Build and test with CUDA 13.2.0

[bdice]

Contributes to rapidsai/build-planning#265

• uses CUDA 13.2.0 to build and test
• updates to CUDA 13.2.0 devcontainers

Notes for Reviewers

This switches GitHub Actions workflows to the cuda-13.2.0 branch from here: rapidsai/shared-workflows#545

A future round of PRs will revert that back to main, once all of RAPIDS is migrated.

[copy-pr-bot]

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.

[greptile-apps]

Greptile Summary

Mechanical bump from CUDA 13.1 to 13.2 across devcontainers, CI workflows, and conda environments. All shared-workflow refs are temporarily pinned to @cuda-13.2.0 as noted in the PR description, with a planned revert to @main once the broader RAPIDS migration completes.

• Devcontainers: Both cuda13.1-conda and cuda13.1-pip directories renamed to cuda13.2-*; all internal CUDA version strings, container names, and BASE image tags updated consistently.
• CI workflows: Every rapidsai/shared-workflows uses: reference across build.yaml, pr.yaml, test.yaml, and trigger-breaking-change-alert.yaml switched from @main to @cuda-13.2.0.
• Conda/deps: Environment files renamed from cuda-131 to cuda-132; dependencies.yaml adds new matrix entries for cuda: \"13.2\" in both the conda cuda-version pin and the pip cuda-toolkit wheel selector.

Confidence Score: 5/5

Safe to merge — all changes are a consistent, mechanical CUDA version bump with no logic modifications.

Every change is a direct string substitution of 13.1 → 13.2 across config, workflow, and environment files. The temporary pinning of shared-workflow refs to the cuda-13.2.0 branch is explicitly acknowledged and intentional. No build logic, API surface, or runtime behavior is altered.

No files require special attention.

Important Files Changed

Filename	Overview
.devcontainer/cuda13.2-conda/devcontainer.json	Renamed from cuda13.1-conda; all CUDA 13.1 references updated to 13.2 consistently.
.devcontainer/cuda13.2-pip/devcontainer.json	Renamed from cuda13.1-pip; CUDA version, BASE image tag, and feature version all updated to 13.2.
.github/workflows/build.yaml	All shared-workflow refs switched from @main to @cuda-13.2.0 (temporary branch, acknowledged in PR description).
.github/workflows/pr.yaml	All shared-workflow refs switched to @cuda-13.2.0; devcontainer cuda input updated from 13.1 to 13.2.
.github/workflows/test.yaml	All shared-workflow refs switched to @cuda-13.2.0; no other changes.
.github/workflows/trigger-breaking-change-alert.yaml	Single shared-workflow ref updated to @cuda-13.2.0.
dependencies.yaml	CUDA matrix updated from 13.1 to 13.2; new conda and pip dependency matrix entries added for cuda 13.2.
conda/environments/all_cuda-132_arch-x86_64.yaml	Renamed from cuda-131; cuda-version pin and environment name updated to 13.2.
conda/environments/all_cuda-132_arch-aarch64.yaml	Renamed from cuda-131; cuda-version pin and environment name updated to 13.2.
python/cugraph-pyg/conda/cugraph_pyg_dev_cuda-132_arch-aarch64.yaml	Renamed from cuda-131; environment name updated to cuda-132.
python/cugraph-pyg/conda/cugraph_pyg_dev_cuda-132_arch-x86_64.yaml	Renamed from cuda-131; environment name updated to cuda-132.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[PR Trigger] --> B[pr.yaml]
    B --> C[shared-workflows cuda-13.2.0 branch]
    C --> D[conda-cpp-build]
    C --> E[wheel-build jobs]
    C --> F[devcontainer build\ncuda 13.2]
    D --> G[conda-python-build]
    D --> H[conda-cpp-tests]
    G --> I[conda-python-tests]
    E --> J[wheel-tests jobs]

    subgraph envs [CUDA 13.2 Conda Environments]
        K[all_cuda-132_arch-x86_64\ncuda-version=13.2]
        L[all_cuda-132_arch-aarch64\ncuda-version=13.2]
        M[cugraph_pyg_dev_cuda-132]
    end

    D -.uses.-> K
    D -.uses.-> L
    G -.uses.-> M

Loading

_{Reviews (3): Last reviewed commit: "Rerun CI" | Re-trigger Greptile}

[jameslamb]

Nice! Happy to see that this just worked, I'd expected PyTorch-related dependencies to need updates on conda-forge.

Thanks very much @h-vetinari @traversaro @carterbox @mgorny for all your hard work in https://github.com/conda-forge/pytorch-cpu-feedstock 😊

[jameslamb]

/merge