` cycles	`265734` cycles	`1.00`
`ML-DSA-87 sign`	`670862` cycles	`667701` cycles	`1.00`
`ML-DSA-87 verify`	`271269` cycles	`270607` cycles	`1.00`

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

oqs-bot

Intel Xeon 3rd gen (c6i)

Benchmark suite	Current: `1fcd573`	Previous: `2c8f312`	Ratio
`ML-DSA-44 keypair`	`76476` cycles	`76456` cycles	`1.00`
`ML-DSA-44 sign`	`216515` cycles	`220291` cycles	`0.98`
`ML-DSA-44 verify`	`85019` cycles	`85557` cycles	`0.99`
`ML-DSA-65 keypair`	`136112` cycles	`135777` cycles	`1.00`
`ML-DSA-65 sign`	`357643` cycles	`363388` cycles	`0.98`
`ML-DSA-65 verify`	`135074` cycles	`135930` cycles	`0.99`
`ML-DSA-87 keypair`	`205390` cycles	`205560` cycles	`1.00`
`ML-DSA-87 sign`	`434173` cycles	`441863` cycles	`0.98`
`ML-DSA-87 verify`	`205850` cycles	`206797` cycles	`1.00`

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

oqs-bot

AMD EPYC 3rd gen (c6a)

Benchmark suite	Current: `1fcd573`	Previous: `2c8f312`	Ratio
`ML-DSA-44 keypair`	`81207` cycles	`80849` cycles	`1.00`
`ML-DSA-44 sign`	`214285` cycles	`225721` cycles	`0.95`
`ML-DSA-44 verify`	`86200` cycles	`87762` cycles	`0.98`
`ML-DSA-65 keypair`	`142258` cycles	`142136` cycles	`1.00`
`ML-DSA-65 sign`	`347838` cycles	`362871` cycles	`0.96`
`ML-DSA-65 verify`	`137115` cycles	`139820` cycles	`0.98`
`ML-DSA-87 keypair`	`230776` cycles	`230443` cycles	`1.00`
`ML-DSA-87 sign`	`454061` cycles	`471376` cycles	`0.96`
`ML-DSA-87 verify`	`225608` cycles	`228079` cycles	`0.99`

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

oqs-bot

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

Benchmark suite	Current: `1fcd573`	Previous: `2c8f312`	Ratio
`ML-DSA-44 keypair`	`153986` cycles	`154048` cycles	`1.00`
`ML-DSA-44 sign`	`516623` cycles	`516877` cycles	`1.00`
`ML-DSA-44 verify`	`165341` cycles	`165471` cycles	`1.00`
`ML-DSA-65 keypair`	`260795` cycles	`260406` cycles	`1.00`
`ML-DSA-65 sign`	`833617` cycles	`833190` cycles	`1.00`
`ML-DSA-65 verify`	`264693` cycles	`264810` cycles	`1.00`
`ML-DSA-87 keypair`	`433114` cycles	`438000` cycles	`0.99`
`ML-DSA-87 sign`	`1070706` cycles	`1080273` cycles	`0.99`
`ML-DSA-87 verify`	`437811` cycles	`442889` cycles	`0.99`

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

oqs-bot

AMD EPYC 4th gen (c7a)

Benchmark suite	Current: `1fcd573`	Previous: `2c8f312`	Ratio
`ML-DSA-44 keypair`	`56105` cycles	`58037` cycles	`0.97`
`ML-DSA-44 sign`	`160387` cycles	`169809` cycles	`0.94`
`ML-DSA-44 verify`	`64308` cycles	`65688` cycles	`0.98`
`ML-DSA-65 keypair`	`103440` cycles	`98195` cycles	`1.05`
`ML-DSA-65 sign`	`263482` cycles	`257411` cycles	`1.02`
`ML-DSA-65 verify`	`100576` cycles	`98329` cycles	`1.02`
`ML-DSA-87 keypair`	`155197` cycles	`149538` cycles	`1.04`
`ML-DSA-87 sign`	`315363` cycles	`313605` cycles	`1.01`
`ML-DSA-87 verify`	`155592` cycles	`150685` cycles	`1.03`

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

oqs-bot

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

Benchmark suite	Current: `1fcd573`	Previous: `2c8f312`	Ratio
`ML-DSA-44 keypair`	`134008` cycles	`134104` cycles	`1.00`
`ML-DSA-44 sign`	`508048` cycles	`508236` cycles	`1.00`
`ML-DSA-44 verify`	`149202` cycles	`149075` cycles	`1.00`
`ML-DSA-65 keypair`	`223397` cycles	`223683` cycles	`1.00`
`ML-DSA-65 sign`	`814824` cycles	`812225` cycles	`1.00`
`ML-DSA-65 verify`	`233403` cycles	`233252` cycles	`1.00`
`ML-DSA-87 keypair`	`368395` cycles	`368108` cycles	`1.00`
`ML-DSA-87 sign`	`1027273` cycles	`1025498` cycles	`1.00`
`ML-DSA-87 verify`	`380965` cycles	`380626` cycles	`1.00`

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

oqs-bot

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

Benchmark suite	Current: `1fcd573`	Previous: `2c8f312`	Ratio
`ML-DSA-44 keypair`	`118302` cycles	`119172` cycles	`0.99`
`ML-DSA-44 sign`	`417091` cycles	`418564` cycles	`1.00`
`ML-DSA-44 verify`	`131102` cycles	`131691` cycles	`1.00`
`ML-DSA-65 keypair`	`200376` cycles	`200755` cycles	`1.00`
`ML-DSA-65 sign`	`668860` cycles	`676135` cycles	`0.99`
`ML-DSA-65 verify`	`204699` cycles	`206045` cycles	`0.99`
`ML-DSA-87 keypair`	`332465` cycles	`335161` cycles	`0.99`
`ML-DSA-87 sign`	`864265` cycles	`869596` cycles	`0.99`
`ML-DSA-87 verify`	`340400` cycles	`341973` cycles	`1.00`

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

github-actions

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

Benchmark suite	Current: `1fcd573`	Previous: `2c8f312`	Ratio
`ML-DSA-44 keypair`	`112349` cycles	`112348` cycles	`1.00`
`ML-DSA-44 sign`	`409302` cycles	`409341` cycles	`1.00`
`ML-DSA-44 verify`	`130067` cycles	`130058` cycles	`1.00`
`ML-DSA-65 keypair`	`192847` cycles	`192877` cycles	`1.00`
`ML-DSA-65 sign`	`658811` cycles	`658871` cycles	`1.00`
`ML-DSA-65 verify`	`207224` cycles	`207182` cycles	`1.00`
`ML-DSA-87 keypair`	`316032` cycles	`316055` cycles	`1.00`
`ML-DSA-87 sign`	`832943` cycles	`832950` cycles	`1.00`
`ML-DSA-87 verify`	`339082` cycles	`339076` cycles	`1.00`

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

github-actions

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

Benchmark suite	Current: `1fcd573`	Previous: `2c8f312`	Ratio
`ML-DSA-44 keypair`	`317762` cycles	`317096` cycles	`1.00`
`ML-DSA-44 sign`	`997799` cycles	`1018044` cycles	`0.98`
`ML-DSA-44 verify`	`341059` cycles	`342860` cycles	`0.99`
`ML-DSA-65 keypair`	`555083` cycles	`555550` cycles	`1.00`
`ML-DSA-65 sign`	`1660502` cycles	`1678586` cycles	`0.99`
`ML-DSA-65 verify`	`543467` cycles	`547237` cycles	`0.99`
`ML-DSA-87 keypair`	`925456` cycles	`918592` cycles	`1.01`
`ML-DSA-87 sign`	`2264430` cycles	`2226269` cycles	`1.02`
`ML-DSA-87 verify`	`920921` cycles	`923214` cycles	`1.00`

