[ROCm] Refine from_recipe_name to support mxfp8 on rocm.#3620
Merged
vkuzo merged 2 commits intopytorch:mainfrom Jan 15, 2026
Merged
[ROCm] Refine from_recipe_name to support mxfp8 on rocm.#3620vkuzo merged 2 commits intopytorch:mainfrom
vkuzo merged 2 commits intopytorch:mainfrom
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/ao/3620
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 60a5e32 with merge base 4b3ebc4 ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
meta-cla
bot
added
the
CLA Signed
vkuzo
reviewed
Jan 12, 2026
| return MXLinearConfig( | ||
| kernel_preference=KernelPreference.AUTO, | ||
| mxfp8_cast_kernel_choice=MXFP8Dim1CastKernelChoice.CUDA, | ||
| mxfp8_cast_kernel_choice=MXFP8Dim1CastKernelChoice.TRITON |
Contributor
There was a problem hiding this comment.
style nit: can we rewrite as below to improve readability
# add a descriptive comment here
mxfp8_cast_kernel_choice = MXFP8Dim1CastKernelChoice.TRITON is if_ROCM() else MXFP8Dim1CastKernelChoice.CUDA
return MXLinearConfig(..., mxfp8_cast_kernel_choice)
Contributor
|
lgtm, could we update the style per my comment |
Contributor
Author
Thanks for your suggestion. But the ruff-format will modify the code if I follow your style. I can't find a method to disable ruff-format in a line. Do you have any further suggestions? |
vkuzo
added
the
topic: improvement
Contributor
Author
|
Hi @vkuzo, I saw my PR was blocked by Dr.CI. Could you help me trigger it ? |
tianyu-l
pushed a commit
to pytorch/torchtitan
that referenced
this pull request
Jan 20, 2026
* Support mxfp8 on gfx950. It depends on TorchAO (pytorch/ao#3620).
jcaip
pushed a commit
that referenced
this pull request
Jan 22, 2026
* [ROCm] Refine from_recipe_name to support mxfp8 on rocm. * resolve reviewer issue
wwwjn
pushed a commit
to wwwjn/torchtitan
that referenced
this pull request
Jan 30, 2026
* Support mxfp8 on gfx950. It depends on TorchAO (pytorch/ao#3620).
xrsrke
pushed a commit
to NousResearch/torchtitan
that referenced
this pull request
Feb 13, 2026
* Support mxfp8 on gfx950. It depends on TorchAO (pytorch/ao#3620).
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
2 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.
Summary
Support mxfp8 on gfx950 by refine
MXLinearConfig.from_recipe_nameIt will encounter error when called triton's quantize kernel with rceil on rocm because PTX instructions was called in kernel. So we chose implementation of
torch.compileto workaround.TODO