Skip to content

Add --disable-cuda-check option to bypass strict CUDA version validation in setup.py #1905

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: master
Choose a base branch
from

Conversation

Adhikari-Ashutosh
Copy link

summary:

  • Introduced a --disable-cuda-check pip install option to skip the strict CUDA version check between PyTorch binary and system CUDA.
  • This is useful when minor version mismatches are known to work but block installation due to conservative checks.
  • The default behavior remains unchanged; check is only skipped when the argument is explicitly passed.

changes:

  • Modified setup.py to parse --disable-cuda-check and set a global flag.
  • Updated the RuntimeError message with guidance on using this flag.
  • Documented the flag usage in README.md.

Added a check for sys arg --disable-cuda-check, to bypass the check_cuda_torch_binary_vs_bare_metal check. 
This is not enforced by default as the check does make sense but is an added convenience for users who don't want to modify setup.py everytime they try to build apex
Update Runtime Error
Solved a logical error there
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.

1 participant