sign stack usage: compute z incrementally #825

mkannwischer · 2025-12-31T03:47:39Z

This commit reduces the stack usage of signing by computing z = y + s1*cp
incrementally (one polynomial at a time) allowing to eliminate the polyvecl
z (at to cost of a single poly z).
The computation of z is moved into a separate function (compute_pack_z) to
vastly speed up the CBMC proofs.

De-facto this saves L-1 KB irrespective of MLD_CONFIG_REDUCE_RAM.

Practically, the same buffer was used early in the function too. Here we
instead introduce a new polyvecl buffer tmp, but that can be placed in a union
together with w1.
Unfortuantely, with the current struct workaround for
diffblue/cbmc#8813, this results in an increase in
stack space by L KB.
This gets eliminated when MLD_CONFIG_REDUCE_RAM is set.

Hoisted out from [TEST] Reduce crypto_sign_signature_internal stack usage #791
Depends on Introduce mld_polymat_get_row() helper function #742
Alternative to sign stack usage: compute z incrementally #821 (which does not move computation of z to a separate function)

Alternative to #822 that I hope to be less controversial. Currently the constant time tests for verification rely on the signature being declassified at the end of verification. This is not ideal. This commit moves this declassification to the constant-time test instead. As suggested in #822 (review), there is more work left to clean up the story around declassifications. This PR is a first step towards cleaning up that story to unblock #825 and #821, but there is more work left. Signed-off-by: Matthias J. Kannwischer <matthias@kannwischer.eu>

mldsa/src/packing.c

mldsa/src/sign.c

hanno-becker

Looks very good overall, thank you @mkannwischer! Some smaller comments

mldsa/src/sign.c

This commit reduces the stack usage of signing by computing z = y + s1*cp incrementally (one polynomial at a time) allowing to eliminate the polyvecl z (at to cost of a single poly z). The computation of z is moved into a separate function (compute_pack_z) to vastly speed up the CBMC proofs. De-facto this saves L-1 KB irrespective of MLD_CONFIG_REDUCE_RAM. Practically, the same buffer was used early in the function too. Here we instead introduce a new polyvecl buffer tmp, but that can be placed in a union together with w1. Unfortuantely, with the current struct workaround for diffblue/cbmc#8813, this results in an increase in stack space by L KB. This gets eliminated when MLD_CONFIG_REDUCE_RAM is set. Hoisted out from #791 Signed-off-by: Matthias J. Kannwischer <matthias@kannwischer.eu>

For some reason the previous (unrelated) commit caused the verify_pre_hash_internal proof to fail due to extra functions in USE_FUNCTION_CONTRACTS. This commit removes the extra functions. Signed-off-by: Matthias J. Kannwischer <matthias@kannwischer.eu>

github-actions

Mac Mini (M1, 2020) benchmarks (opt)

Details

Benchmark suite	Current: `d28fb9f`	Previous: `75ab215`	Ratio
`ML-DSA-44 keypair`	`46412` cycles	`46417` cycles	`1.00`
`ML-DSA-44 sign`	`132123` cycles	`131302` cycles	`1.01`
`ML-DSA-44 verify`	`47772` cycles	`47780` cycles	`1.00`
`ML-DSA-65 keypair`	`81327` cycles	`81321` cycles	`1.00`
`ML-DSA-65 sign`	`217249` cycles	`216907` cycles	`1.00`
`ML-DSA-65 verify`	`80053` cycles	`80071` cycles	`1.00`
`ML-DSA-87 keypair`	`132572` cycles	`132544` cycles	`1.00`
`ML-DSA-87 sign`	`278635` cycles	`277775` cycles	`1.00`
`ML-DSA-87 verify`	`130415` cycles	`130361` 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: `d28fb9f`	Previous: `75ab215`	Ratio
`ML-DSA-44 keypair`	`114381` cycles	`114381` cycles	`1`
`ML-DSA-44 sign`	`418716` cycles	`428553` cycles	`0.98`
`ML-DSA-44 verify`	`122380` cycles	`122432` cycles	`1.00`
`ML-DSA-65 keypair`	`195803` cycles	`195935` cycles	`1.00`
`ML-DSA-65 sign`	`683828` cycles	`696798` cycles	`0.98`
`ML-DSA-65 verify`	`197606` cycles	`197557` cycles	`1.00`
`ML-DSA-87 keypair`	`323082` cycles	`322999` cycles	`1.00`
`ML-DSA-87 sign`	`866048` cycles	`879409` cycles	`0.98`
`ML-DSA-87 verify`	`328303` cycles	`328164` 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: `d28fb9f`	Previous: `75ab215`	Ratio
`ML-DSA-44 keypair`	`276165` cycles	`277431` cycles	`1.00`
`ML-DSA-44 sign`	`818209` cycles	`828306` cycles	`0.99`
`ML-DSA-44 verify`	`274469` cycles	`274630` cycles	`1.00`
`ML-DSA-65 keypair`	`472161` cycles	`472268` cycles	`1.00`
`ML-DSA-65 sign`	`1338336` cycles	`1350849` cycles	`0.99`
`ML-DSA-65 verify`	`449216` cycles	`449649` cycles	`1.00`
`ML-DSA-87 keypair`	`804093` cycles	`802074` cycles	`1.00`
`ML-DSA-87 sign`	`1813859` cycles	`1808884` cycles	`1.00`
`ML-DSA-87 verify`	`774374` cycles	`771390` cycles	`1.00`

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

oqs-bot

Intel Xeon 4th gen (c7i)

Details

Benchmark suite	Current: `d28fb9f`	Previous: `75ab215`	Ratio
`ML-DSA-44 keypair`	`34600` cycles	`34698` cycles	`1.00`
`ML-DSA-44 sign`	`121014` cycles	`120452` cycles	`1.00`
`ML-DSA-44 verify`	`38094` cycles	`38255` cycles	`1.00`
`ML-DSA-65 keypair`	`61083` cycles	`61720` cycles	`0.99`
`ML-DSA-65 sign`	`202128` cycles	`200582` cycles	`1.01`
`ML-DSA-65 verify`	`62534` cycles	`62863` cycles	`0.99`
`ML-DSA-87 keypair`	`94058` cycles	`93574` cycles	`1.01`
`ML-DSA-87 sign`	`235645` cycles	`231081` cycles	`1.02`
`ML-DSA-87 verify`	`93724` cycles	`93987` cycles	`1.00`

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

oqs-bot

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

Details

Benchmark suite	Current: `d28fb9f`	Previous: `75ab215`	Ratio
`ML-DSA-44 keypair`	`94261` cycles	`96015` cycles	`0.98`
`ML-DSA-44 sign`	`333473` cycles	`344139` cycles	`0.97`
`ML-DSA-44 verify`	`100019` cycles	`100919` cycles	`0.99`
`ML-DSA-65 keypair`	`161116` cycles	`160878` cycles	`1.00`
`ML-DSA-65 sign`	`546754` cycles	`548011` cycles	`1.00`
`ML-DSA-65 verify`	`161418` cycles	`161909` cycles	`1.00`
`ML-DSA-87 keypair`	`268049` cycles	`267516` cycles	`1.00`
`ML-DSA-87 sign`	`715711` cycles	`710718` cycles	`1.01`
`ML-DSA-87 verify`	`269969` cycles	`269917` cycles	`1.00`

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

oqs-bot

AMD EPYC 3rd gen (c6a)

Details

Benchmark suite	Current: `d28fb9f`	Previous: `75ab215`	Ratio
`ML-DSA-44 keypair`	`69242` cycles	`69243` cycles	`1.00`
`ML-DSA-44 sign`	`188051` cycles	`184050` cycles	`1.02`
`ML-DSA-44 verify`	`69161` cycles	`69175` cycles	`1.00`
`ML-DSA-65 keypair`	`119740` cycles	`119178` cycles	`1.00`
`ML-DSA-65 sign`	`303540` cycles	`294814` cycles	`1.03`
`ML-DSA-65 verify`	`115187` cycles	`115292` cycles	`1.00`
`ML-DSA-87 keypair`	`204195` cycles	`204047` cycles	`1.00`
`ML-DSA-87 sign`	`396708` cycles	`387431` cycles	`1.02`
`ML-DSA-87 verify`	`195232` cycles	`195361` cycles	`1.00`

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

oqs-bot

Intel Xeon 3rd gen (c6i)

Details

Benchmark suite	Current: `d28fb9f`	Previous: `75ab215`	Ratio
`ML-DSA-44 keypair`	`56501` cycles	`56515` cycles	`1.00`
`ML-DSA-44 sign`	`182196` cycles	`179924` cycles	`1.01`
`ML-DSA-44 verify`	`61023` cycles	`61223` cycles	`1.00`
`ML-DSA-65 keypair`	`98891` cycles	`99047` cycles	`1.00`
`ML-DSA-65 sign`	`301535` cycles	`296975` cycles	`1.02`
`ML-DSA-65 verify`	`100427` cycles	`100792` cycles	`1.00`
`ML-DSA-87 keypair`	`152820` cycles	`153381` cycles	`1.00`
`ML-DSA-87 sign`	`357365` cycles	`353572` cycles	`1.01`
`ML-DSA-87 verify`	`153222` cycles	`153278` cycles	`1.00`

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

oqs-bot

Graviton4

Details

Benchmark suite	Current: `d28fb9f`	Previous: `75ab215`	Ratio
`ML-DSA-44 keypair`	`68812` cycles	`68650` cycles	`1.00`
`ML-DSA-44 sign`	`203587` cycles	`202247` cycles	`1.01`
`ML-DSA-44 verify`	`70760` cycles	`70601` cycles	`1.00`
`ML-DSA-65 keypair`	`121769` cycles	`121601` cycles	`1.00`
`ML-DSA-65 sign`	`334123` cycles	`331430` cycles	`1.01`
`ML-DSA-65 verify`	`118199` cycles	`117927` cycles	`1.00`
`ML-DSA-87 keypair`	`199203` cycles	`198920` cycles	`1.00`
`ML-DSA-87 sign`	`430244` cycles	`428106` cycles	`1.00`
`ML-DSA-87 verify`	`194450` cycles	`195053` 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: `d28fb9f`	Previous: `75ab215`	Ratio
`ML-DSA-44 keypair`	`135410` cycles	`135092` cycles	`1.00`
`ML-DSA-44 sign`	`527859` cycles	`534269` cycles	`0.99`
`ML-DSA-44 verify`	`147490` cycles	`147219` cycles	`1.00`
`ML-DSA-65 keypair`	`227108` cycles	`226958` cycles	`1.00`
`ML-DSA-65 sign`	`862532` cycles	`870870` cycles	`0.99`
`ML-DSA-65 verify`	`235084` cycles	`235062` cycles	`1.00`
`ML-DSA-87 keypair`	`371417` cycles	`371508` cycles	`1.00`
`ML-DSA-87 sign`	`1082333` cycles	`1086997` cycles	`1.00`
`ML-DSA-87 verify`	`382983` cycles	`383200` 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: `d28fb9f`	Previous: `75ab215`	Ratio
`ML-DSA-44 keypair`	`157796` cycles	`159017` cycles	`0.99`
`ML-DSA-44 sign`	`550650` cycles	`568055` cycles	`0.97`
`ML-DSA-44 verify`	`169552` cycles	`170439` cycles	`0.99`
`ML-DSA-65 keypair`	`268363` cycles	`268942` cycles	`1.00`
`ML-DSA-65 sign`	`906244` cycles	`916893` cycles	`0.99`
`ML-DSA-65 verify`	`273642` cycles	`274044` cycles	`1.00`
`ML-DSA-87 keypair`	`448947` cycles	`448757` cycles	`1.00`
`ML-DSA-87 sign`	`1161143` cycles	`1171199` cycles	`0.99`
`ML-DSA-87 verify`	`458213` cycles	`457924` cycles	`1.00`

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

oqs-bot

Graviton3

Details

Benchmark suite	Current: `d28fb9f`	Previous: `75ab215`	Ratio
`ML-DSA-44 keypair`	`72708` cycles	`72739` cycles	`1.00`
`ML-DSA-44 sign`	`214077` cycles	`212398` cycles	`1.01`
`ML-DSA-44 verify`	`75636` cycles	`75721` cycles	`1.00`
`ML-DSA-65 keypair`	`128015` cycles	`128130` cycles	`1.00`
`ML-DSA-65 sign`	`354152` cycles	`350108` cycles	`1.01`
`ML-DSA-65 verify`	`125570` cycles	`125368` cycles	`1.00`
`ML-DSA-87 keypair`	`206752` cycles	`209477` cycles	`0.99`
`ML-DSA-87 sign`	`448406` cycles	`450700` cycles	`0.99`
`ML-DSA-87 verify`	`205849` cycles	`205401` cycles	`1.00`

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

oqs-bot

Graviton4 (no-opt)

Details

Benchmark suite	Current: `d28fb9f`	Previous: `75ab215`	Ratio
`ML-DSA-44 keypair`	`128947` cycles	`128725` cycles	`1.00`
`ML-DSA-44 sign`	`449000` cycles	`456379` cycles	`0.98`
`ML-DSA-44 verify`	`136818` cycles	`137029` cycles	`1.00`
`ML-DSA-65 keypair`	`220984` cycles	`220722` cycles	`1.00`
`ML-DSA-65 sign`	`729869` cycles	`737657` cycles	`0.99`
`ML-DSA-65 verify`	`221208` cycles	`221346` cycles	`1.00`
`ML-DSA-87 keypair`	`365833` cycles	`365707` cycles	`1.00`
`ML-DSA-87 sign`	`929448` cycles	`939501` cycles	`0.99`
`ML-DSA-87 verify`	`370148` cycles	`370387` cycles	`1.00`

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

oqs-bot

AMD EPYC 4th gen (c7a)

Details

Benchmark suite	Current: `d28fb9f`	Previous: `75ab215`	Ratio
`ML-DSA-44 keypair`	`41523` cycles	`42364` cycles	`0.98`
`ML-DSA-44 sign`	`134242` cycles	`133559` cycles	`1.01`
`ML-DSA-44 verify`	`44664` cycles	`45481` cycles	`0.98`
`ML-DSA-65 keypair`	`72408` cycles	`72453` cycles	`1.00`
`ML-DSA-65 sign`	`216861` cycles	`211467` cycles	`1.03`
`ML-DSA-65 verify`	`73309` cycles	`73397` cycles	`1.00`
`ML-DSA-87 keypair`	`107598` cycles	`110157` cycles	`0.98`
`ML-DSA-87 sign`	`251916` cycles	`254019` cycles	`0.99`
`ML-DSA-87 verify`	`109156` cycles	`111808` cycles	`0.98`

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

oqs-bot

Graviton2

Details

Benchmark suite	Current: `d28fb9f`	Previous: `75ab215`	Ratio
`ML-DSA-44 keypair`	`114000` cycles	`113905` cycles	`1.00`
`ML-DSA-44 sign`	`358680` cycles	`360308` cycles	`1.00`
`ML-DSA-44 verify`	`117919` cycles	`117940` cycles	`1.00`
`ML-DSA-65 keypair`	`197620` cycles	`197696` cycles	`1.00`
`ML-DSA-65 sign`	`594164` cycles	`594054` cycles	`1.00`
`ML-DSA-65 verify`	`194968` cycles	`194953` cycles	`1.00`
`ML-DSA-87 keypair`	`324651` cycles	`324147` cycles	`1.00`
`ML-DSA-87 sign`	`758979` cycles	`759328` cycles	`1.00`
`ML-DSA-87 verify`	`320806` cycles	`320484` cycles	`1.00`

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: `d28fb9f`	Previous: `75ab215`	Ratio
`ML-DSA-44 keypair`	`465085` cycles	`464882` cycles	`1.00`
`ML-DSA-44 sign`	`2150114` cycles	`2213472` cycles	`0.97`
`ML-DSA-44 verify`	`547454` cycles	`547002` cycles	`1.00`
`ML-DSA-65 keypair`	`779566` cycles	`780337` cycles	`1.00`
`ML-DSA-65 sign`	`3521130` cycles	`3593147` cycles	`0.98`
`ML-DSA-65 verify`	`848308` cycles	`848702` cycles	`1.00`
`ML-DSA-87 keypair`	`1256100` cycles	`1257888` cycles	`1.00`
`ML-DSA-87 sign`	`4343431` cycles	`4428098` cycles	`0.98`
`ML-DSA-87 verify`	`1363073` cycles	`1361393` cycles	`1.00`

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

oqs-bot

Graviton3 (no-opt)

Details

Benchmark suite	Current: `d28fb9f`	Previous: `75ab215`	Ratio
`ML-DSA-44 keypair`	`139121` cycles	`138888` cycles	`1.00`
`ML-DSA-44 sign`	`485471` cycles	`492916` cycles	`0.98`
`ML-DSA-44 verify`	`148368` cycles	`148562` cycles	`1.00`
`ML-DSA-65 keypair`	`243026` cycles	`242771` cycles	`1.00`
`ML-DSA-65 sign`	`796203` cycles	`804857` cycles	`0.99`
`ML-DSA-65 verify`	`240904` cycles	`241160` cycles	`1.00`
`ML-DSA-87 keypair`	`396786` cycles	`397265` cycles	`1.00`
`ML-DSA-87 sign`	`1016444` cycles	`1026974` cycles	`0.99`
`ML-DSA-87 verify`	`401593` cycles	`402066` cycles	`1.00`

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

oqs-bot

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

Details

Benchmark suite	Current: `d28fb9f`	Previous: `75ab215`	Ratio
`ML-DSA-44 keypair`	`122581` cycles	`120166` cycles	`1.02`
`ML-DSA-44 sign`	`455680` cycles	`452709` cycles	`1.01`
`ML-DSA-44 verify`	`131381` cycles	`130091` cycles	`1.01`
`ML-DSA-65 keypair`	`204249` cycles	`204195` cycles	`1.00`
`ML-DSA-65 sign`	`728522` cycles	`738873` cycles	`0.99`
`ML-DSA-65 verify`	`210149` cycles	`209224` cycles	`1.00`
`ML-DSA-87 keypair`	`337896` cycles	`337432` cycles	`1.00`
`ML-DSA-87 sign`	`925559` cycles	`934927` cycles	`0.99`
`ML-DSA-87 verify`	`346908` cycles	`346862` cycles	`1.00`

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

oqs-bot

Graviton2 (no-opt)

Details

Benchmark suite	Current: `d28fb9f`	Previous: `75ab215`	Ratio
`ML-DSA-44 keypair`	`213689` cycles	`213754` cycles	`1.00`
`ML-DSA-44 sign`	`763922` cycles	`779981` cycles	`0.98`
`ML-DSA-44 verify`	`229483` cycles	`229404` cycles	`1.00`
`ML-DSA-65 keypair`	`381937` cycles	`381467` cycles	`1.00`
`ML-DSA-65 sign`	`1256384` cycles	`1282011` cycles	`0.98`
`ML-DSA-65 verify`	`371810` cycles	`371341` cycles	`1.00`
`ML-DSA-87 keypair`	`607103` cycles	`605515` cycles	`1.00`
`ML-DSA-87 sign`	`1601574` cycles	`1618590` cycles	`0.99`
`ML-DSA-87 verify`	`618524` cycles	`617714` cycles	`1.00`

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: `d28fb9f`	Previous: `75ab215`	Ratio
`ML-DSA-44 keypair`	`244241` cycles	`227155` cycles	`1.08`
`ML-DSA-44 sign`	`661391` cycles	`634926` cycles	`1.04`
`ML-DSA-44 verify`	`235482` cycles	`223942` cycles	`1.05`
`ML-DSA-65 keypair`	`395832` cycles	`396222` cycles	`1.00`
`ML-DSA-65 sign`	`1036776` cycles	`1033955` cycles	`1.00`
`ML-DSA-65 verify`	`373209` cycles	`382272` cycles	`0.98`
`ML-DSA-87 keypair`	`692726` cycles	`662437` cycles	`1.05`
`ML-DSA-87 sign`	`1433908` cycles	`1393666` cycles	`1.03`
`ML-DSA-87 verify`	`670754` cycles	`647777` cycles	`1.04`

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: `d28fb9f`	Previous: `75ab215`	Ratio
`ML-DSA-44 keypair`	`244241` cycles	`227155` cycles	`1.08`
`ML-DSA-44 sign`	`661391` cycles	`634926` cycles	`1.04`
`ML-DSA-44 verify`	`235482` cycles	`223942` cycles	`1.05`
`ML-DSA-87 keypair`	`692726` cycles	`662437` cycles	`1.05`
`ML-DSA-87 verify`	`670754` cycles	`647777` 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)

Details

Benchmark suite	Current: `d28fb9f`	Previous: `75ab215`	Ratio
`ML-DSA-44 keypair`	`315610` cycles	`331789` cycles	`0.95`
`ML-DSA-44 sign`	`1176470` cycles	`1306541` cycles	`0.90`
`ML-DSA-44 verify`	`344598` cycles	`363529` cycles	`0.95`
`ML-DSA-65 keypair`	`568926` cycles	`607535` cycles	`0.94`
`ML-DSA-65 sign`	`1954395` cycles	`2078057` cycles	`0.94`
`ML-DSA-65 verify`	`541671` cycles	`577791` cycles	`0.94`
`ML-DSA-87 keypair`	`857076` cycles	`893687` cycles	`0.96`
`ML-DSA-87 sign`	`2427084` cycles	`2605336` cycles	`0.93`
`ML-DSA-87 verify`	`892479` cycles	`928223` cycles	`0.96`

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: `d28fb9f`	Previous: `75ab215`	Ratio
`ML-DSA-44 keypair`	`831908` cycles	`830625` cycles	`1.00`
`ML-DSA-44 sign`	`3243025` cycles	`3342511` cycles	`0.97`
`ML-DSA-44 verify`	`921924` cycles	`921888` cycles	`1.00`
`ML-DSA-65 keypair`	`1410709` cycles	`1405524` cycles	`1.00`
`ML-DSA-65 sign`	`5330162` cycles	`5453246` cycles	`0.98`
`ML-DSA-65 verify`	`1473728` cycles	`1471261` cycles	`1.00`
`ML-DSA-87 keypair`	`2312729` cycles	`2307928` cycles	`1.00`
`ML-DSA-87 sign`	`6666114` cycles	`6787920` cycles	`0.98`
`ML-DSA-87 verify`	`2410121` cycles	`2408200` cycles	`1.00`

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: `d28fb9f`	Previous: `75ab215`	Ratio
`ML-DSA-44 keypair`	`113886` cycles	`113817` cycles	`1.00`
`ML-DSA-44 sign`	`358367` cycles	`359993` cycles	`1.00`
`ML-DSA-44 verify`	`117782` cycles	`117902` cycles	`1.00`
`ML-DSA-65 keypair`	`197634` cycles	`197711` cycles	`1.00`
`ML-DSA-65 sign`	`593784` cycles	`593720` cycles	`1.00`
`ML-DSA-65 verify`	`194997` cycles	`194920` cycles	`1.00`
`ML-DSA-87 keypair`	`324116` cycles	`323552` cycles	`1.00`
`ML-DSA-87 sign`	`758139` cycles	`758497` cycles	`1.00`
`ML-DSA-87 verify`	`320455` cycles	`320209` cycles	`1.00`

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

github-actions

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

Details

Benchmark suite	Current: `d28fb9f`	Previous: `75ab215`	Ratio
`ML-DSA-44 keypair`	`213239` cycles	`213302` cycles	`1.00`
`ML-DSA-44 sign`	`762799` cycles	`783978` cycles	`0.97`
`ML-DSA-44 verify`	`229058` cycles	`229037` cycles	`1.00`
`ML-DSA-65 keypair`	`381608` cycles	`381483` cycles	`1.00`
`ML-DSA-65 sign`	`1256317` cycles	`1272131` cycles	`0.99`
`ML-DSA-65 verify`	`371520` cycles	`371484` cycles	`1.00`
`ML-DSA-87 keypair`	`606449` cycles	`605847` cycles	`1.00`
`ML-DSA-87 sign`	`1603106` cycles	`1617275` cycles	`0.99`
`ML-DSA-87 verify`	`618254` cycles	`618103` cycles	`1.00`

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

mkannwischer mentioned this pull request Dec 31, 2025

sign stack usage: compute z incrementally #821

Closed

mkannwischer force-pushed the stack-stream-z-2 branch from 0ee075f to 6c05449 Compare December 31, 2025 09:37

mkannwischer marked this pull request as ready for review December 31, 2025 10:13

mkannwischer requested a review from a team as a code owner December 31, 2025 10:13

mkannwischer mentioned this pull request Jan 3, 2026

CT: Declassify sig in test instead of at the end of signing #835

Closed

mkannwischer force-pushed the stack-stream-z-2 branch from 6c05449 to 3026327 Compare January 3, 2026 00:59

mkannwischer force-pushed the stack-stream-z-2 branch 3 times, most recently from b91f0c7 to 5b0673e Compare January 5, 2026 03:35