NVIDIA/numba-cuda v0.21.0

v0.21.0

Repository: NVIDIA/numba-cuda

Tag: v0.21.0

Published: 2025-11-20T14:56:10Z

Prerelease: no

Release notes:

What's Changed

• [Refactor][NFC] Vendor-in more cpython registries for future CUDA-specific changes by @atmnp in https://github.com/NVIDIA/numba-cuda/pull/478
• [Refactor][NFC] Vendor-in numba.extending for future CUDA-specific changes by @atmnp in https://github.com/NVIDIA/numba-cuda/pull/466
• [test] Use numpy's tolerance for float16 by @ashermancinelli in https://github.com/NVIDIA/numba-cuda/pull/491
• Change dangling imports of numba.core.lowering to numba.cuda.lowering by @VijayKandiah in https://github.com/NVIDIA/numba-cuda/pull/475
• [test] Remove dependency on cpu_target by @ashermancinelli in https://github.com/NVIDIA/numba-cuda/pull/490
• Vendor in optional, boxing for CUDA-specific changes, fix dangling imports by @VijayKandiah in https://github.com/NVIDIA/numba-cuda/pull/476
• [Refactor][NFC] Vendor in numba.misc for CUDA-specific changes by @VijayKandiah in https://github.com/NVIDIA/numba-cuda/pull/457
• Vendor in core.registry for CUDA-specific changes by @VijayKandiah in https://github.com/NVIDIA/numba-cuda/pull/485
• Add compile_all API by @isVoid in https://github.com/NVIDIA/numba-cuda/pull/484
• Improve debug value range coverage by @jiel-nv in https://github.com/NVIDIA/numba-cuda/pull/461
• [MNT] Managed UM memset fallback and skip CUDA IPC tests on WSL2 by @rparolin in https://github.com/NVIDIA/numba-cuda/pull/488
• [Refactor] Vendor in _dispatcher, _devicearray, mviewbuf C extension for CUDA-specific customization by @VijayKandiah in https://github.com/NVIDIA/numba-cuda/pull/373
• Vendor in components from numba.core.runtime for CUDA-specific changes by @VijayKandiah in https://github.com/NVIDIA/numba-cuda/pull/498
• Fix Bf16 Test OB Error by @isVoid in https://github.com/NVIDIA/numba-cuda/pull/509
• [Refactor][NFC] Vendor-in core Numba analysis utils for CUDA-specific changes by @atmnp in https://github.com/NVIDIA/numba-cuda/pull/433
• Ensure NUMBA can manipulate memory from CUDA graphs before the graph is launched by @caugonnet in https://github.com/NVIDIA/numba-cuda/pull/437
• Remove C extension loading hacks by @gmarkall in https://github.com/NVIDIA/numba-cuda/pull/506
• Make the CUDA target the default for CUDA overload decorators by @gmarkall in https://github.com/NVIDIA/numba-cuda/pull/511
• [Refactor][NFC] Vendor-in numba.np modules for future CUDA-specific changes by @atmnp in https://github.com/NVIDIA/numba-cuda/pull/494
• [Refactor][NFC] Vendor-in numba.cpython modules for future CUDA-specific changes by @atmnp in https://github.com/NVIDIA/numba-cuda/pull/493
• Vendor in typeconv for future CUDA-specific changes by @VijayKandiah in https://github.com/NVIDIA/numba-cuda/pull/499
• Vendor in _helperlib cext for CUDA-specific changes by @VijayKandiah in https://github.com/NVIDIA/numba-cuda/pull/512
• Remove some unnecessary uses of ContextResettingTestCase by @gmarkall in https://github.com/NVIDIA/numba-cuda/pull/507
• Updating .gitignore with binaries in the testing folder by @rparolin in https://github.com/NVIDIA/numba-cuda/pull/516
• Switch back to stable cuDF release in thirdparty tests by @brandon-b-miller in https://github.com/NVIDIA/numba-cuda/pull/518
• Don't use MemoryLeakMixin for tests that don't use NRT by @gmarkall in https://github.com/NVIDIA/numba-cuda/pull/519
• Vendor the imputils module for CUDA refactoring by @ashermancinelli in https://github.com/NVIDIA/numba-cuda/pull/448
• chore(dev-deps): add pixi by @cpcloud in https://github.com/NVIDIA/numba-cuda/pull/505
• build: allow parallelization of nvcc testing builds by @cpcloud in https://github.com/NVIDIA/numba-cuda/pull/521
• Vendor in dispatcher, entrypoints, pretty_annotate for CUDA-specific changes by @VijayKandiah in https://github.com/NVIDIA/numba-cuda/pull/502
• [MNT] Drop NUMBA_CUDA_USE_NVIDIA_BINDING; always use cuda.core and cuda.bindings as fallback by @rparolin in https://github.com/NVIDIA/numba-cuda/pull/479
• Use numba.core.config when available, otherwise use numba.cuda.core.config by @VijayKandiah in https://github.com/NVIDIA/numba-cuda/pull/497
• Vendor in numba.core.typing for CUDA-specific changes by @VijayKandiah in https://github.com/NVIDIA/numba-cuda/pull/473
• pyproject.toml: add search path for Pyrefly by @gmarkall in https://github.com/NVIDIA/numba-cuda/pull/524
• Draft: Vendor in the IR module by @ashermancinelli in https://github.com/NVIDIA/numba-cuda/pull/439
• refactor: fully remove USE_NV_BINDING by @cpcloud in https://github.com/NVIDIA/numba-cuda/pull/525
• test(pixi): update pixi testing command to work with the new testing directory by @cpcloud in https://github.com/NVIDIA/numba-cuda/pull/522
• test: add benchmarks for kernel launch for reproducibility by @cpcloud in https://github.com/NVIDIA/numba-cuda/pull/528
• [Refactor][NFC] Vendor-in vectorize decorators for future CUDA-specific changes by @atmnp in https://github.com/NVIDIA/numba-cuda/pull/513
• refactor: remove unnecessary custom map and set implementations by @cpcloud in https://github.com/NVIDIA/numba-cuda/pull/530
• perf: reduce the number of __cuda_array_interface__ accesses by @cpcloud in https://github.com/NVIDIA/numba-cuda/pull/538
• perf: remove context threading in various pointer abstractions by @cpcloud in https://github.com/NVIDIA/numba-cuda/pull/536
• perf: speed up kernel launch by @cpcloud in https://github.com/NVIDIA/numba-cuda/pull/510
• test: speed up ipc tests by ~6.5x by @cpcloud in https://github.com/NVIDIA/numba-cuda/pull/527
• chore(perf): add torch to benchmark by @cpcloud in https://github.com/NVIDIA/numba-cuda/pull/539
• perf: remove duplicated size computation by @cpcloud in https://github.com/NVIDIA/numba-cuda/pull/537
• perf: cache dimension computations by @cpcloud in https://github.com/NVIDIA/numba-cuda/pull/542
• bench: add cupy to array constructor kernel launch benchmarks by @cpcloud in https://github.com/NVIDIA/numba-cuda/pull/547
• refactor: cleanup device constructor by @cpcloud in https://github.com/NVIDIA/numba-cuda/pull/548
• Vendor in types and datamodel for CUDA-specific changes by @VijayKandiah in https://github.com/NVIDIA/numba-cuda/pull/533
• feat: add support for math.exp2 by @kaeun97 in https://github.com/NVIDIA/numba-cuda/pull/541
• Handle cuda.core.Stream in driver operations by @brandon-b-miller in https://github.com/NVIDIA/numba-cuda/pull/401
• ci: add timeout to avoid blocking the job queue by @cpcloud...