CI: drop container in test-cuda

[njzjz]

Summary by CodeRabbit

• Chores
  • Updated the CUDA test workflow to install CUDA components directly on the runner instead of using a fixed container image.
  • The workflow now determines the operating system version at runtime, downloads the appropriate NVIDIA package sources, and installs the current CUDA toolkit and cuDNN for the environment.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 82.21%. Comparing base (d67e75b) to head (310afb8).
⚠️ Report is 288 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #5543      +/-   ##
==========================================
- Coverage   82.46%   82.21%   -0.26%     
==========================================
  Files         830      892      +62     
  Lines       88414   101562   +13148     
  Branches     4232     4242      +10     
==========================================
+ Hits        72913    83497   +10584     
- Misses      14215    16762    +2547     
- Partials     1286     1303      +17     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[njzjz]

This PR saves about 15~20 minutes, so I think it's worth merging.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro

Run ID: 890147e7-b255-488e-be4b-1e99edae079f

📥 Commits

Reviewing files that changed from the base of the PR and between 3b2a769 and 310afb8.

📒 Files selected for processing (1)

• .github/workflows/test_cuda.yml

🚧 Files skipped from review as they are similar to previous changes (1)

• .github/workflows/test_cuda.yml

---

📝 Walkthrough

Walkthrough

The CUDA CI workflow drops the nvidia/cuda:12.9.1-cudnn-devel-ubuntu22.04 Docker container in favor of running directly on the nvidia runner. A new step dynamically determines the Ubuntu version, downloads the corresponding NVIDIA CUDA keyring .deb, and installs cuda-toolkit-12-9 and cudnn9-cuda-12, replacing the previous hardcoded CUDA 12.3 and libcudnn8 installation.

Changes

CUDA CI Provisioning Overhaul

Layer / File(s)	Summary
Drop container, add dynamic CUDA/cuDNN provisioning .github/workflows/test_cuda.yml	Comments out the Docker container block and the prior hardcoded Ubuntu 22.04 keyring + CUDA 12.3/cuDNN 8 install step. Adds a new step that resolves the Ubuntu version at runtime, fetches the NVIDIA CUDA keyring .deb, and installs cuda-toolkit-12-9 plus cudnn9-cuda-12. Also removes the separate sudo/git apt step that was needed only inside the container.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately reflects the main change: removing the container from the test-cuda CI workflow by using an nvidia runner instead.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 1

🧹 Nitpick comments (1)

.github/workflows/test_cuda.yml (1)

46-46: 💤 Low value

Remove stale comment referencing the nvidia image.

This comment # if: false # skip as we use nvidia image is now outdated since the container is no longer in use. Consider removing it to avoid confusion.

Suggested change

          && sudo apt-get -y install cuda-toolkit-12-9 cudnn9-cuda-12
-        # if: false # skip as we use nvidia image

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In @.github/workflows/test_cuda.yml at line 46, The comment on line 46 of the
test_cuda.yml workflow file that states "if: false # skip as we use nvidia
image" is outdated and references an nvidia image that is no longer in use.
Remove this entire comment line to avoid confusion about the current state of
the workflow.

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In @.github/workflows/test_cuda.yml:
- Around line 37-46: The CUDA toolkit is installed at /usr/local/cuda-12.9/ but
the CUDA_PATH environment variable is never exported, causing subsequent
references to $CUDA_PATH/lib64 in LD_LIBRARY_PATH exports (at lines 72 and 82)
to expand to empty strings and create malformed paths. After the
cuda-toolkit-12-9 installation command completes in the run block, add an export
command to set CUDA_PATH=/usr/local/cuda-12.9/ so that the environment variable
is properly defined for the C++ test runtime steps that depend on it.

---

Nitpick comments:
In @.github/workflows/test_cuda.yml:
- Line 46: The comment on line 46 of the test_cuda.yml workflow file that states
"if: false # skip as we use nvidia image" is outdated and references an nvidia
image that is no longer in use. Remove this entire comment line to avoid
confusion about the current state of the workflow.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro

Run ID: 121b70aa-8a54-4fa1-97b9-122f8d1fba56

📥 Commits

Reviewing files that changed from the base of the PR and between d3834e2 and 3b2a769.

📒 Files selected for processing (1)

• .github/workflows/test_cuda.yml