This is the release note of v10.0.0b2. See here for the complete list of solved issues and merged PRs.
We are running a Gitter chat for general discussions and quick questions. Feel free to join the channel to talk with developers and users!
Highlights
Support for CUDA Python (#5638)
CuPy is one of the first libraries providing support for the newly released CUDA Python bindings. To try it, install cuda-python manually and set the CUPY_USE_CUDA_PYTHON=1 environment variable when building CuPy as written in the documentation.
Support for AMD ROCm 4.3
Support for ROCm 4.3 has been added in the latest release and binary wheels are provided as well. Note that there is currently an issue with ROCm 4.3 that prevents it from running in several environments. The current workaround is to set the LLVM_PATH variable to the llvm folder included in ROCm 4.3 installation (e.g., export LLVM_PATH=/opt/rocm-4.3/llvm).
Announcements
Removal of Alpha/Beta/RC Wheels from PyPI
-
As per the discussion in #5671, we will stop uploading pre-release binary wheels to PyPI for the health of the ecosystem. Pre-release wheels can now be downloaded from the assets section of each GitHub release page (e.g.,
pip install cupy-cudaXXX -f https://github.com/cupy/cupy/releases/tag/v10.0.0b2) . Note that the sdist package is available in PyPI for all versions. -
We are also going to remove outdated (v8.0.0rc1 or earlier) pre-release binary wheels from PyPI on September 20th. See #5667 for details.
Changes
New Features
- Support batched QR solver (#5583)
- Add
cupyx.scipy.sparse.linalg.minres(#5585) - Add Log Series distribution to
cupy.random.Generator(#5618) - Add Power distribution to
cupy.random.Generator(#5624) - Add support for CUDA Python (#5638)
- Add Chi-square distribution to
cupy.random.Generator(#5645) - Add Dirichlet distribution to
cupy.random.Generator(#5648) - Add F distribution to
cupy.random.Generator(#5655)
Enhancements
- Add
ncclAvgandncclBfloat16for NCCL (#5545) - Add new eigensolvers from
rocSOLVER(#5555) - Add support for array input in
betadistribution ofcupy.random.Generator(#5573) - Release the GIL for several NCCL ops (#5574)
- Allow to compile using PTX with an envvar (#5622)
- Show CUDA Python version (#5651)
- Fix version check for new ROCm version definition (#5657)
- Rest of version check fix for new ROCm version definition (#5660)
- Add ROCm 4.3 in duplicate detection (#5669)
Bug Fixes
- Fix compute capability check (#5600)
- Fix FFT convolve for shapes containing 1 (#5609)
- Fix squareness checks (#5642)
- Fix
uniquefor empty array (#5654)
Code Fixes
Documentation
- Update Sphinx to 4.1.2 (#5612)
- Fix random docstring (#5628)
- Support ROCm v4.3 in document (#5633)
__array_function__feature by default (#5644)
Tests
- Fix
skipTestintest_decomp_lu(#5593) - Mark
lsmrtests xfail for CSR matrices on HIP (#5597) - Increase test timeout (#5601)
- Fix cubic
for_all_dtypes_combinationtests (#5629) - Add CI for ROCm 4.3 (#5630)
- Reload GPG key for ROCm 4.2 test (#5636)
- Fix branch name of cuda-python (#5650)
- Add a workaround for ROCm 4.3.0 for testing (#5662)
Others
Contributors
The CuPy Team would like to thank all those who contributed to this release!