Repository navigation
Rebuild CPU-specific fbtools wheels on their CI runner - #21
Merged
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The free-boundary merge-group job reused a cached fbtools wheel and stopped with SIGILL before collecting its tests. Pinned fbtools14a5c4a7 defaults to
-march=native, while its ordinary Linux wheel tag does not encode the build CPU ISA. Build that dependency with--no-cache-diron the runner that loads it, and split the three import probes so a future failure names its module. The original combined import does not uniquely identify the faulting module. No dependency revision, test selection or numerical gate changes.Validation: Slurm3362468 rebuilt pinned fbtools on node1, passed each NumPy/fbtools/GVEC import, and passed all47 actual original free-boundary tests with zero skips. Runtime and immutable GVEC-package guards passed before and after (4CPUs,6m26s,1.06GiB MaxRSS). This workflow-only diff reuses exact implementation/test proof from3362376:1314 Python tests and all17 native resident tests. The actual merge-group CI remains the hosted-runner guarantee.
Chris&AI