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

github-actions

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

Benchmark suite	Current: `1fcd573`	Previous: `2c8f312`	Ratio
`ML-DSA-44 keypair`	`452588` cycles	`451876` cycles	`1.00`
`ML-DSA-44 sign`	`2008448` cycles	`2006209` cycles	`1.00`
`ML-DSA-44 verify`	`526344` cycles	`525659` cycles	`1.00`
`ML-DSA-65 keypair`	`761801` cycles	`759548` cycles	`1.00`
`ML-DSA-65 sign`	`3334109` cycles	`3318976` cycles	`1.00`
`ML-DSA-65 verify`	`819837` cycles	`817753` cycles	`1.00`
`ML-DSA-87 keypair`	`1228908` cycles	`1223157` cycles	`1.00`
`ML-DSA-87 sign`	`4143200` cycles	`4111725` cycles	`1.01`
`ML-DSA-87 verify`	`1316509` cycles	`1312412` cycles	`1.00`

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

github-actions

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

Benchmark suite	Current: `1fcd573`	Previous: `2c8f312`	Ratio
`ML-DSA-44 keypair`	`128536` cycles	`128662` cycles	`1.00`
`ML-DSA-44 sign`	`418466` cycles	`419335` cycles	`1.00`
`ML-DSA-44 verify`	`142201` cycles	`142486` cycles	`1.00`
`ML-DSA-65 keypair`	`240068` cycles	`240198` cycles	`1.00`
`ML-DSA-65 sign`	`690896` cycles	`694792` cycles	`0.99`
`ML-DSA-65 verify`	`230790` cycles	`231005` cycles	`1.00`
`ML-DSA-87 keypair`	`361956` cycles	`361975` cycles	`1.00`
`ML-DSA-87 sign`	`872989` cycles	`879257` cycles	`0.99`
`ML-DSA-87 verify`	`375760` cycles	`375200` cycles	`1.00`

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

github-actions

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

Benchmark suite	Current: `1fcd573`	Previous: `2c8f312`	Ratio
`ML-DSA-44 keypair`	`939171` cycles	`937991` cycles	`1.00`
`ML-DSA-44 sign`	`4325879` cycles	`4322008` cycles	`1.00`
`ML-DSA-44 verify`	`1071821` cycles	`1070846` cycles	`1.00`
`ML-DSA-65 keypair`	`1556170` cycles	`1555235` cycles	`1.00`
`ML-DSA-65 sign`	`7143723` cycles	`7134424` cycles	`1.00`
`ML-DSA-65 verify`	`1692624` cycles	`1691055` cycles	`1.00`
`ML-DSA-87 keypair`	`2525632` cycles	`2522308` cycles	`1.00`
`ML-DSA-87 sign`	`8747724` cycles	`8729409` cycles	`1.00`
`ML-DSA-87 verify`	`2710156` cycles	`2706680` 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)

Benchmark suite	Current: `1fcd573`	Previous: `2c8f312`	Ratio
`ML-DSA-44 keypair`	`209655` cycles	`209811` cycles	`1.00`
`ML-DSA-44 sign`	`722179` cycles	`721462` cycles	`1.00`
`ML-DSA-44 verify`	`228283` cycles	`228512` cycles	`1.00`
`ML-DSA-65 keypair`	`375596` cycles	`375810` cycles	`1.00`
`ML-DSA-65 sign`	`1186352` cycles	`1185979` cycles	`1.00`
`ML-DSA-65 verify`	`370168` cycles	`370610` cycles	`1.00`
`ML-DSA-87 keypair`	`596584` cycles	`595808` cycles	`1.00`
`ML-DSA-87 sign`	`1513903` cycles	`1514568` cycles	`1.00`
`ML-DSA-87 verify`	`612620` cycles	`613735` cycles	`1.00`

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

github-actions

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

Benchmark suite	Current: `1fcd573`	Previous: `2c8f312`	Ratio
`ML-DSA-44 keypair`	`251270` cycles	`237197` cycles	`1.06`
`ML-DSA-44 sign`	`749767` cycles	`703842` cycles	`1.07`
`ML-DSA-44 verify`	`259927` cycles	`240763` cycles	`1.08`
`ML-DSA-65 keypair`	`476884` cycles	`463657` cycles	`1.03`
`ML-DSA-65 sign`	`1216737` cycles	`1186377` cycles	`1.03`
`ML-DSA-65 verify`	`435656` cycles	`418257` cycles	`1.04`
`ML-DSA-87 keypair`	`691818` cycles	`696260` cycles	`0.99`
`ML-DSA-87 sign`	`1535844` cycles	`1539476` cycles	`1.00`
`ML-DSA-87 verify`	`693012` cycles	`690657` cycles	`1.00`

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

github-actions

⚠️ Performance Alert ⚠️

Possible performance regression was detected for benchmark 'Arm Cortex-A72 (Raspberry Pi 4) benchmarks (opt)'.
Benchmark result of this commit is worse than the previous benchmark result exceeding threshold 1.03.

Benchmark suite	Current: `1fcd573`	Previous: `2c8f312`	Ratio
`ML-DSA-44 keypair`	`251270` cycles	`237197` cycles	`1.06`
`ML-DSA-44 sign`	`749767` cycles	`703842` cycles	`1.07`
`ML-DSA-44 verify`	`259927` cycles	`240763` cycles	`1.08`
`ML-DSA-65 verify`	`435656` cycles	`418257` cycles	`1.04`

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

github-actions

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

Benchmark suite	Current: `1fcd573`	Previous: `2c8f312`	Ratio
`ML-DSA-44 keypair`	`291947` cycles	`310373` cycles	`0.94`
`ML-DSA-44 sign`	`1080188` cycles	`1132919` cycles	`0.95`
`ML-DSA-44 verify`	`319505` cycles	`332037` cycles	`0.96`
`ML-DSA-65 keypair`	`546967` cycles	`559733` cycles	`0.98`
`ML-DSA-65 sign`	`1793387` cycles	`1850373` cycles	`0.97`
`ML-DSA-65 verify`	`528504` cycles	`536497` cycles	`0.99`
`ML-DSA-87 keypair`	`834990` cycles	`842080` cycles	`0.99`
`ML-DSA-87 sign`	`2260558` cycles	`2348501` cycles	`0.96`
`ML-DSA-87 verify`	`864553` cycles	`880866` cycles	`0.98`

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

mldsa/native/x86_64/src/pointwise.S

Signed-off-by: Matthias J. Kannwischer <[email protected]>

oqs-bot

⚠️ Performance Alert ⚠️

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

Benchmark suite	Current: `1fcd573`	Previous: `2c8f312`	Ratio
`ML-DSA-65 keypair`	`103440` cycles	`98195` cycles	`1.05`
`ML-DSA-87 keypair`	`155197` cycles	`149538` cycles	`1.04`
`ML-DSA-87 verify`	`155592` cycles	`150685` cycles	`1.03`

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

mkannwischer mentioned this pull request Jul 2, 2025

AVX2: Add poly_pointwise_montgomery assembly #337

Open

mkannwischer added the benchmark label Jul 2, 2025

oqs-bot reviewed Jul 2, 2025

View reviewed changes

This comment was marked as outdated.

Sign in to view

oqs-bot reviewed Jul 2, 2025

View reviewed changes

mkannwischer added benchmark and removed benchmark labels Jul 2, 2025

github-actions bot reviewed Jul 2, 2025

View reviewed changes

mkannwischer force-pushed the poly-pointwise-montgomery branch from 8b6bd85 to dc976dc Compare July 2, 2025 04:50

AArch64: Native poly_pointwise_montgomery

eeb9204

Signed-off-by: Matthias J. Kannwischer <[email protected]>

mkannwischer force-pushed the poly-pointwise-montgomery branch from dc976dc to a4856a7 Compare July 2, 2025 04:53

mkannwischer added benchmark and removed benchmark labels Jul 2, 2025

oqs-bot reviewed Jul 2, 2025

View reviewed changes

github-actions bot reviewed Jul 2, 2025

View reviewed changes

jakemas reviewed Jul 2, 2025

View reviewed changes

mldsa/native/x86_64/src/pointwise.S Outdated Show resolved Hide resolved

mkannwischer added 2 commits July 3, 2025 14:46

AVX2: Native poly_pointwise_montgomery

47a3e21

Signed-off-by: Matthias J. Kannwischer <[email protected]>

[TEST]: Apply SLOTHY to AArch64 pointwise_montgomery.S

1fcd573

Signed-off-by: Matthias J. Kannwischer <[email protected]>

mkannwischer force-pushed the poly-pointwise-montgomery branch from a4856a7 to 1fcd573 Compare July 3, 2025 06:46

mkannwischer added benchmark and removed benchmark labels Jul 3, 2025

oqs-bot reviewed Jul 3, 2025

View reviewed changes

AVX2/AArch64: Add native poly_pointwise_montgomery #346

Are you sure you want to change the base?

AVX2/AArch64: Add native poly_pointwise_montgomery #346

Uh oh!

Conversation

mkannwischer commented Jul 2, 2025

Uh oh!

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

Uh oh!

Choose a reason for hiding this comment

Graviton4

Uh oh!

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

Uh oh!

Choose a reason for hiding this comment

Graviton4 (no-opt)

Uh oh!

This comment was marked as outdated.

Uh oh!

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

Uh oh!

Choose a reason for hiding this comment

Graviton2

Uh oh!

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

Uh oh!

Choose a reason for hiding this comment

Graviton3

Uh oh!

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

Uh oh!

Choose a reason for hiding this comment

Graviton2 (no-opt)

Uh oh!

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

Uh oh!

Choose a reason for hiding this comment

Graviton3 (no-opt)

Uh oh!

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

Uh oh!

Choose a reason for hiding this comment

Mac Mini (M1, 2020) benchmarks (opt)

Uh oh!

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

Uh oh!

Choose a reason for hiding this comment

Intel Xeon 4th gen (c7i)

Uh oh!

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

Uh oh!

Choose a reason for hiding this comment

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

Uh oh!

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

Uh oh!

Choose a reason for hiding this comment

Intel Xeon 3rd gen (c6i)

Uh oh!

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

Uh oh!

Choose a reason for hiding this comment

AMD EPYC 3rd gen (c6a)

Uh oh!

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

Uh oh!

Choose a reason for hiding this comment

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

Uh oh!

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

Uh oh!

Choose a reason for hiding this comment

AMD EPYC 4th gen (c7a)

Uh oh!

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

Uh oh!

Choose a reason for hiding this comment

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

Uh oh!

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

AVX2/AArch64: Add native `poly_pointwise_montgomery` #346

AVX2/AArch64: Add native `poly_pointwise_montgomery` #346

oqs-bot left a comment •

edited

Loading

oqs-bot left a comment •

edited

Loading

oqs-bot left a comment •

edited

Loading

oqs-bot left a comment •

edited

Loading

oqs-bot left a comment •

edited

Loading

oqs-bot left a comment •

edited

Loading

github-actions bot left a comment •

edited

Loading

oqs-bot left a comment •

edited

Loading

oqs-bot left a comment •

edited

Loading

oqs-bot left a comment •

edited

Loading

oqs-bot left a comment •

edited

Loading

oqs-bot left a comment •

edited

Loading

oqs-bot left a comment •

edited

Loading

oqs-bot left a comment •

edited

Loading

oqs-bot left a comment •

edited

Loading

github-actions bot left a comment •

edited

Loading

github-actions bot left a comment •

edited

Loading

github-actions bot left a comment •

edited

Loading

github-actions bot left a comment •

edited

Loading

github-actions bot left a comment •

edited

Loading

github-actions bot left a comment •

edited

Loading

github-actions bot left a comment •

edited

Loading

github-actions bot left a comment •

edited

Loading

github-actions bot left a comment •

edited

Loading