Skip to content

Cutlass Upgrade to 4.8 - #3443

Open
vthumbe1503 wants to merge 4 commits into
NVIDIA:mainfrom
vthumbe1503:cutlass_update
Open

Cutlass Upgrade to 4.8#3443
vthumbe1503 wants to merge 4 commits into
NVIDIA:mainfrom
vthumbe1503:cutlass_update

Conversation

@vthumbe1503

Copy link
Copy Markdown
Collaborator

Description

Please include a brief summary of the changes, relevant motivation and context.

Fixes # (issue)

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:

  • Change A
  • Change B

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: Varun Thumbe <vthumbe@nvidia.com>
@vthumbe1503 vthumbe1503 changed the title Cutlass Upgrade Cutlass Upgrade to 4.8 Aug 31, 2026
@vthumbe1503

Copy link
Copy Markdown
Collaborator Author

/te-ci L1

@greptile-apps

greptile-apps Bot commented Aug 31, 2026

Copy link
Copy Markdown
Contributor

Greptile Summary

Updates the Cutlass submodule and adapts SM100 Hadamard-transform kernels to Cutlass 4.8’s TMEM capacity namespace.

  • Advances the bundled Cutlass revision.
  • Uses cute::TMEM::Sm100TmemCapacityColumns consistently for TMEM allocation and deallocation across five fused CUDA kernels.

Confidence Score: 5/5

The PR appears safe to merge because no blocking failure remains.

No blocking failure remains.

Important Files Changed

Filename Overview
3rdparty/cutlass Advances the Cutlass submodule revision for the 4.8 upgrade.
transformer_engine/common/hadamard_transform/graph_safe_group_row_cast_col_hadamard_transform_cast_fusion.cu Updates TMEM allocation and deallocation to use the Cutlass 4.8 capacity namespace.
transformer_engine/common/hadamard_transform/group_hadamard_transform_cast_fusion.cu Updates the grouped Hadamard kernel’s matching TMEM allocation and free calls.
transformer_engine/common/hadamard_transform/group_row_cast_col_hadamard_transform_cast_fusion.cu Adapts grouped row/column Hadamard TMEM management to the upgraded Cutlass API.
transformer_engine/common/hadamard_transform/hadamard_transform_cast_fusion.cu Adapts the base fused Hadamard kernel to the upgraded TMEM constant location.
transformer_engine/common/hadamard_transform/row_cast_col_hadamard_transform_cast_fusion.cu Updates row/column Hadamard TMEM allocation and release to use the new namespace.

Reviews (3): Last reviewed commit: "Merge branch 'main' into cutlass_update" | Re-trigger Greptile

@ptrendx ptrendx added the 2.20 label Aug 31, 2026
Signed-off-by: Varun Thumbe <vthumbe@nvidia.com>
@vthumbe1503

Copy link
Copy Markdown
Collaborator Author

/te-ci L1

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants