Skip to content

mxfp8 (for all gemm layouts) is not supported on 120+ arch yet #1939

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.

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

Open
wants to merge 5 commits into
base: main
Choose a base branch
from

Conversation

sudhakarsingh27
Copy link
Collaborator

@sudhakarsingh27 sudhakarsingh27 commented Jul 9, 2025

Description

MXFP8 is not supported on 120+ arch yet (for all gemm layouts). Add check to not default to MXFP8 recipe for those archs.

Fixes # (internal)

Type of change

  • Documentation change (change only to the documentation, either a fix or a new content)
  • Bug fix (non-breaking change which fixes an issue)
  • New feature (non-breaking change which adds functionality)
  • Breaking change (fix or feature that would cause existing functionality to not work as expected)
  • Infra/Build change
  • Code refactoring

Changes

Please list the changes introduced in this PR:

  • check_mxfp8_support: returns False for arcs 120+
  • get_default_fp8_recipe: uses check_mxfp8_support instead of directly checking get_device_compute_capability.

Checklist:

  • I have read and followed the contributing guidelines
  • The functionality is complete
  • I have commented my code, particularly in hard-to-understand areas
  • I have made corresponding changes to the documentation
  • My changes generate no new warnings
  • I have added tests that prove my fix is effective or that my feature works
  • New and existing unit tests pass locally with my changes

Signed-off-by: Sudhakar Singh <sudhakars@nvidia.com>
timmoon10
timmoon10 previously approved these changes Jul 9, 2025
@@ -64,7 +66,7 @@ def check_fp8_block_scaling_support() -> Tuple[bool, str]:

def get_default_fp8_recipe() -> Recipe:
"""FP8 recipe with default args."""
if get_device_compute_capability() >= (10, 0): # blackwell and above
if check_mxfp8_support()[0]:
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Let's use the subchannel fp8 recipe for default for 10+ archs that don't have mxfp8 support

Copy link
Member

@ksivaman ksivaman left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Default recipe

Signed-off-by: Sudhakar Singh <sudhakars@nvidia.com>
@sudhakarsingh27 sudhakarsingh27 changed the title mxfp8 is not supported on 120+ arch yet mxfp8 (for all gemm layouts) is not supported on 120+ arch yet Jul 9, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants