-
Notifications
You must be signed in to change notification settings - Fork 21.4k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[cuBLAS] Add an option to disable reduced precision reductions for BF16 GEMM #89172
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/89172
Note: Links to docs will display an error until the docs builds have been completed. ⏳ No Failures, 1 PendingAs of commit 909a989: This comment was automatically generated by Dr. CI and updates every 15 minutes. |
@pytorchmergebot rebase |
@pytorchbot successfully started a rebase job. Check the current status here |
Successfully rebased |
e64bf93
to
6133bcc
Compare
@pytorchmergebot rebase |
@pytorchbot successfully started a rebase job. Check the current status here |
Successfully rebased |
2d9b2e3
to
82d6337
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Approving, but I'd defer to @ptrblck for what the default value should be, just making tests pass may not be reason enough to flip the default, as long as real workloads still converge.
@pytorchmergebot rebase |
@pytorchbot successfully started a rebase job. Check the current status here |
Successfully rebased |
81898db
to
909a989
Compare
@pytorchmergebot merge |
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 |
…16 GEMM (pytorch#89172) Essentially the same change as pytorch#67946, except that the default is to disallow reduced precision reductions in `BFloat16` GEMMs (for now). If performance is severely regressed, we can change the default, but this option appears to be necessary to pass some `addmm` `BFloat16` tests on H100. CC @ptrblck @ngimel Pull Request resolved: pytorch#89172 Approved by: https://github.com/ngimel
Essentially the same change as #67946, except that the default is to disallow reduced precision reductions in
BFloat16
GEMMs (for now). If performance is severely regressed, we can change the default, but this option appears to be necessary to pass someaddmm
BFloat16
tests on H100.CC @ptrblck @ngimel