[Bugfix] Only use triton_kernels for MXFP4 on SM90 and SM100 #29339
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
Signed-off-by: mgoin <[email protected]>
mgoin
requested review from
pavanimajety,
robertgshaw2-redhat,
tlrmchlsmth and
yewentao256
as code owners
mgoin
changed the title
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Code Review
This pull request correctly restricts the use of Triton kernels for MXFP4 to SM90 and SM100 architectures, addressing issues on newer GPUs. The logic is sound, but I've pointed out a small redundancy in the conditional check that can be simplified for better code clarity and maintainability.
Signed-off-by: mgoin <[email protected]>
Contributor
|
LGTM! Thanks for the fix @mgoin |
varun-sundar-rabindranath
approved these changes
Nov 24, 2025
yewentao256
approved these changes
Nov 24, 2025
Member
yewentao256
left a comment
yewentao256
left a comment
There was a problem hiding this comment.
LGTM, thanks for the work!
github-project-automation
bot
moved this from To Triage
to Ready
in gpt-oss Issues & Enhancements
github-project-automation
bot
moved this from Ready
to Done
in gpt-oss Issues & Enhancements
bringlein
pushed a commit
to bringlein/vllm
that referenced
this pull request
Nov 26, 2025
…oject#29339) Signed-off-by: mgoin <[email protected]>
devpatelio
pushed a commit
to SumanthRH/vllm
that referenced
this pull request
Nov 29, 2025
…oject#29339) Signed-off-by: mgoin <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 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Outdated suggestions cannot be applied.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.
Purpose
After we landed triton_kernels as a default dep in #28788, we started seeing some issues appear for gpt-oss users on sm110 and sm120 (see #29317 and https://vllm-dev.slack.com/archives/C0990U53QBV/p1764009918189409).
This PR disables that path for now and uses Marlin for those GPUs.
Test Plan
Test Result
Essential Elements of an Effective PR Description Checklist
supported_models.mdandexamplesfor a new model.