Repository navigation
More changes for dropping 3.10 / adding 3.15 - #3113
Conversation
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info
📝 Summary
Priority: ➖ Normal Change: Feature Merge Risk: ⚪ Minimal · up to
|
There was a problem hiding this comment.
Actionable comments posted: 1
ℹ️ Review info
⚙️ Run configuration
- Configuration used: Repository: NVIDIA/cuda-python/.coderabbit.yaml
- Review profile: CHILL
- Plan: Enterprise
- Run ID:
71519d4f-cb36-4799-aa1a-f1d7eac9d187
📒 Files selected for processing: 77
.github/actions/check-doc-links/action.yml.github/workflows/build-wheel.yml.github/workflows/test-wheel-linux.yml.github/workflows/test-wheel-windows.ymlci/ci-pipeline.mmdci/tools/download-wheelscuda_bindings/pixi.tomlcuda_bindings/pyproject.tomlcuda_core/AGENTS.mdcuda_core/cuda/core/_cpp/rt/DESIGN.mdcuda_core/cuda/core/_memory/_copy_enums.pycuda_core/cuda/core/_utils/cuda_utils.pyxcuda_core/cuda/core/_utils/pycompat.pycuda_core/cuda/core/system/typing.pycuda_core/pixi.tomlcuda_core/tests/graph/test_graph_builder.pycuda_core/tests/test_enum_coverage.pycuda_core/tests/test_error_handling.pycuda_python_test_helpers/pyproject.tomlsamples/cuda_bindings/0_Introduction/clockNvrtc/README.mdsamples/cuda_bindings/0_Introduction/simpleCubemapTexture/README.mdsamples/cuda_bindings/0_Introduction/systemWideAtomics/README.mdsamples/cuda_bindings/0_Introduction/vectorAddDriverApi/README.mdsamples/cuda_bindings/0_Introduction/vectorAddMmap/README.mdsamples/cuda_bindings/2_Concepts_and_Techniques/streamOrderedAllocation/README.mdsamples/cuda_bindings/3_CUDA_Features/cudaGraphsManualNodes/README.mdsamples/cuda_bindings/3_CUDA_Features/globalToShmemAsyncCopy/README.mdsamples/cuda_bindings/4_CUDA_Libraries/conjugateGradientMultiBlockCG/README.mdsamples/cuda_bindings/4_CUDA_Libraries/nvidiaSmi/README.mdsamples/cuda_bindings/extra/isoFdModelling/README.mdsamples/cuda_bindings/extra/jitProgram/README.mdsamples/cuda_core/binarySearch/README.mdsamples/cuda_core/blockwiseSum/README.mdsamples/cuda_core/blurImageUnifiedMemory/README.mdsamples/cuda_core/copyImageArraytoGPU/README.mdsamples/cuda_core/cudaComputeLambdas/README.mdsamples/cuda_core/cudaGraphs/README.mdsamples/cuda_core/customPyTorchKernel/README.mdsamples/cuda_core/customTensorFlowKernel/README.mdsamples/cuda_core/deviceQuery/README.mdsamples/cuda_core/fftSignalAnalysis/README.mdsamples/cuda_core/glInteropFluid/README.mdsamples/cuda_core/glInteropMipmapLod/README.mdsamples/cuda_core/glInteropPlasma/README.mdsamples/cuda_core/greenContext/README.mdsamples/cuda_core/ipcMemoryPool/README.mdsamples/cuda_core/jitLtoLinking/README.mdsamples/cuda_core/kernelNsysProfile/README.mdsamples/cuda_core/launchConfigTuning/README.mdsamples/cuda_core/launchConfigTuning/requirements.txtsamples/cuda_core/matrixMulSharedMem/README.mdsamples/cuda_core/memoryResources/README.mdsamples/cuda_core/multiGPUGradientAverage/README.mdsamples/cuda_core/pageRank/README.mdsamples/cuda_core/parallelHistogram/README.mdsamples/cuda_core/parallelHistogram/requirements.txtsamples/cuda_core/parallelReduction/README.mdsamples/cuda_core/persistentProgramCache/README.mdsamples/cuda_core/prefixSum/README.mdsamples/cuda_core/prefixSum/requirements.txtsamples/cuda_core/processCheckpoint/README.mdsamples/cuda_core/reduction/README.mdsamples/cuda_core/reductionMultiBlockCG/README.mdsamples/cuda_core/simpleMultiGpu/README.mdsamples/cuda_core/simpleP2P/README.mdsamples/cuda_core/simplePrint/README.mdsamples/cuda_core/simpleZeroCopy/README.mdsamples/cuda_core/streamingCopyComputeOverlap/README.mdsamples/cuda_core/streamingCopyComputeOverlap/requirements.txtsamples/cuda_core/stridedMemoryViewConstructors/README.mdsamples/cuda_core/stridedMemoryViewCpu/README.mdsamples/cuda_core/stridedMemoryViewGpu/README.mdsamples/cuda_core/systemInfo/README.mdsamples/cuda_core/textureSample/README.mdsamples/cuda_core/threadBlockCluster/README.mdsamples/cuda_core/tmaTensorMap/README.mdsamples/cuda_core/vectorAdd/README.md
💤 Files with no reviewable changes: 5
- cuda_bindings/pixi.toml
- cuda_core/tests/test_enum_coverage.py
- cuda_core/pixi.toml
- cuda_bindings/pyproject.toml
- cuda_python_test_helpers/pyproject.toml
Included review availability: This review used your included allowance. Your plan provides up to 12 included reviews per hour; 11 remain after this review.
There was a problem hiding this comment.
Actionable comments posted: 1
ℹ️ Review info
⚙️ Run configuration
- Configuration used: Repository: NVIDIA/cuda-python/.coderabbit.yaml
- Review profile: CHILL
- Plan: Enterprise
- Run ID:
d221f98e-9aed-411f-9400-ef784dabcbf3
📒 Files selected for processing: 7
.github/workflows/test-wheel-linux.yml.github/workflows/test-wheel-windows.ymlcuda_bindings/pyproject.tomlcuda_core/pyproject.tomlcuda_pathfinder/pyproject.tomlcuda_pathfinder/tests/test_find_nvidia_headers.pycuda_python/pyproject.toml
💤 Files with no reviewable changes: 2
- .github/workflows/test-wheel-linux.yml
- .github/workflows/test-wheel-windows.yml
🚧 Files skipped from review as they are similar to previous changes: 1
- cuda_bindings/pyproject.toml
Included review availability: This review used your included allowance. Your plan provides up to 12 included reviews per hour; 8 remain after this review.
|
Andy-Jost
left a comment
There was a problem hiding this comment.
Approving. CI is green, and the 3.10 cleanup and the 3.15 guard removals look right. Three non-blocking follow-ups:
-
No CUDA 12 test job on Python 3.15. All 3.15 jobs run on CUDA 13.4.2 (
ci/test-matrix.ymlL60-L61, L132). So CI never imports the CUDA 12 half of a cp315cuda-corewheel. cuda-bindings 12.9.10 now has cp315 wheels on PyPI, sopip install "cuda-core[cu12]"resolves on 3.15. Only the test coverage is missing. Suggestion: add aPY_VER: '3.15',CUDA_VER: '12.9.1'row. #3108 adds the same coverage on the 1.2.x branch. -
The ml-dtypes cap keeps a guard whose TODO is now met. ml-dtypes 0.6.0 has cp315 and cp315t wheels for
win_amd64,win_arm64, and manylinux x86_64/aarch64. The<0.6.0cap holds CI at 0.5.4, which has no cp315 wheels (cuda_core/pyproject.tomlL84-L85). As a result, the Windows 3.15 job skipstest_bfloat16_mappingwith "ml_dtypes not installed", and the Linux 3.15 jobs build ml-dtypes 0.5.4 from source. If the tests pass with 0.6.x, raise the cap and remove the Windows 3.15 guard and its TODO. -
Stale numpy cap in the cuda-bindings
testgroup.cuda_bindings/pyproject.tomlL48-L49 keepsnumpy>=1.21.1,<=2.6.0.dev0and its comment about scientific-python nightlies. This PR removes the nightly index, and numpy 2.5.4 has cp315 wheels, so the comment no longer applies. The cap will also exclude numpy 2.6.0 when it ships.<2.6gives the same result for releases without the comment. Or remove the cap if CI should run numpy 2.6.
This is a follow-on with a bunch of things that got missed in #3080.