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
114 lines
3.1 KiB
Bash
Executable File
114 lines
3.1 KiB
Bash
Executable File
#!/usr/bin/env bash
|
|
|
|
set -euo pipefail
|
|
|
|
readonly matx_repo=https://github.com/NVIDIA/MatX.git
|
|
readonly matx_branch=main
|
|
|
|
# Ensure the script is being executed in the root cccl directory:
|
|
cd "$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )/../..";
|
|
|
|
log_vars() {
|
|
for var in "$@"; do
|
|
echo "${var}=${!var}"
|
|
done
|
|
}
|
|
|
|
# Take two version strings and return the greater of the two:
|
|
version_max() {
|
|
local v1="${1}"
|
|
local v2="${2}"
|
|
if ci/util/version_compare.sh "$v1" ge "$v2"; then
|
|
echo "$v1"
|
|
else
|
|
echo "$v2"
|
|
fi
|
|
}
|
|
|
|
# Get the current CCCL info:
|
|
readonly cccl_repo="${PWD}"
|
|
|
|
# Define CCCL_TAG to override the default CCCL SHA. Otherwise the current HEAD of the local checkout is used.
|
|
echo "CCCL_TAG (override): ${CCCL_TAG-}";
|
|
if test -n "${CCCL_TAG-}"; then
|
|
# If CCCL_TAG is defined, fetch it to the local checkout
|
|
git -C "${cccl_repo}" fetch origin "${CCCL_TAG}";
|
|
cccl_sha="$(git -C "${cccl_repo}" rev-parse FETCH_HEAD)";
|
|
else
|
|
cccl_sha="$(git -C "${cccl_repo}" rev-parse HEAD)";
|
|
fi
|
|
|
|
cccl_repo_version="$(git -C "${cccl_repo}" describe "${cccl_sha}"| grep -Eo '[0-9]+\.[0-9]+\.[0-9]+')"
|
|
readonly cccl_repo_version
|
|
|
|
# Define CCCL_VERSION to override the version used by rapids-cmake to patch CCCL.
|
|
echo "CCCL_VERSION (override): ${CCCL_VERSION-}";
|
|
if test -n "${CCCL_VERSION-}"; then
|
|
readonly cccl_rapids_cmake_version="${CCCL_VERSION}"
|
|
else
|
|
cccl_rapids_cmake_version="${cccl_repo_version}"
|
|
# shellcheck disable=SC2034
|
|
readonly cccl_rapids_cmake_version
|
|
fi
|
|
|
|
# If the current version is less than 2.8.0, use 2.8.0 for the rapids-cmake version.
|
|
# This is to allow rapids-cmake to correctly patch the CCCL install rules on current `main`.
|
|
cccl_version=$(version_max "${cccl_repo_version}" "2.8.0")
|
|
readonly cccl_version
|
|
|
|
readonly workdir="${cccl_repo}/build/${CCCL_BUILD_INFIX:-}/matx"
|
|
readonly version_file="${workdir}/MatX/cmake/versions.json"
|
|
readonly version_override_file="${workdir}/versions-override.json"
|
|
|
|
log_vars \
|
|
matx_repo matx_branch \
|
|
cccl_repo cccl_sha cccl_repo_version cccl_rapids_cmake_version \
|
|
workdir \
|
|
version_file version_override_file
|
|
|
|
mkdir -p "${workdir}"
|
|
cd "${workdir}"
|
|
|
|
# Python deps:
|
|
pip install numpy
|
|
|
|
# Clone MatX
|
|
rm -rf MatX
|
|
git clone "${matx_repo}" -b "${matx_branch}"
|
|
|
|
cd MatX
|
|
echo "MatX HEAD:"
|
|
git log -1 --format=short
|
|
cd ..
|
|
|
|
# Write out version override file
|
|
jq -r ".packages.CCCL *=
|
|
{
|
|
\"git_url\": \"${cccl_repo}\",
|
|
\"git_tag\": \"${cccl_sha}\",
|
|
\"version\": \"${cccl_version}\",
|
|
\"always_download\": true
|
|
}" \
|
|
"${version_file}" > "${version_override_file}"
|
|
|
|
echo "Overriding MatX versions.json file:"
|
|
cat "$version_override_file"
|
|
|
|
# Configure and build
|
|
rm -rf build
|
|
|
|
SCCACHE_NO_DIST_COMPILE=1 cmake \
|
|
-B build -S MatX -G Ninja \
|
|
"-DCMAKE_CUDA_ARCHITECTURES=75;120" \
|
|
"-DRAPIDS_CMAKE_CPM_OVERRIDE_VERSION_FILE=${version_override_file}" \
|
|
-DMATX_BUILD_TESTS=ON \
|
|
-DMATX_BUILD_EXAMPLES=ON \
|
|
-DMATX_BUILD_BENCHMARKS=ON \
|
|
-DMATX_EN_CUTENSOR=ON
|
|
|
|
# Disabled because `cmake --build -j ""` is invalid, but so is
|
|
# `cmake --build -j8`. CMake expects a space between `-j` and
|
|
# the numeric argument, or no argument at all.
|
|
# shellcheck disable=SC2086
|
|
time cmake --build build -j ${PARALLEL_LEVEL:-}
|