NVIDIA
Add clang21 to CI.
Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually.
Contributors can view more details about this message here.
![copy-pr-bot[bot] avatar](https://avatars.githubusercontent.com/in/354886?v=4 "Published by a pub.dev verified publisher"){kind=small}
😬 CI Workflow Results
🟥 Finished in 2h 15m: Pass: 91%/244 | Total: 2d 21h | Max: 1h 18m | Hits: 97%/294222
See results here.
![github-actions[bot] avatar](https://avatars.githubusercontent.com/in/15368?v=4 "Published by a pub.dev verified publisher"){kind=small}
We love our unsupported compiler thingy https://github.com/NVIDIA/cccl/actions/runs/18144608121/job/51643523655?pr=6083#step:4:533
I dont know how we can circumvent that for CMAKE
I dont know how we can circumvent that for CMAKE
I made it work locally in the past by setting CUDAFLAGS=-allow-unsupported-compiler
Yeah, we can disable those checks, but at the same time, I see little value in spending CI time testing configurations that are unsupported 🙂
@alliepiper I think it would make sense, because from what I've seen from the bug reports, the support is being tested internally and if test new compilers in advance, we can catch errors earlier.
Plus we can test newer version of clang-cuda which comes with support for newer CUDA features, too :)
the support is being tested internally and if test new compilers in advance, we can catch errors earlier.
Could we test them as part of the cuda 99.8 CI images? Do we even run them? They appear in matrix.yaml but seem to not be used by any job.
