8000 Fix Assertion for Rowwise Scaling by petrex · Pull Request #161243 · pytorch/pytorch · GitHub
[go: up one dir, main page]
Skip to content

Fix Assertion for Rowwise Scaling #161243

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.

Sign up for GitHub

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

Jump to bottom
Closed
wants to merge 1 commit into from
Closed

Fix Assertion for Rowwise Scaling #161243

wants to merge 1 commit into from

Conversation

@petrex
Copy link
Contributor
@petrex petrex commented Aug 22, 2025 •
edited
Loading

This pull request updates the assertion logic for rowwise scaling support in the scaled_gemm function, clarifying compatibility with CUDA versions.

forward-fix : #151360
fix pytorch/ao#2843

Compatibility check improvement:

  • In aten/src/ATen/cuda/CUDABlas.cpp, the assertion for rowwise scaling now explicitly checks for CUDA version 12.9 or above, providing a clearer error message about support limitations for scaled_gemm with rowwise scaling.

@petrex petrex requested review from eqy and syed-ahmed as code owners August 22, 2025 03:43
@pytorch-bot
Copy link
pytorch-bot bot commented Aug 22, 2025 •
edited
Loading

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/161243

Note: Links to docs will display an error until the docs builds have been completed.

✅ No Failures

As of commit fef116d with merge base f5bf514 (image):
💚 Looks good so far! There are no failures yet. 💚

This comment was automatically generated by Dr. CI and updates every 15 minutes.

@petrex
Copy link
Contributor Author
petrex commented Aug 22, 2025

FYI @drisspg @weifengpy

@petrex
Copy link
Contributor Author
petrex commented Aug 22, 2025

@pytorchbot label "topic: not user facing"

@pytorch-bot pytorch-bot bot added the topic: not user facing topic category label Aug 22, 2025
@petrex petrex mentioned this pull request Aug 22, 2025
Open
@eqy
Copy link
Collaborator
eqy commented Aug 23, 2025 •
edited
Loading

I think the tests also need to be adjust for CUDA due to dtype support matrix chagnes, see also https://github.com/pytorch/pytorch/pull/161305/

If you can add that we can go ahead and land this

@petrex
Copy link
Contributor Author
petrex commented Aug 27, 2025

I think the tests also need to be adjust for CUDA due to dtype support matrix chagnes, see also https://github.com/pytorch/pytorch/pull/161305/

If you can add that we can go ahead and land this

Thanks, but could you clarify which test you're referring to? CI looks green to me at the moment.

@jeffdaily
Copy link
Collaborator
jeffdaily commented Oct 14, 2025

No longer needed.

@jeffdaily jeffdaily closed this Oct 14, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@weifengpy weifengpy weifengpy approved these changes

@eqy eqy Awaiting requested review from eqy

@syed-ahmed syed-ahmed Awaiting requested review from syed-ahmed

Assignees

No one assigned

Labels

open source topic: not user facing topic category

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

RuntimeError: use_rowwise == false INTERNAL ASSERT FAILED at "/pytorch/aten/src/ATen/cuda/CUDABlas.cpp":1971

5 participants

@petrex @eqy @jeffdaily @weifengpy @pytorchbot
0