The HIP Environment and ROCm Kit - A lightweight open source build system for HIP and ROCm
pip install CppHeaderParser
sudo apt install \
repo git-lfs libnuma-dev ninja-build cmake g++ pkg-config libdrm-dev \
libelf-dev xxd libgl1-mesa-dev
We want ROCm sources checked out into the sources/ directory or if you check it out elsewhere create a symlink called ln -s /path/to/rocm sources
python ./build_tools/fetch_sources.py
This will use the repo tool with modified ROCm manifest. It will also apply patches to some of the repositories.
Alternatively, for a specific ROCm version e.g. 6.3.0
python ./build_tools/fetch_sources.py \
--manifest-url https://github.com/ROCm/ROCm.git \
--manifest-name tools/rocm-build/rocm-6.3.1.xml \
--manifest-branch roc-6.3.x
This will also apply the patches to the downloaded source files.
Checkout the latest development branch with
mkdir -p ~/github/rocm
cd ~/github/rocm
repo init -u https://github.com/ROCm/ROCm.git -m tools/rocm-build/rocm-6.3.1.xml -b roc-6.3.x
repo sync -j16
Checkout TheRock build tools
cd ~/github/
git clone https://github.com/nod-ai/TheRock
cd TheRock
ln -s </path/to/rocm> sources
Note that you must specify GPU targets or families to build for with either
-DTHEROCK_AMDGPU_FAMILIES= or -DTHEROCK_AMDGPU_TARGETS= and will get an
error if there is an issue. Supported families and targets are in the
therock_amdgpu_targets.cmake file. Not
all combinations are presently supported.
cmake -B build -GNinja . -DTHEROCK_AMDGPU_FAMILIES=gfx110X-dgpu
# Or if iterating and wishing to cache:
# cmake -Bbuild -GNinja -DCMAKE_C_COMPILER_LAUNCHER=ccache -DCMAKE_CXX_COMPILER_LAUNCHER=ccache .
cmake --build build
# Install per component.
cmake --install build --component amdgpu-runtime
cmake --install build --component amdgpu-runtime-dev
cmake --install build --component amdgpu-compiler
# Create archives.
(cd build && cpack -G TBZ2)
The following does not replace a robust test suite. However, it will tell you whether the toolchain you have just crafted is viable at all (as in can load and enumerate devices).
./build/dlopen-hip install/lib/libamdhip64.so
HIP enabled IREE can also be used to enumerate:
LD_LIBRARY_PATH=install/lib iree-run-module --dump_devices=hip
Our CI builds run in such a container, and it can be useful to run locally.
docker run --rm -it -v $PWD:$PWD --entrypoint /bin/bash ghcr.io/nod-ai/manylinux_ghr_x86_64:main
Packages needed:
yum install -y numactl-devel elfutils-libelf-devel vim-common git-lfs
pip install CppHeaderParser