Skip to content

Conversation

ysiraichi
Copy link
Collaborator

@ysiraichi ysiraichi commented Aug 23, 2025

This PR removes the OpenXLA CUDA fallback implementation, and also the _XLA_cuda_functions.so extension, completely. Starting on this PR, the fallback shall be run only on CPU.

Key Changes:

  • Remove aten_cuda_functions.cpp and aten_cuda_functions.h
  • Remove the OpenXLA CUDA fallback functions from aten_fallback.cpp
  • Remove the _XLAC_cuda_functions.so library from BUILD
  • Remove the Python _XLAC_cuda_functions.so extension from setup.py
  • Remove the conditional loading of _XLAC_cuda_functions.so Python extension from torch_xla/init.py

Copy link
Collaborator

@zhanyong-wan zhanyong-wan left a comment

Choose a reason for hiding this comment

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

Great! Thank you!

@ysiraichi ysiraichi merged commit e9a1c5f into master Aug 25, 2025
24 checks passed
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.

2 participants