* Propagate generated path finder changes * Rebase to 13.0 RC14 * Update license headers on examples * Regenerate after merging upstream * Update benchmarks and run pre-commit * cython-gen output, NO manual changes * Bump cuda/bindings/_version.py → 13.0.0 * `path_finder` and cybind updates for CTK 13.0 (#81) * Update SUPPORTED_WINDOWS_DLLS: kitpicks/cuda-r13-0/13.0.0/013/local_installers/cuda_13.0.0_windows.exe * Update SUPPORTED_LINUX_SONAMES: kitpicks/cuda-r13-0/13.0.0/013/local_installers/cuda_13.0.0_580.31_linux.run * 013 → 014: SUPPORTED_LINUX_SONAMES unchanged * 013 → 014: SUPPORTED_WINDOWS_DLLS unchanged * cybind update with 13.0.0 headers (014) * Bump cuda/bindings/_version.py → 13.0.0 * test_nvjitlink.py: remove sm_60, add sm_100 * Updates from cybind after removing all 11.x headers (affects "automatically generated" comments only). * Add new toolshed/reformat_cuda_enums_as_py.py (reads cuda.h, driver_types.h headers directly). * Use new toolshed/reformat_cuda_enums_as_py.py to regenerate driver_cu_result_explanations.py, runtime_cuda_error_explanations.py * Use `driver.cuDeviceGetUuid()` instead of `driver.cuDeviceGetUuid_v2()` with CTK 13. * Adjustments for locating nvvm directory in CTK 13 installations. * Fixes from windows testing (#89) * Add missing error handling (tests/test_nvjitlink.py) * Add missing `const` in cudaMemcpyBatchAsync call (cuda/bindings/runtime.pyx.in) * Add qa/13.0.0/01_linux.sh * Remove qa/13.0.0/01_linux.sh after it was moved to a new upstream qa branch. * Strictly correct casts for cudaMemcpyBatchAsync (generated by cython_gen). * Pragmatic minimal fix for cudaMemcpyBatchAsync casts (works with Linux and Windows). (generated with cython-gen) * print *prog pointers in nvrtcCreateProgram, nvrtcCompileProgram bindings * Remove stray `"nvrtc64_*_0.alt.dll"` entries in `SUPPORTED_WINDOWS_DLLS` * Revert "print *prog pointers in nvrtcCreateProgram, nvrtcCompileProgram bindings" This reverts commit 104abbd. * _find_lib_dir_using_cuda_home(): Windows CTK 13 → bin\x64 * getLocalRuntimeVersion(): Search for libcudart.so.13 * SUPPORTED_LINUX_SONAMES: Add CTK 13 soname values * Update path_finder/supported_libs.py from kitpicks 13.0.0/025 (#96) * Linux update from cuda_13.0.0_580.46_kitpicks025_linux.run: no-op b/o NVIDIA-dev/cuda-python-private#95 * Windows update from cuda_13.0.0_kitpicks025_windows.exe * This trivial change should have been included in PR #81, but was overlooked. Direct commit for simplicity. * cuda_core forward compatibility changes (private development branch) (#94) * CCCL_INCLUDE_PATH fixes in test_event.py, test_launcher.py * Add new file (accidentally missing in a prior commit). * Fix pre-commit errors in new tests/helpers.py * 12→13 compatibility fixes in cuda/core/experimental/_graph.py * CTK 12 compatibility (tests/test_cuda_utils.py) * Make the cuda/core/experimental/_graph.py changes backwards compatible. * Do not try to hide `13` in cuda_core/tests/test_cuda_utils.py * More elegant handling of `CCCL_INCLUDE_PATHS` in cuda_core/tests/helpers.py * Remove stray empty line (cuda_core/tests/conftest.py). * Fix logic error computing CCCL_INCLUDE_PATHS in cuda_core/tests/helpers.py * Fix `cuda_bindings` and `cuda_core` examples (#98) * Unmask globalToShmemAsyncCopy_test.py error: explicit pytest_skipif_cuda_include_not_found(), pytest_skipif_compute_capability_too_low() * Update cuda_bindings/examples/common/common.py for CTK 13 compatibility, to fix globalToShmemAsyncCopy_test.py * Update cuda_core/examples/thread_block_cluster.py for CTK 13 compatibility. * Update