cuda.core v0.1.1

cuda.core v0.1.1 release announcement

Release note

• https://nvidia.github.io/cuda-python/cuda-core/latest/release/0.1.1-notes.html

All functionalities are currently hosted under the cuda.core.experimental namespace. Once the features become stable they will be moved out of experimental.

Key Features and Enhancements

• Added Linker for runtime linking (using nvJitLink or driver APIs)
• Added StridedMemoryView and @args_viewable_as_strided_memory to arbitrary Python objects that support either DLPack or CUDA Array Interface
• Support pip installation
• Public, GitHub-Action-based CI infrastructure

For full details please refer to the release note above.

Documentation

• https://nvidia.github.io/cuda-python/cuda-core/0.1.1/

Sample codes

• https://github.com/NVIDIA/cuda-python/tree/main/cuda_core/examples

What's Changed

• Fix _util.device_from_ctx by @ksimpson-work in #203
• Update docs to reflect recent releases by @leofang in #239
• rectify the supported types for ObjectCode and Program by @ksimpson-work in #224
• Clean up testsuite by @ksimpson-work in #213
• Fix the build portion of the gh-actions by @ksimpson-work in #249
• fix some test issues caused by bad merging on browser by @ksimpson-work in #254
• Add ruff linter by @ksimpson-work in #201
• update the nvjitlink bindings test by @ksimpson-work in #228
• Systematically replace __del__ with weakref.finalize() by @rwgk in #246
• Add docs, tests, and samples for StridedMemoryView/@args_viewable_as_strided_memory by @leofang in #247
• Convert line endings from CRLF to LF by @trxcllnt in #263
• Some CI tweaks by @leofang in #266
• Full CI support for public builds + switch to use cibuildwheel by @leofang in #267
• Add fallback memory resource for TCC devices by @ksimpson-work in #257
• Add the cuda.core.experimental.Linker class by @ksimpson-work in #229
• Add support for CI testing by @sandeepd-nv in #124
• Add the cuda.core.experimental.system singleton by @ksimpson-work in #256
• Lazy load code modules by @ksimpson-work in #269
• Bump cuda.core version to v0.1.1 by @leofang in #290
• Add GPU runner for linux-aarch64 by @leofang in #289
• Add cluster to LaunchConfig to support thread block clusters on Hopper by @leofang in #261
• Make the link to different object types explicit for the docs by @ksimpson-work in #258
• Add an doc page for interoperatibility by @leofang in #298
• Prepare for wheel release + populate PyPI page information by @leofang in #296
• Fix StridedMemoryView by deferring the check for whether a capsule is versioned by @leofang in #292
• Fix test_is_done without a sync by @vzhurba01 in #305
• CI refactoring to cover more test support by @leofang in #302
• add nvjitlink to bindings documentation by @ksimpson-work in #291
• cuda.core v0.1.1 final doc touch by @leofang in #301

New Contributors

• @rwgk made their first contribution in #246
• @trxcllnt made their first contribution in #263

Full Changelog: cuda-core-v0.1.0...cuda-core-v0.1.1