Fix incorrect cast from BF16 to FP32 in SBGEMM#5712
Merged
martin-frbg merged 1 commit intoOpenMathLib:developfrom Mar 27, 2026
Merged
Fix incorrect cast from BF16 to FP32 in SBGEMM#5712martin-frbg merged 1 commit intoOpenMathLib:developfrom
martin-frbg merged 1 commit intoOpenMathLib:developfrom
Conversation
This change fixes a regression in SBGEMM where C is assumed to be BF16, and so unconditionally casts the output to FP32 resulting in incorrect outputs when beta=1.
Contributor
|
@martin-frbg is there a way we can get a patch release for OpenBLAS? |
Collaborator
|
Yes, I'll try to do 0.3.33 this weekend - though the spurious utest failure bothers me a lot less than the weird DDOT bug that my workaround for the weird Neoverse SDOT bug introduced. Coincidentally the Reference-LAPACK team is committing a flurry of fixes for their upcoming 3.13, so it looks like a good chance to import those as well. |
Contributor
|
that'll be much appreciated! 🙏 |
Contributor
|
I think this is what I've been pointing out for a while now. |
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
4 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
This change fixes a regression in SBGEMM where C is assumed to be BF16, and so unconditionally attempts to cast elements of C from BF16 to FP32 resulting in incorrect outputs when beta=1.