Skip to content

fix(kernels): don't fail all2all build on CUDA 12.8 deprecation warnings - #264

Open
blue (Blueyyds) wants to merge 1 commit into
Lightricks:mainfrom
wgdl666:fix/kernels-cuda-128-build
Open

fix(kernels): don't fail all2all build on CUDA 12.8 deprecation warnings#264
blue (Blueyyds) wants to merge 1 commit into
Lightricks:mainfrom
wgdl666:fix/kernels-cuda-128-build

Conversation

@Blueyyds

Copy link
Copy Markdown

CUDA 12.8's cuSPARSE headers declare legacy API types as deprecated, and PyTorch pulls those declarations in transitively. With -Werror in effect the third-party deprecation turned into a hard build failure for the all2all_cpp extension on our CUDA 12.8 hosts.

Add -Wno-deprecated-declarations so the strict warning policy still applies to this extension's own sources without being hostage to third-party headers.

CUDA 12.8's cuSPARSE headers declare legacy API types as deprecated, and
PyTorch pulls those declarations in transitively. With -Werror in effect
the third-party deprecation turned into a hard build failure for the
all2all_cpp extension on our CUDA 12.8 hosts.

Add -Wno-deprecated-declarations so the strict warning policy still
applies to this extension's own sources without being hostage to
third-party headers.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
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