s390x: update using vectorization builtins #124396
Conversation
With gcc >= 12 on s390x store builtins are accidentally optimized out due to bad type aliasing. Ensure that proper corresponding types are used, and if types do mismatch, first store data into array of correct type and then memcpy it to destination pointer. See also: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114676
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/124396
Note: Links to docs will display an error until the docs builds have been completed. ❌ 3 New Failures, 1 Unrelated FailureAs of commit 9ab6d41 with merge base 7cd7a7a ( NEW FAILURES - The following jobs have failed:
BROKEN TRUNK - The following job failed but were present on the merge base:👉 Rebase onto the `viable/strict` branch to avoid these failures
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
pytorch-bot
bot
added
the
module: cpu
|
Hmm @malfet - tentatively tagged you, but do you know who a good person would be to look at CPU vectorization changes? |
bdhirsh
added
the
triaged
malfet
added
release notes: performance_as_product
|
@pytorchbot merge |
To be frank, I don't have a good idea what is going on here, but to the best of my knowledge we don't have binary builds nor ways to test it, so anything is fine there, as long as there aren't consequitive PRs that change the same line back and forth |
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
Merge failedReason: 1 mandatory check(s) failed. The first few are: Dig deeper by viewing the failures on hud |
|
@pytorchbot merge -f "Failures are unrelated" |
Merge startedYour change will be merged immediately since you used the force (-f) flag, bypassing any CI checks (ETA: 1-5 minutes). Please use Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
With gcc >= 12 on s390x store builtins are accidentally optimized out due to bad type aliasing. Ensure that proper corresponding types are used, and if types do mismatch, first store data into array of correct type and then memcpy it to destination pointer. See also: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114676 Pull Request resolved: pytorch#124396 Approved by: https://github.com/malfet
With gcc >= 12 on s390x store builtins
are accidentally optimized out due to
bad type aliasing.
Ensure that proper corresponding types are used,
and if types do mismatch,
first store data into array of correct type
and then memcpy it to destination pointer.
See also:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114676
cc @jgong5 @mingfeima @XiaobingSuper @sanchitintel @ashokei @jingxu10