Fix quantize kernels on rocm 6.4 #4708
Closed
+12
−10
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
✅ Deploy Preview for pytorch-fbgemm-docs ready!
To edit notification comments on pull requests, go to your Netlify project configuration. |
|
No ciflow labels are configured for this repo. |
|
This pull request was exported from Phabricator. Differential Revision: D80309166 |
jwfromm
added a commit
to jwfromm/FBGEMM
that referenced
this pull request
Aug 15, 2025
Summary: X-link: facebookresearch/FBGEMM#1731 Interestingly, ROCM6.4 technically allows both OCP and FNUZ floating point formats. We have a check in our quantize kernels that sees if OCP formats are allowed and uses them if so. However, for pretty much any integration, FNUZ is still expected. This small diff fixes the behavior by checking env vars more carefully and exposes rowwise quantization on AMD to unit tests. Reviewed By: q10 Differential Revision: D80309166
jwfromm
force-pushed
the
export-D80309166
branch
from
33f7e9f to
d09b525
Compare
|
This pull request was exported from Phabricator. Differential Revision: D80309166 |
Summary: X-link: facebookresearch/FBGEMM#1731 Interestingly, ROCM6.4 technically allows both OCP and FNUZ floating point formats. We have a check in our quantize kernels that sees if OCP formats are allowed and uses them if so. However, for pretty much any integration, FNUZ is still expected. This small diff fixes the behavior by checking env vars more carefully and exposes rowwise quantization on AMD to unit tests. Reviewed By: q10 Differential Revision: D80309166
jwfromm
force-pushed
the
export-D80309166
branch
from
d09b525 to
a5778ea
Compare
|
This pull request was exported from Phabricator. Differential Revision: D80309166 |
|
This pull request has been merged in 2d4e455. |
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.
Summary: Interestingly, ROCM6.4 technically allows both OCP and FNUZ floating point formats. We have a check in our quantize kernels that sees if OCP formats are allowed and uses them if so. However, for pretty much any integration, FNUZ is still expected. This small diff fixes the behavior by checking env vars more carefully and exposes rowwise quantization on AMD to unit tests.
Differential Revision: D80309166