Skip to content

ci : add ccache action to windows-cublas job #2893

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

Merged
merged 14 commits into from
Mar 19, 2025

Conversation

danbev
Copy link
Collaborator

@danbev danbev commented Mar 17, 2025

This commit adds the ccache action to the windows-cublas job. This will allow us to cache the build artifacts and hopefully speed up the build process.

danbev added 14 commits March 17, 2025 16:12
This commit adds the ccache action to the windows-cublas job. This will
allow us to cache the build artifacts and hopefully speed up the build
process.
This commit adds a directory to store the cuda compilation cache and
also caches this directory in github actions.

The motivation is to reduce the compilation time of the CUDA code. The
ccache should still help with the host part of the compilation but it
did not help with the CUDA code.

This also addes the --parallel (-j) flag to the cmake build command to
use all the available cores.
This commit add sccache compiler launcher for windows-cublas-ci to
hopefully enable the sccache to be used.
This commit updates the CUDA_CACHE_PATH to use backslashes for Windows.
This commit moves the installation of the CUDA Toolkit before
configuring sccache. This is to aviod it being cached.
This commit removes the dependency on Jimver/cuda-toolkit and manually
installs the CUDA toolkits.
@danbev danbev marked this pull request as ready for review March 18, 2025 19:28
@danbev danbev merged commit d2aaffd into ggml-org:master Mar 19, 2025
47 checks passed
@danbev
Copy link
Collaborator Author

danbev commented Mar 19, 2025

I spent quite some time trying to get sccache working with nvcc.exe but without success. I'll give it another try and also I'll clean up the current windows-cublas job in a separate PR. I'm not doing it in this PR as this improves the build times as is, the windows-cublas are down to around 1.5 from around 3.5, which is nice to get in sooner rather than later. Also the iterations with github actions is somewhat slow especially for windows as I don't have a windows environment set up yet.

@ggerganov
Copy link
Member

Nice work - this is already quite helpful.

buxuku pushed a commit to buxuku/whisper.cpp that referenced this pull request Mar 26, 2025
* ci : add ccache action to windows-cublas job

This commit adds the ccache action to the windows-cublas job. This will
allow us to cache the build artifacts and hopefully speed up the build
process.

Refs: ggml-org#2781
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.

2 participants