Error during installation: Error: The process '/usr/bin/sudo' failed with exit code 134

[pplmx]

hi,

The cuda cannot be installed correctly by action

Run Jimver/cuda-toolkit@master
  with:
    cuda: 1[2](https://github.com/x-pt/example-cuda-cmake/actions/runs/10771750645/job/29867925725#step:3:2).5.0
    sub-packages: []
    non-cuda-sub-packages: []
    method: local
    linux-local-args: ["--toolkit", "--samples"]
    use-github-cache: true
    use-local-cache: true
    log-file-suffix: log.txt
/usr/bin/tar --posix -cf cache.tzst --exclude cache.tzst -P -C /home/runner/work/example-cuda-cmake/example-cuda-cmake --files-from manifest.txt --use-compress-program zstdmt
Cache Size: ~4087 MB (4285841[3](https://github.com/x-pt/example-cuda-cmake/actions/runs/10771750645/job/29867925725#step:3:3)97 B)
Cache saved successfully
/usr/bin/sudo /home/runner/work/example-cuda-cmake/example-cuda-cmake/cuda_installer-linux-6.5.0-1025-azure-12.5.0/cuda_installer-linux-6.5.0-1025-azure_12.5.0.run --silent --toolkit --samples
terminate called after throwing an instance of 'boost::filesystem::filesystem_error'
  what():  boost::filesystem::create_symlink: No such file or directory: "libnvvm.so.[4](https://github.com/x-pt/example-cuda-cmake/actions/runs/10771750645/job/29867925725#step:3:4).0.0", "/usr/local/cuda-12.[5](https://github.com/x-pt/example-cuda-cmake/actions/runs/10771750645/job/29867925725#step:3:5)/nvvm/lib64/libnvvm.so.4"
Aborted (core dumped)
Warning: Error during installation: Error: The process '/usr/bin/sudo' failed with exit code 134
Artifact name is valid!
Root directory input is valid!
Beginning upload of artifact content to blob storage
Uploaded bytes 972
Finished uploading artifact content to blob storage!
SHA25[6](https://github.com/x-pt/example-cuda-cmake/actions/runs/10771750645/job/29867925725#step:3:6) hash of uploaded artifact zip is 8cfc0d6d[7](https://github.com/x-pt/example-cuda-cmake/actions/runs/10771750645/job/29867925725#step:3:7)d32ba554c5da7deb254b30ee01b242da4a01d81ce25bcfc0a57b042
Finalizing artifact upload

[pplmx]

Retrying can bypass the above error, but the test still won't run — it can only be built.

You can see a similar error here:

CUDA error: CUDA driver version is insufficient for CUDA runtime version at /home/runner/work/example-cuda-cmake/example-cuda-cmake/src/matrix_add.cu:20

[pplmx]

Hi, @Jimver

Do you have any insights on this issue?

After upgrading to version 0.2.17, I’m still encountering the same error, although the error line has shifted:

CUDA error: CUDA driver version is insufficient for CUDA runtime version at src/matrix_add.cu:48

Any guidance would be appreciated. Thanks in advance!

[nazar-pc]

Looks like you need to update driver version like error says due to newer default CUDA version 🤷

[pplmx]

Looks like you need to update driver version like error says due to newer default CUDA version 🤷

Where should I update the version? Is it in the CMakeLists.txt?

It works fine on my local machine, but fails in GitHub Actions when using cuda-toolkit.

[nazar-pc]

Just like I told you above and as error tells you as well, you need to update the driver install on your OS or explicitly specify older version of CUDA that you want.