Skip to content

Conversation

@LoserCheems
Copy link
Collaborator

Summary

  • Introduces benchmark tests for the gemm operation across multiple backends (Python, PyTorch, Triton, and Cute) to evaluate performance on different devices and data types.

Root Cause

  • The need for performance comparison across various implementations of the gemm operation prompted the addition of these benchmarks.

Changes

  • Added a new test suite for gemm benchmarks, including parameterized tests for different devices and data types.

Reproduction

  • Run the benchmark tests using the provided test suite to compare performance across implementations.

Tests

  • Added tests for the gemm operation benchmarks, confirming successful execution across supported backends.

Compatibility

  • No migration concerns or backwards compatibility issues identified.

Checklist

Introduces parameterized GEMM benchmarks that compare multiple backend implementations across devices and dtypes to validate performance regressions early
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