Add rounding mode options to mx_github #2933
Closed
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
Summary: We found the rounding mode when selecting the block scales in MX formats is critical to numeric accuracies. Josh has added rounding mode options in the fp32 to mx4 conversion kernels in D59562029. Here is to add the options to mx_github library which supports more mx formats. To use different options from the default 'floor' method in the cuda kernel, need to specify custom_cuda=False in the mx specs. Differential Revision: D60682029
✅ Deploy Preview for pytorch-fbgemm-docs ready!
To edit notification comments on pull requests, go to your Netlify site configuration. |
|
This pull request was exported from Phabricator. Differential Revision: D60682029 |
|
This pull request has been merged in b0c16d7. |
q10
pushed a commit
to q10/FBGEMM
that referenced
this pull request
Apr 10, 2025
Summary: Pull Request resolved: facebookresearch/FBGEMM#34 X-link: pytorch#2933 We found the rounding mode when selecting the block scales in MX formats is critical to numeric accuracies. Josh has added rounding mode options in the fp32 to mx4 conversion kernels in D59562029. Here is to add the options to mx_github library which supports more mx formats. To use different options from the default 'floor' method in the cuda kernel, need to specify custom_cuda=False in the mx specs. Reviewed By: jwfromm Differential Revision: D60682029 fbshipit-source-id: 561d668b96cdcf02cd4b9673c3dffe523cf4d577
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Outdated suggestions cannot be applied.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: We found the rounding mode when selecting the block scales in MX formats is critical to numeric accuracies. Josh has added rounding mode options in the fp32 to mx4 conversion kernels in D59562029. Here is to add the options to mx_github library which supports more mx formats. To use different options from the default 'floor' method in the cuda kernel, need to specify custom_cuda=False in the mx specs.
Differential Revision: D60682029