-
Notifications
You must be signed in to change notification settings - Fork 109
Update release container to CUDA 13.0.1 #1068
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
Conversation
|
/build |
3 similar comments
|
/build |
|
/build |
|
/build |
cliffburdick
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
should we use a variable so it's only updated in one place?
|
/build |
820bde1 to
89e3ef8
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Greptile Overview
Greptile Summary
This PR attempts to update MatX's development and CI infrastructure to use CUDA 13.0.1, modifying three files: .devcontainer/setup.sh sets the default container version tag, .github/workflows/build-docs.yml updates the documentation build workflow container, and .devcontainer/recipe.py changes the base Docker image and CuPy dependency. These changes are part of a coordinated effort to standardize the project's container environment on a newer CUDA version, ensuring consistency across local development (devcontainer setup), CI documentation builds, and Python package dependencies. The changes follow the project's existing container-based development workflow structure, where .devcontainer/ defines the VS Code development container configuration and GitHub Actions workflows rely on pre-built release containers from ghcr.io/nvidia/matx/release.
Important Files Changed
| Filename | Score | Overview |
|---|---|---|
| .devcontainer/recipe.py | 1/5 | Updates base Docker image to nvidia/cuda:13.0.1-devel-ubuntu24.04 and CuPy to cupy-cuda13x, but CUDA 13.0.1 does not exist and will cause build failures |
| .devcontainer/setup.sh | 1/5 | Changes default MATX_VERSION_TAG from 12.9.1_ubuntu24.04 to 13.0.1_ubuntu24.04, referencing a non-existent container version |
| .github/workflows/build-docs.yml | 1/5 | Updates documentation build container image to ghcr.io/nvidia/matx/release:13.0.1_ubuntu24.04-amd64, which does not exist |
Confidence score: 0/5
- This PR will immediately fail and cannot be merged - it references a non-existent CUDA version that will break all container builds and CI workflows
- Score reflects critical blocker issues: CUDA 13.0.1does not exist (current CUDA versions are in the 12.x series), causing Docker image pull failures, pip package installation failures, and complete CI pipeline breakage across all three modified files
- All files require immediate correction - the intended CUDA version should be verified and all references updated to an actual existing CUDA release (likely 12.6.x or similar) before this PR can proceed
Sequence Diagram
sequenceDiagram
participant Dev as Developer
participant GH as GitHub Actions
participant Auth as Blossom Authorization
participant Container as Docker Container<br/>(CUDA 13.0.1)
participant Build as Build System
participant Pages as GitHub Pages
Note over Dev,Pages: Container Setup Flow (.devcontainer/setup.sh)
Dev->>Dev: Run setup.sh
Dev->>Dev: Set MATX_VERSION_TAG="13.0.1_ubuntu24.04"
Dev->>Dev: Detect platform (x86_64/aarch64)
Dev->>Dev: Configure container instance
Note over Dev,Pages: Container Recipe Build Flow (.devcontainer/recipe.py)
Dev->>Container: Pull nvidia/cuda:13.0.1-devel-ubuntu24.04
Container->>Container: Install system packages (clang-tidy, cmake, etc)
Container->>Container: Install GNU toolchain
Container->>Container: Install CMake 3.30.4
Container->>Container: Install Nsight Compute & Systems
Container->>Container: Build & install Doxygen 1.14.0
Container->>Container: Build & install FFTW 3.3.10 (float & double)
Container->>Container: Install Coveralls
Container->>Container: Build & install BLIS 1.0
Container->>Container: Setup fixuid for user matx (uid:2000)
Container->>Container: Create Python venv at /opt/nvidia/venv
Container->>Container: Install Python packages (cupy-cuda13x, sphinx, etc)
Note over Dev,Pages: Documentation Build & Deploy Flow (.github/workflows/build-docs.yml)
Dev->>GH: Push to main branch
GH->>Auth: Check authorization
Auth->>Auth: Verify actor in approved list
Auth-->>GH: Authorization granted
GH->>Container: Pull ghcr.io/nvidia/matx/release:13.0.1_ubuntu24.04-amd64
GH->>Container: Checkout repository
Container->>Build: Create build directory
Build->>Build: Run cmake with MATX_BUILD_DOCS=ON via /opt/nvidia/run_from_venv.sh
Build->>Build: Run make to build documentation
Build-->>GH: Documentation artifacts (build/docs_input/sphinx/)
GH->>Pages: Configure GitHub Pages
GH->>Pages: Upload documentation artifact
GH->>Pages: Deploy to GitHub Pages
Pages-->>Dev: Documentation URL available
3 files reviewed, 2 comments
|
/build |
Greptile is wrong, CUDA 13.0.1 most certainly does exist. |
No description provided.