Skip to content

Conversation

@junhaha666
Copy link
Contributor

No description provided.

Copilot AI review requested due to automatic review settings July 21, 2025 03:04
Copy link
Contributor

Copilot AI left a comment

Choose a reason for hiding this comment

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

Pull Request Overview

This PR fixes API inconsistencies in the GEMM A8W8 bpreshuffle operation by standardizing type annotations and updating the configuration lookup logic. The changes focus on improving type consistency and modifying the fallback behavior for unsupported configurations.

  • Standardizes type annotations from generic Tensor to torch.Tensor and adds return type annotation
  • Updates configuration caching to support multiple tuned files simultaneously
  • Removes fallback to ASM implementation and adds explicit assertions for supported data types
Comments suppressed due to low confidence (1)

aiter/ops/gemm_op_a8w8.py:36

  • Parameter name 'Out' uses inconsistent capitalization. It should be 'out' to match the original naming convention and align with the variable 'Y' used in the function body.
    Out: torch.Tensor,

Copy link
Collaborator

@valarLip valarLip left a comment

Choose a reason for hiding this comment

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

LGTM

@valarLip valarLip merged commit 0439b31 into main Jul 21, 2025
13 checks passed
@valarLip valarLip deleted the gemm_a8w8_bpreshuffle_api_fix branch July 21, 2025 06:34
cagrikymk pushed a commit that referenced this pull request Jul 30, 2025
* add check before gemm_a8w8_bpreshuffle_ck

* disable gemm_a8w8_ASM in gemm_a8w8_bpreshuffle api

* fix get_CKGEMM_config
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