Reimplement AArch64 NTT/iNTT #570

mkannwischer · 2025-10-30T03:08:56Z

The current NTT + iNTT in mldsa-native originate from the SLOTHY repository and
are, hence, MIT-licensed.
Relicensing was attempted in
#236
without success.

This commit re-implements both iNTT and NTT based on the "Apache-2.0 OR ISC OR MIT"-licensed NTT and iNTT from mlkem-native.
The first commit switches to the exact files from mlkem-native, and the second commit makes minimal changes to make it work for ML-DSA.

Resolves AArch64: Relicense or rewrite ntt.s and intt.s #381

github-actions

Mac Mini (M1, 2020) benchmarks (opt)

Benchmark suite	Current: `2726c27`	Previous: `0d8fcae`	Ratio
`ML-DSA-44 keypair`	`46224` cycles	`46214` cycles	`1.00`
`ML-DSA-44 sign`	`132632` cycles	`132516` cycles	`1.00`
`ML-DSA-44 verify`	`47870` cycles	`47875` cycles	`1.00`
`ML-DSA-65 keypair`	`81165` cycles	`81134` cycles	`1.00`
`ML-DSA-65 sign`	`219234` cycles	`219047` cycles	`1.00`
`ML-DSA-65 verify`	`80151` cycles	`80131` cycles	`1.00`
`ML-DSA-87 keypair`	`132347` cycles	`132358` cycles	`1.00`
`ML-DSA-87 sign`	`281155` cycles	`280907` cycles	`1.00`
`ML-DSA-87 verify`	`130365` cycles	`130354` 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: `2726c27`	Previous: `0d8fcae`	Ratio
`ML-DSA-44 keypair`	`115054` cycles	`115060` cycles	`1.00`
`ML-DSA-44 sign`	`431657` cycles	`431622` cycles	`1.00`
`ML-DSA-44 verify`	`122217` cycles	`122202` cycles	`1.00`
`ML-DSA-65 keypair`	`197059` cycles	`197106` cycles	`1.00`
`ML-DSA-65 sign`	`701051` cycles	`701081` cycles	`1.00`
`ML-DSA-65 verify`	`197720` cycles	`197681` cycles	`1.00`
`ML-DSA-87 keypair`	`325231` cycles	`325228` cycles	`1.00`
`ML-DSA-87 sign`	`884685` cycles	`884716` cycles	`1.00`
`ML-DSA-87 verify`	`328830` cycles	`328859` 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: `2726c27`	Previous: `0d8fcae`	Ratio
`ML-DSA-44 keypair`	`288315` cycles	`287853` cycles	`1.00`
`ML-DSA-44 sign`	`936816` cycles	`917885` cycles	`1.02`
`ML-DSA-44 verify`	`292070` cycles	`294498` cycles	`0.99`
`ML-DSA-65 keypair`	`488360` cycles	`486448` cycles	`1.00`
`ML-DSA-65 sign`	`1528831` cycles	`1508569` cycles	`1.01`
`ML-DSA-65 verify`	`482094` cycles	`476811` cycles	`1.01`
`ML-DSA-87 keypair`	`837518` cycles	`829526` cycles	`1.01`
`ML-DSA-87 sign`	`2096524` cycles	`2031020` cycles	`1.03`
`ML-DSA-87 verify`	`820254` cycles	`811735` cycles	`1.01`

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

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: `2726c27`	Previous: `0d8fcae`	Ratio
`ML-DSA-87 sign`	`2096524` cycles	`2031020` cycles	`1.03`

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

oqs-bot

Intel Xeon 4th gen (c7i)