driver_cu_result_explanations.py, runtime_cuda_error_explanations.py (#100) * kitpicks/cuda-r13-0/13.0.0/033: CUDA_HOME=/usr/local/cuda python cython_gen.py --target-lib driver runtime nvrtc --out ../unreleased-13.0 (#107) * Update cuda_pathfinder supported_nvidia_libs.py from kitpicks 13.0.0/036 (NO CHANGES compared to 025) * Update driver_cu_result_explanations.py, runtime_cuda_error_explanations.py from kitpicks 13.0.0/036 (NO CHANGES compared to 025) * Update cuda_pathfinder supported_nvidia_libs.py EXPECTED_LIB_SYMBOLS for libnpp* The newly chosen symbols appear in all CTK 12.x releases and 13.0.0: https://gitlab-master.nvidia.com/rgrossekunst/rwgk_config_nvidia/-/blob/a1c2f29decd9b93fc7af9611bdc60565446b0cd3/bin/check_libnpp_symbols.sh * cython-gen changes due to PR #101 (#115) * cython-gen changes due to release_gil_revert_leos_commits_fix_cast_error (#116) * cython-gen changes due to cython-gen PR #118 (#120) * test_cufile.py: pytest.skip("NEEDS DEBUGGING (unreleased-13.0)") * cython-gen changes on top of `git merge world-main -X ours` product. * Add missing imports (related to #769). These got lost due to merging with `-X ours` * Revert obsolete aec7d10 Made obsolete by #778 * Remove cuda_bindings/site-packages entirely. * Change test_batch_io_large_operations to avoid a flood of output (`assert read_data == expected_data` failure). * Remove `(scope="module")` from `cufile_env_json` fixture: resolves test_batch_io_large_operations failure. * [unreleased-13.0] `test_cufile.py`: Remove fallback to `/etc/cufile.json` (#126) * test_cufile.py: NEVER USE /etc/cufile.json * Remove /etc/cufile.json code entirely. * update win driver to 580.88 * change backport branch to 12.9.x * update build ver to 13.0.0 * crt headers are now split from cudart (or nvcc?) * remove the outdated cufile skip condition (it was buggy anyway) * remove 11.8 CI and add 13.0 CI * update cuda-bindings optional dependencies * update release notes * update cuda-bindings docs * update cuda-python docs * libnvvm is also split out * ensure using sanitizer from the latest release of the same major ver * Remove -cu12 suffixes and add nvidia-nvvm in cuda_pathfinder/pyproject.toml. Make related changes in .github/workflows * fix backport branch's ci name * restore nvidia_wheels_cu12 * remove tests * always test 12.9.x with the latest driver * ensure fetch_ctk works with 12.x * Fix Linux libnvvm site-packages search for CTK 13 CTK 12: site-packages/nvidia/cuda_nvcc/nvvm/lib64/libnvvm.so CTK 13: site-packages/nvidia/cu13/lib/libnvvm.so.4 * update docs * add PTX ISA 9.0 to utils * sync 13.0.0 docs * Fix Windows site-packages search for CTK 13 * Also add "nvidia-nvvm~=13.0" in cuda_bindings/pyproject.toml * Add _work_around_known_bugs() in load_dl_linux.py To resolve this issue: #792 (comment) * driver_cu_result_explanations.py, runtime_cuda_error_explanations.py refresh (no-op) * SUPPORTED_LINUX_SONAMES refresh (no-op) * SUPPORTED_WINDOWS_DLLS refresh (no-op) * Update generated files: nvjitlink, nvvm (trivial changes, functional no-op) * update release dates --------- Co-authored-by: Vladislav Zhurba <vzhurba@nvidia.com> Co-authored-by: Vladislav Zhurba <53052066+vzhurba01@users.noreply.github.com> Co-authored-by: Ralf W. Grosse-Kunstleve <rgrossekunst@nvidia.com> Co-authored-by: Ralf W. Grosse-Kunstleve <rwgkio@gmail.com> Co-authored-by: Robert Maynard <rmaynard@nvidia.com>

Complete docstrings for public APIs #96

Description

Activity

leofang commented on Oct 15, 2024

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions