Thanks to visit codestin.com
Credit goes to github.com

Skip to content

Conversation

@kuhar
Copy link
Member

@kuhar kuhar commented Oct 10, 2025

Reverts #21922.

This PR re-introduced matmul failures on rdna4.

@krzysz00 krzysz00 enabled auto-merge (squash) October 10, 2025 22:04
@krzysz00 krzysz00 merged commit cdce159 into main Oct 10, 2025
50 checks passed
@krzysz00 krzysz00 deleted the revert-21922-rdna4-proper-fix branch October 10, 2025 22:30
@krzysz00
Copy link
Contributor

Ok, so, to use this PR as the ticket

  1. This isn't a recent regression - the problem also occurs at the point where I landed the MLIR-side change to use fences
  2. This wasn't caught before because I think this test didn't exist before
  3. It works with -O0
  4. It's almost certainly the case that dropping the MMRA metadata fixes it (the revert also takes one to s.barrier instead of s.barrier.signal/s.barrier.wait
  5. (Interestingly, the change to barriers had some other knock-on effects)
  6. Time to go check what a nightly wheel of IREE includes to see if it works well for creating a reproducer (when combined with executable substitution)

@krzysz00
Copy link
Contributor

Poked at packaging up a reproducer and it turns out the O0 is nondeterministically a failure.

krzysz00 added a commit to krzysz00/iree that referenced this pull request Oct 15, 2025
This reverts commit cdce159.

This re-applies iree-org#21922.

This reverts iree-org#22278.

The changes in iree-org#22198 appear to have solved the problem.
krzysz00 added a commit that referenced this pull request Oct 16, 2025
This reverts commit cdce159.

This re-applies #21922.

This reverts #22278.

The changes in #22198 appear to have solved the problem.

ci-extra:test_torch
weidel-p pushed a commit to weidel-p/iree that referenced this pull request Oct 21, 2025
Reverts iree-org#21922.

This PR re-introduced matmul failures on rdna4.

Signed-off-by: Philipp <[email protected]>
weidel-p pushed a commit to weidel-p/iree that referenced this pull request Oct 21, 2025
…2326)

This reverts commit cdce159.

This re-applies iree-org#21922.

This reverts iree-org#22278.

The changes in iree-org#22198 appear to have solved the problem.

ci-extra:test_torch
Signed-off-by: Philipp <[email protected]>
newling added a commit to newling/iree that referenced this pull request Oct 23, 2025
newling added a commit to newling/iree that referenced this pull request Oct 23, 2025
pstarkcdpr pushed a commit to pstarkcdpr/iree that referenced this pull request Nov 28, 2025
Reverts iree-org#21922.

This PR re-introduced matmul failures on rdna4.
pstarkcdpr pushed a commit to pstarkcdpr/iree that referenced this pull request Nov 28, 2025
…2326)

This reverts commit cdce159.

This re-applies iree-org#21922.

This reverts iree-org#22278.

The changes in iree-org#22198 appear to have solved the problem.

ci-extra:test_torch
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants