Repository navigation
Conversation
Since CUDA 13, nvvm is shipped in a separate libnvvm package, so cuda_nvcc has no nvvm/bin. The glob then expands to nothing and cp fails with a missing destination operand, breaking any build that sets ptxasVersion to a 13.x release.
drbh
added a commit
to huggingface/kernels-community
that referenced
this pull request
Oct 30, 2026
Temporary pin to torch-2.15.0-rc5 plus huggingface/kernels#954, so the 13.x ptxas overrides can build.
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
since cuda 13, nvvm ships in a separate
libnvvmpackage, socuda_nvcchas nonvvm/bin. whenptxasVersionpoints at a 13.x release, the glob in the ptxas override expands to nothing and thecuda_nvccbuild fails:this pr only copies
nvvm/binwhen it exists. no version defaults change, this just unblocks kernel-specific ptxas overrides like huggingface/kernels-community#1285 (failing job: https://github.com/huggingface/kernels-community/actions/runs/40942703764/job/123690362546)this is the nvvm part of #952, which was closed in favor of the kernel-specific override. it was tested locally there: cuda 13.0 nvcc with ptxas 13.4 on aarch64-linux, and cuda 13.2 nvcc with ptxas 13.4 on x86_64-linux both build