CI: gate SLOTHY and add label to trigger it - #1384
Conversation
L-series
commented
Aug 17, 2026
- CI: Only run SLOTHY when its inputs change
- CI: Allow requesting SLOTHY with the slothy-tests label
SLOTHY starts a c8g.8xlarge for every PR push and merge_group run, regardless of what changed. Gate it on a changed-files job, as in hol_light.yml. Signed-off-by: Andreas Hatziiliou <andreas.hatziiliou@savoirfairelinux.com>
This commit introduces a slothy-tests label for PR's which can manually trigger SLOTHY upon addition. - Resolves #1175 Signed-off-by: Andreas Hatziiliou <andreas.hatziiliou@savoirfairelinux.com>
|
Resolves: #1175 |
|
@L-series, please next time check if someone is already assigned to an issue. I have this half finished locally, but you can take over. |
CBMC Results (ML-DSA-65, REDUCE-RAM)
Full Results (210 proofs)
|
CBMC Results (ML-DSA-44, REDUCE-RAM)
Full Results (210 proofs)
|
CBMC Results (ML-DSA-44)
Full Results (210 proofs)
|
CBMC Results (ML-DSA-87, REDUCE-RAM)
Full Results (210 proofs)
|
CBMC Results (ML-DSA-65)
Full Results (210 proofs)
|
CBMC Results (ML-DSA-87)
Full Results (210 proofs)
|
|
@mkannwischer apologies, its late and i completely glossed over the assignee tab.. Won't do that again. On the topic of SLOTHY however, what do you think of the addition of a per-file gate? I see that most of the time is spent optimizing NTT and iNTT, and so it might make sense to just have a finer grain gate for these two routines (i.e if any other routine is modified, only re-optimize what has been affected) |
No worries! Thank you for your contributions.
We are not planning many more asm changes at this point, so the vast majority of re-runs will be triggered by nix flake changes, I believe. For those we have to re-run for all files anyway. I prefer to keep it simple for now. |