Benchmark suite	Current: `2726c27`	Previous: `0d8fcae`	Ratio
`ML-DSA-44 keypair`	`34907` cycles	`34885` cycles	`1.00`
`ML-DSA-44 sign`	`121063` cycles	`120963` cycles	`1.00`
`ML-DSA-44 verify`	`38343` cycles	`38317` cycles	`1.00`
`ML-DSA-65 keypair`	`61999` cycles	`62663` cycles	`0.99`
`ML-DSA-65 sign`	`200215` cycles	`201214` cycles	`1.00`
`ML-DSA-65 verify`	`62514` cycles	`62474` cycles	`1.00`
`ML-DSA-87 keypair`	`94914` cycles	`94634` cycles	`1.00`
`ML-DSA-87 sign`	`235124` cycles	`233708` cycles	`1.01`
`ML-DSA-87 verify`	`93941` cycles	`94546` 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: `2726c27`	Previous: `0d8fcae`	Ratio
`ML-DSA-44 keypair`	`95055` cycles	`95105` cycles	`1.00`
`ML-DSA-44 sign`	`348626` cycles	`348797` cycles	`1.00`
`ML-DSA-44 verify`	`100722` cycles	`100891` cycles	`1.00`
`ML-DSA-65 keypair`	`164854` cycles	`164366` cycles	`1.00`
`ML-DSA-65 sign`	`567460` cycles	`567110` cycles	`1.00`
`ML-DSA-65 verify`	`165353` cycles	`165083` cycles	`1.00`
`ML-DSA-87 keypair`	`267189` cycles	`267179` cycles	`1.00`
`ML-DSA-87 sign`	`721701` cycles	`722256` cycles	`1.00`
`ML-DSA-87 verify`	`272143` cycles	`271890` cycles	`1.00`

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

oqs-bot

Intel Xeon 3rd gen (c6i)

Benchmark suite	Current: `2726c27`	Previous: `0d8fcae`	Ratio
`ML-DSA-44 keypair`	`57131` cycles	`57327` cycles	`1.00`
`ML-DSA-44 sign`	`180912` cycles	`180385` cycles	`1.00`
`ML-DSA-44 verify`	`61419` cycles	`61211` cycles	`1.00`
`ML-DSA-65 keypair`	`99741` cycles	`99428` cycles	`1.00`
`ML-DSA-65 sign`	`296897` cycles	`296541` cycles	`1.00`
`ML-DSA-65 verify`	`100189` cycles	`99935` cycles	`1.00`
`ML-DSA-87 keypair`	`153657` cycles	`153664` cycles	`1.00`
`ML-DSA-87 sign`	`353986` cycles	`352456` cycles	`1.00`
`ML-DSA-87 verify`	`152885` cycles	`153187` cycles	`1.00`

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

oqs-bot

AMD EPYC 4th gen (c7a)

Benchmark suite	Current: `2726c27`	Previous: `0d8fcae`	Ratio
`ML-DSA-44 keypair`	`42040` cycles	`42710` cycles	`0.98`
`ML-DSA-44 sign`	`130621` cycles	`130749` cycles	`1.00`
`ML-DSA-44 verify`	`45006` cycles	`44253` cycles	`1.02`
`ML-DSA-65 keypair`	`72376` cycles	`73757` cycles	`0.98`
`ML-DSA-65 sign`	`211462` cycles	`211376` cycles	`1.00`
`ML-DSA-65 verify`	`72885` cycles	`72870` cycles	`1.00`
`ML-DSA-87 keypair`	`109455` cycles	`110490` cycles	`0.99`
`ML-DSA-87 sign`	`248738` cycles	`249050` cycles	`1.00`
`ML-DSA-87 verify`	`110765` cycles	`109887` cycles	`1.01`

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

oqs-bot

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

Benchmark suite	Current: `2726c27`	Previous: `0d8fcae`	Ratio
`ML-DSA-44 keypair`	`158253` cycles	`157889` cycles	`1.00`
`ML-DSA-44 sign`	`566352` cycles	`566893` cycles	`1.00`
`ML-DSA-44 verify`	`170172` cycles	`169660` cycles	`1.00`
`ML-DSA-65 keypair`	`270655` cycles	`270428` cycles	`1.00`
`ML-DSA-65 sign`	`926442` cycles	`926410` cycles	`1.00`
`ML-DSA-65 verify`	`276134` cycles	`275530` cycles	`1.00`
`ML-DSA-87 keypair`	`451653` cycles	`453802` cycles	`1.00`
`ML-DSA-87 sign`	`1182253` cycles	`1189145` cycles	`0.99`
`ML-DSA-87 verify`	`460960` cycles	`462952` 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: `2726c27`	Previous: `0d8fcae`	Ratio
`ML-DSA-44 keypair`	`120029` cycles	`120561` cycles	`1.00`
`ML-DSA-44 sign`	`455292` cycles	`455880` cycles	`1.00`
`ML-DSA-44 verify`	`130189` cycles	`130918` cycles	`0.99`
`ML-DSA-65 keypair`	`205204` cycles	`206047` cycles	`1.00`
`ML-DSA-65 sign`	`737270` cycles	`739174` cycles	`1.00`
`ML-DSA-65 verify`	`210582` cycles	`210656` cycles	`1.00`
`ML-DSA-87 keypair`	`337660` cycles	`337444` cycles	`1.00`
`ML-DSA-87 sign`	`925392` cycles	`927610` cycles	`1.00`
`ML-DSA-87 verify`	`345045` cycles	`345586` 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: `2726c27`	Previous: `0d8fcae`	Ratio
`ML-DSA-44 keypair`	`466361` cycles	`466242` cycles	`1.00`
`ML-DSA-44 sign`	`2214442` cycles	`2212266` cycles	`1.00`
`ML-DSA-44 verify`	`550729` cycles	`550372` cycles	`1.00`
`ML-DSA-65 keypair`	`778060` cycles	`778748` cycles	`1.00`
`ML-DSA-65 sign`	`3645402` cycles	`3635941` cycles	`1.00`
`ML-DSA-65 verify`	`850119` cycles	`854033` cycles	`1.00`
`ML-DSA-87 keypair`	`1261462` cycles	`1268682` cycles	`0.99`
`ML-DSA-87 sign`	`4521620` cycles	`4497329` cycles	`1.01`
`ML-DSA-87 verify`	`1370691` cycles	`1375035` cycles	`1.00`

