2a7ca101d7
Sparse-checkout from NVIDIA/cccl main branch to complete cccl_upstream: Added: - python/cuda_cccl/ (226 files) — Python bindings for device-level algorithms Critical for muh toolchain: cuda.compute.reduce_into, scan, radix_sort, etc. Includes 204 .py files with full test coverage for all 27 algorithms - ci/ (163 files) — Build/test infrastructure build_cub.sh, test_cub.sh, build_and_test_targets.sh, matrix.yaml Directly maps to our [INFRA-CI] and [INFRA-BUILD] items - .agent/skills/ (7 files) — NVIDIA's own agent skills for CCCL cccl-style/SKILL.md, cccl-test/SKILL.md, sass-diff/SKILL.md - docs/ (491 files) — Official CCCL documentation CI references, CMake guides, Python compute docs, libcudacxx PTX docs - test/ (12 files) — Top-level integration tests (cuda_smoke, stdpar) - Root configs: .clang-format, .clang-tidy, CONTRIBUTING.md, pyproject.toml - CLAUDE.md symlink → AGENTS.md (NVIDIA's standard) cccl_upstream now mirrors full NVIDIA/cccl structure: Before: 42M (cub + thrust + libcudacxx + cudax + c + examples + benchmarks) After: 53M (+python +ci +docs +.agent +test +configs) This completes the CCCL base needed for: - [muh-bench] items: ci/util/build_and_test_targets.sh for targeted builds - [CCCL-verify] items: python/cuda_cccl/tests/ as reference implementations - [CCCL-test] items: ci/test_cub.sh, ci/test_thrust.sh - Agent workflow: .agent/skills/ for consistent style and test patterns
51 lines
2.1 KiB
Bash
Executable File
51 lines
2.1 KiB
Bash
Executable File
#!/usr/bin/env bash
|
|
|
|
set -euo pipefail
|
|
|
|
ci_dir="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
|
|
source "$ci_dir/pyenv_helper.sh"
|
|
|
|
# Parse common arguments
|
|
source "$ci_dir/util/python/common_arg_parser.sh"
|
|
parse_python_args "$@"
|
|
# Pin cuda-toolkit to the container's CTK minor and set cuda_version /
|
|
# cuda_major_version (-ctk-mode latest opts out). See pyenv_helper.sh.
|
|
pin_cuda_toolkit "${ctk_mode}"
|
|
|
|
# Setup Python environment
|
|
setup_python_env "${py_version}"
|
|
|
|
# Fetch or build the cuda_cccl wheel:
|
|
if [[ -n "${GITHUB_ACTIONS:-}" ]]; then
|
|
wheel_artifact_name=$("$ci_dir/util/workflow/get_wheel_artifact_name.sh")
|
|
"$ci_dir/util/artifacts/download.sh" "${wheel_artifact_name}" /home/coder/cccl/
|
|
else
|
|
"$ci_dir/build_cuda_cccl_python.sh" -py-version "${py_version}"
|
|
fi
|
|
|
|
# Install cuda_cccl. The extra flavor is "cu" (pip-installed toolkit) or "sysctk"
|
|
# (system-provided toolkit) depending on the -ctk-mode arg.
|
|
CUDA_CCCL_WHEEL_PATH="$(ls /home/coder/cccl/wheelhouse/cuda_cccl-*.whl)"
|
|
ctk_flavor="$(ctk_extra_flavor "${ctk_mode}")"
|
|
python -m pip install "${CUDA_CCCL_WHEEL_PATH}[test-${ctk_flavor}${cuda_major_version}]"
|
|
|
|
# Run tests for compute module.
|
|
# On the v2 (HostJIT) backend, abort on first failure — the suite is still
|
|
# stabilizing and a single early failure is enough signal to investigate
|
|
# without scrolling through hundreds of subsequent passes.
|
|
pytest_extra=()
|
|
if [[ "${CCCL_PYTHON_USE_V2:-}" =~ ^(1|true|TRUE|on|ON)$ ]]; then
|
|
pytest_extra+=(-x)
|
|
fi
|
|
|
|
cd "/home/coder/cccl/python/cuda_cccl/tests/"
|
|
if [[ "${CCCL_PYTHON_USE_V2:-}" =~ ^(1|true|TRUE|on|ON)$ ]]; then
|
|
# The test isolates itself in a fresh subprocess (LLVM initialization is
|
|
# process-wide and only cold once), but it carries the free_threading marker,
|
|
# so it must be selected by node-id here or the sweeps below never run it.
|
|
python -m pytest "${pytest_extra[@]}" -n 0 -v \
|
|
compute/test_free_threading_stress.py::test_v2_concurrent_cold_llvm_initialization
|
|
fi
|
|
python -m pytest "${pytest_extra[@]}" -n 6 -v compute/ -m "not large and not free_threading"
|
|
python -m pytest "${pytest_extra[@]}" -n 0 -v compute/ -m "large and not free_threading"
|