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

oqs-bot

AMD EPYC 3rd gen (c6a)

Benchmark suite	Current: `2726c27`	Previous: `0d8fcae`	Ratio
`ML-DSA-44 keypair`	`69402` cycles	`69123` cycles	`1.00`
`ML-DSA-44 sign`	`185402` cycles	`186494` cycles	`0.99`
`ML-DSA-44 verify`	`69157` cycles	`69176` cycles	`1.00`
`ML-DSA-65 keypair`	`121024` cycles	`119562` cycles	`1.01`
`ML-DSA-65 sign`	`299933` cycles	`296189` cycles	`1.01`
`ML-DSA-65 verify`	`116947` cycles	`115372` cycles	`1.01`
`ML-DSA-87 keypair`	`202031` cycles	`201764` cycles	`1.00`
`ML-DSA-87 sign`	`386717` cycles	`385754` cycles	`1.00`
`ML-DSA-87 verify`	`193861` cycles	`193267` cycles	`1.00`

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

oqs-bot

Graviton2

Benchmark suite	Current: `2726c27`	Previous: `0d8fcae`	Ratio
`ML-DSA-44 keypair`	`115212` cycles	`115338` cycles	`1.00`
`ML-DSA-44 sign`	`377502` cycles	`377797` cycles	`1.00`
`ML-DSA-44 verify`	`120434` cycles	`120362` cycles	`1.00`
`ML-DSA-65 keypair`	`199186` cycles	`199199` cycles	`1.00`
`ML-DSA-65 sign`	`624251` cycles	`623684` cycles	`1.00`
`ML-DSA-65 verify`	`198357` cycles	`198333` cycles	`1.00`
`ML-DSA-87 keypair`	`326376` cycles	`326299` cycles	`1.00`
`ML-DSA-87 sign`	`791842` cycles	`791197` cycles	`1.00`
`ML-DSA-87 verify`	`325180` cycles	`325134` cycles	`1.00`

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

oqs-bot

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

Benchmark suite	Current: `2726c27`	Previous: `0d8fcae`	Ratio
`ML-DSA-44 keypair`	`134922` cycles	`134866` cycles	`1.00`
`ML-DSA-44 sign`	`539696` cycles	`539900` cycles	`1.00`
`ML-DSA-44 verify`	`148744` cycles	`148419` cycles	`1.00`
`ML-DSA-65 keypair`	`228639` cycles	`228041` cycles	`1.00`
`ML-DSA-65 sign`	`890793` cycles	`890286` cycles	`1.00`
`ML-DSA-65 verify`	`238719` cycles	`238030` cycles	`1.00`
`ML-DSA-87 keypair`	`374177` cycles	`372651` cycles	`1.00`
`ML-DSA-87 sign`	`1109288` cycles	`1105492` cycles	`1.00`
`ML-DSA-87 verify`	`388030` cycles	`387568` cycles	`1.00`

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

oqs-bot

Graviton4

Benchmark suite	Current: `2726c27`	Previous: `0d8fcae`	Ratio
`ML-DSA-44 keypair`	`69473` cycles	`69256` cycles	`1.00`
`ML-DSA-44 sign`	`215180` cycles	`214215` cycles	`1.00`
`ML-DSA-44 verify`	`72765` cycles	`72454` cycles	`1.00`
`ML-DSA-65 keypair`	`122913` cycles	`123009` cycles	`1.00`
`ML-DSA-65 sign`	`353022` cycles	`352613` cycles	`1.00`
`ML-DSA-65 verify`	`120784` cycles	`120850` cycles	`1.00`
`ML-DSA-87 keypair`	`200531` cycles	`200453` cycles	`1.00`
`ML-DSA-87 sign`	`451469` cycles	`450752` cycles	`1.00`
`ML-DSA-87 verify`	`198605` cycles	`198288` cycles	`1.00`

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

oqs-bot

Graviton3

Benchmark suite	Current: `2726c27`	Previous: `0d8fcae`	Ratio
`ML-DSA-44 keypair`	`73848` cycles	`73607` cycles	`1.00`
`ML-DSA-44 sign`	`228437` cycles	`227471` cycles	`1.00`
`ML-DSA-44 verify`	`78181` cycles	`77864` cycles	`1.00`
`ML-DSA-65 keypair`	`129680` cycles	`129703` cycles	`1.00`
`ML-DSA-65 sign`	`378385` cycles	`376720` cycles	`1.00`
`ML-DSA-65 verify`	`129105` cycles	`128985` cycles	`1.00`
`ML-DSA-87 keypair`	`208328` cycles	`210411` cycles	`0.99`
`ML-DSA-87 sign`	`475098` cycles	`478519` cycles	`0.99`
`ML-DSA-87 verify`	`208464` cycles	`210010` cycles	`0.99`

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

oqs-bot

Graviton4 (no-opt)

Benchmark suite	Current: `2726c27`	Previous: `0d8fcae`	Ratio
`ML-DSA-44 keypair`	`132772` cycles	`132736` cycles	`1.00`
`ML-DSA-44 sign`	`498728` cycles	`497985` cycles	`1.00`
`ML-DSA-44 verify`	`144884` cycles	`144837` cycles	`1.00`
`ML-DSA-65 keypair`	`226579` cycles	`226300` cycles	`1.00`
`ML-DSA-65 sign`	`813455` cycles	`813075` cycles	`1.00`
`ML-DSA-65 verify`	`231767` cycles	`231206` cycles	`1.00`
`ML-DSA-87 keypair`	`374712` cycles	`374564` cycles	`1.00`
`ML-DSA-87 sign`	`1020930` cycles	`1021102` cycles	`1.00`
`ML-DSA-87 verify`	`383904` cycles	`383658` cycles	`1.00`

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

oqs-bot

Graviton2 (no-opt)

Benchmark suite	Current: `2726c27`	Previous: `0d8fcae`	Ratio
`ML-DSA-44 keypair`	`213697` cycles	`213517` cycles	`1.00`
`ML-DSA-44 sign`	`784515` cycles	`794796` cycles	`0.99`
`ML-DSA-44 verify`	`230134` cycles	`229984` cycles	`1.00`
`ML-DSA-65 keypair`	`384987` cycles	`384988` cycles	`1.00`
`ML-DSA-65 sign`	`1309061` cycles	`1308906` cycles	`1.00`
`ML-DSA-65 verify`	`376307` cycles	`376450` cycles	`1.00`
`ML-DSA-87 keypair`	`605064` cycles	`605742` cycles	`1.00`
`ML-DSA-87 sign`	`1625253` cycles	`1626304` cycles	`1.00`
`ML-DSA-87 verify`	`616808` cycles	`617507` cycles	`1.00`

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

oqs-bot

Graviton3 (no-opt)

Benchmark suite	Current: `2726c27`	Previous: `0d8fcae`	Ratio
`ML-DSA-44 keypair`	`138492` cycles	`138358` cycles	`1.00`
`ML-DSA-44 sign`	`493164` cycles	`493462` cycles	`1.00`
`ML-DSA-44 verify`	`148520` cycles	`148380` cycles	`1.00`
`ML-DSA-65 keypair`	`241747` cycles	`241583` cycles	`1.00`
`ML-DSA-65 sign`	`809994` cycles	`809602` cycles	`1.00`
`ML-DSA-65 verify`	`240773` cycles	`240577` cycles	`1.00`
`ML-DSA-87 keypair`	`395991` cycles	`395743` cycles	`1.00`
`ML-DSA-87 sign`	`1027497` cycles	`1027225` cycles	`1.00`
`ML-DSA-87 verify`	`401639` cycles	`401343` 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: `2726c27`	Previous: `0d8fcae`	Ratio
`ML-DSA-44 keypair`	`825007` cycles	`822298` cycles	`1.00`
`ML-DSA-44 sign`	`3333179` cycles	`3328889` cycles	`1.00`
`ML-DSA-44 verify`	`921539` cycles	`919881` cycles	`1.00`
`ML-DSA-65 keypair`	`1398851` cycles	`1399516` cycles	`1.00`
`ML-DSA-65 sign`	`5455024` cycles	`5464226` cycles	`1.00`
`ML-DSA-65 verify`	`1466984` cycles	`1467271` cycles	`1.00`
`ML-DSA-87 keypair`	`2302372` cycles	`2302736` cycles	`1.00`
`ML-DSA-87 sign`	`6816040` cycles	`6829767` cycles	`1.00`
`ML-DSA-87 verify`	`2404751` cycles	`2397322` 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: `2726c27`	Previous: `0d8fcae`	Ratio
`ML-DSA-44 keypair`	`115005` cycles	`115154` cycles	`1.00`
`ML-DSA-44 sign`	`377208` cycles	`377500` cycles	`1.00`
`ML-DSA-44 verify`	`120241` cycles	`120247` cycles	`1.00`
`ML-DSA-65 keypair`	`198985` cycles	`199032` cycles	`1.00`
`ML-DSA-65 sign`	`623927` cycles	`623317` cycles	`1.00`
`ML-DSA-65 verify`	`198166` cycles	`198232` cycles	`1.00`
`ML-DSA-87 keypair`	`325843` cycles	`326045` cycles	`1.00`
`ML-DSA-87 sign`	`790782` cycles	`790668` cycles	`1.00`
`ML-DSA-87 verify`	`324887` cycles	`324882` 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: `2726c27`	Previous: `0d8fcae`	Ratio
`ML-DSA-44 keypair`	`213046` cycles	`213136` cycles	`1.00`
`ML-DSA-44 sign`	`782804` cycles	`782135` cycles	`1.00`
`ML-DSA-44 verify`	`229810` cycles	`230342` cycles	`1.00`
`ML-DSA-65 keypair`	`384499` cycles	`384139` cycles	`1.00`
`ML-DSA-65 sign`	`1307550` cycles	`1315123` cycles	`0.99`
`ML-DSA-65 verify`	`375891` cycles	`375712` cycles	`1.00`
`ML-DSA-87 keypair`	`604919` cycles	`605390` cycles	`1.00`
`ML-DSA-87 sign`	`1622524` cycles	`1623045` cycles	`1.00`
`ML-DSA-87 verify`	`616996` cycles	`617470` 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: `2726c27`	Previous: `0d8fcae`	Ratio
`ML-DSA-44 keypair`	`233548` cycles	`235495` cycles	`0.99`
`ML-DSA-44 sign`	`667993` cycles	`662800` cycles	`1.01`
`ML-DSA-44 verify`	`234753` cycles	`226921` cycles	`1.03`
`ML-DSA-65 keypair`	`392605` cycles	`412469` cycles	`0.95`
`ML-DSA-65 sign`	`1092773` cycles	`1122974` cycles	`0.97`
`ML-DSA-65 verify`	`377801` cycles	`392531` cycles	`0.96`
`ML-DSA-87 keypair`	`656111` cycles	`688346` cycles	`0.95`
`ML-DSA-87 sign`	`1467364` cycles	`1474034` cycles	`1.00`
`ML-DSA-87 verify`	`651523` cycles	`663133` cycles	`0.98`

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: `2726c27`	Previous: `0d8fcae`	Ratio
`ML-DSA-44 verify`	`234753` cycles	`226921` cycles	`1.03`

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: `2726c27`	Previous: `0d8fcae`	Ratio
`ML-DSA-44 keypair`	`300059` cycles	`308602` cycles	`0.97`
`ML-DSA-44 sign`	`1181409` cycles	`1178008` cycles	`1.00`
`ML-DSA-44 verify`	`328487` cycles	`335471` cycles	`0.98`
`ML-DSA-65 keypair`	`556668` cycles	`557372` cycles	`1.00`
`ML-DSA-65 sign`	`1972952` cycles	`1916650` cycles	`1.03`
`ML-DSA-65 verify`	`531684` cycles	`527142` cycles	`1.01`
`ML-DSA-87 keypair`	`876554` cycles	`873711` cycles	`1.00`
`ML-DSA-87 sign`	`2481949` cycles	`2484925` cycles	`1.00`
`ML-DSA-87 verify`	`908833` cycles	`886721` cycles	`1.02`

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

dev/aarch64_clean/src/intt.S

The current NTT + iNTT in mldsa-native originate from the SLOTHY repository and are, hence, MIT-licensed. Relicensing was attempted in #236 without success. This commit and the following commit do a clean-room re-implementation of the same techniques. We start of with the "Apache-2.0 OR ISC OR MIT"-licensed NTT and iNTT from mlkem-native and make the necessary changes to turn it into a ML-DSA NTT. Signed-off-by: Matthias J. Kannwischer <[email protected]>

This is the second commit reimplementing the NTT based on the "Apache-2.0 OR ISC OR MIT"-licensed NTT and iNTT from mlkem-native. This commit changes the ML-KEM NTT into a ML-DSA NTT using minimal changes. Resolves #381 Signed-off-by: Matthias J. Kannwischer <[email protected]>

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

#570 reimplemented the iNTT/NTT to achieve uniform Apache-2.0 OR ISC OR MIT licensing. This commit updates the LICENSE to remove the outdated license for the old NTT and INTT - these were the last files covered by that license. Signed-off-by: Matthias J. Kannwischer <[email protected]>

mkannwischer force-pushed the ntt-again branch from ffd6e03 to 2726c27 Compare October 30, 2025 03:09

mkannwischer added the benchmark label Oct 30, 2025

github-actions bot reviewed Oct 30, 2025

View reviewed changes

oqs-bot reviewed Oct 30, 2025

View reviewed changes

github-actions bot reviewed Oct 30, 2025

View reviewed changes

oqs-bot reviewed Oct 30, 2025

View reviewed changes

github-actions bot reviewed Oct 30, 2025

View reviewed changes

mkannwischer marked this pull request as ready for review October 30, 2025 03:33

mkannwischer requested a review from a team as a code owner October 30, 2025 03:33

mkannwischer force-pushed the ntt-again branch from 2726c27 to 487bdb6 Compare October 30, 2025 03:35

mkannwischer force-pushed the ntt-again branch from 487bdb6 to 655aaa6 Compare October 31, 2025 10:53

hanno-becker reviewed Nov 2, 2025

View reviewed changes

dev/aarch64_clean/src/intt.S Show resolved Hide resolved

hanno-becker reviewed Nov 2, 2025

View reviewed changes

dev/aarch64_clean/src/intt.S Show resolved Hide resolved

mkannwischer mentioned this pull request Nov 2, 2025

AArch64 iNTT: Fix unaligned twiddle loads #590

Closed

mkannwischer added 3 commits November 2, 2025 05:07

Lint: Remove license exceptions for ntt/intt

520e0ed

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

hanno-becker force-pushed the ntt-again branch from 655aaa6 to 520e0ed Compare November 2, 2025 05:07

hanno-becker approved these changes Nov 2, 2025

View reviewed changes

mkannwischer mentioned this pull request Nov 2, 2025

RELICENSE: Add frame for gathering consent to relicense AArch64 ASM #236

Closed

mkannwischer merged commit d518e10 into main Nov 2, 2025
247 checks passed

mkannwischer deleted the ntt-again branch November 2, 2025 05:36

mkannwischer mentioned this pull request Nov 2, 2025

Adjust LICENSE after AArch64 iNTT/NTT reimplementation #593

Merged

Reimplement AArch64 NTT/iNTT #570

Reimplement AArch64 NTT/iNTT #570

Uh oh!

Conversation

mkannwischer commented Oct 30, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions bot left a comment

Choose a reason for hiding this comment

Mac Mini (M1, 2020) benchmarks (opt)

Uh oh!

github-actions bot left a comment

Choose a reason for hiding this comment

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

Uh oh!

github-actions bot left a comment

Choose a reason for hiding this comment

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

Uh oh!

github-actions bot left a comment

Choose a reason for hiding this comment

⚠️ Performance Alert ⚠️

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 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

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

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

Uh oh!

github-actions bot left a comment

Choose a reason for hiding this comment

Arm Cortex-A55 (Snapdragon 888) 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

Graviton2

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

Graviton4

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

Graviton4 (no-opt)

Uh oh!

oqs-bot left a comment

Choose a reason for hiding this comment

Graviton2 (no-opt)

Uh oh!

oqs-bot left a comment

mkannwischer commented Oct 30, 2025 •

edited

Loading