diff options
author | V3n3RiX <venerix@koprulu.sector> | 2023-05-08 23:39:55 +0100 |
---|---|---|
committer | V3n3RiX <venerix@koprulu.sector> | 2023-05-08 23:39:55 +0100 |
commit | 2c359e29620429f94a60fc1e832048e54ee41eb5 (patch) | |
tree | 3a24152899725be0796179577e84caf957a80ce3 /sci-libs/caffe2/caffe2-2.0.0-r4.ebuild | |
parent | d6a3ca89e16356791af128f18ca5d5aeaf1323e0 (diff) |
gentoo auto-resync : 08:05:2023 - 23:39:55
Diffstat (limited to 'sci-libs/caffe2/caffe2-2.0.0-r4.ebuild')
-rw-r--r-- | sci-libs/caffe2/caffe2-2.0.0-r4.ebuild | 195 |
1 files changed, 195 insertions, 0 deletions
diff --git a/sci-libs/caffe2/caffe2-2.0.0-r4.ebuild b/sci-libs/caffe2/caffe2-2.0.0-r4.ebuild new file mode 100644 index 000000000000..f6a3684a96a0 --- /dev/null +++ b/sci-libs/caffe2/caffe2-2.0.0-r4.ebuild @@ -0,0 +1,195 @@ +# Copyright 2022-2023 Gentoo Authors +# Distributed under the terms of the GNU General Public License v2 + +EAPI=8 + +PYTHON_COMPAT=( python3_{9..11} ) +inherit python-single-r1 cmake cuda flag-o-matic + +MYPN=pytorch +MYP=${MYPN}-${PV} + +DESCRIPTION="A deep learning framework" +HOMEPAGE="https://pytorch.org/" +SRC_URI="https://github.com/pytorch/${MYPN}/archive/refs/tags/v${PV}.tar.gz + -> ${MYP}.tar.gz" + +LICENSE="BSD" +SLOT="0" +KEYWORDS="~amd64" +IUSE="cuda distributed fbgemm ffmpeg gloo mpi nnpack +numpy opencl opencv openmp qnnpack tensorpipe xnnpack" +RESTRICT="test" +REQUIRED_USE=" + ${PYTHON_REQUIRED_USE} + ffmpeg? ( opencv ) + mpi? ( distributed ) + tensorpipe? ( distributed ) + gloo? ( distributed ) +" # ?? ( cuda rocm ) + +# CUDA 12 not supported yet: https://github.com/pytorch/pytorch/issues/91122 +RDEPEND=" + ${PYTHON_DEPS} + dev-cpp/gflags:= + >=dev-cpp/glog-0.5.0 + dev-libs/cpuinfo + dev-libs/libfmt + dev-libs/protobuf:= + dev-libs/pthreadpool + dev-libs/sleef + sci-libs/lapack + >=sci-libs/onnx-1.12.0 + sci-libs/foxi + cuda? ( + =dev-libs/cudnn-8* + dev-libs/cudnn-frontend:0/8 + <dev-util/nvidia-cuda-toolkit-12:=[profiler] + ) + fbgemm? ( dev-libs/FBGEMM ) + ffmpeg? ( media-video/ffmpeg:= ) + gloo? ( sci-libs/gloo[cuda?] ) + mpi? ( sys-cluster/openmpi ) + nnpack? ( sci-libs/NNPACK ) + numpy? ( $(python_gen_cond_dep ' + dev-python/numpy[${PYTHON_USEDEP}] + ') ) + opencl? ( virtual/opencl ) + opencv? ( media-libs/opencv:= ) + qnnpack? ( sci-libs/QNNPACK ) + tensorpipe? ( sci-libs/tensorpipe ) + xnnpack? ( >=sci-libs/XNNPACK-2022.12.22 ) +" +DEPEND=" + ${RDEPEND} + dev-cpp/eigen + cuda? ( dev-libs/cutlass ) + dev-libs/psimd + dev-libs/FP16 + dev-libs/FXdiv + dev-libs/pocketfft + dev-libs/flatbuffers + sci-libs/kineto + $(python_gen_cond_dep ' + dev-python/pyyaml[${PYTHON_USEDEP}] + dev-python/pybind11[${PYTHON_USEDEP}] + ') +" + +S="${WORKDIR}"/${MYP} + +PATCHES=( + "${FILESDIR}"/${P}-gentoo.patch + "${FILESDIR}"/${PN}-1.13.0-install-dirs.patch + "${FILESDIR}"/${PN}-1.12.0-glog-0.6.0.patch + "${FILESDIR}"/${PN}-1.13.1-tensorpipe.patch + "${FILESDIR}"/${P}-gcc13.patch + "${FILESDIR}"/${P}-cudnn_include_fix.patch +) + +src_prepare() { + filter-lto #bug 862672 + sed -i \ + -e "/third_party\/gloo/d" \ + cmake/Dependencies.cmake \ + || die + cmake_src_prepare + pushd torch/csrc/jit/serialization || die + flatc --cpp --gen-mutable --scoped-enums mobile_bytecode.fbs || die + popd +} + +src_configure() { + if use cuda && [[ -z ${TORCH_CUDA_ARCH_LIST} ]]; then + ewarn "WARNING: caffe2 is being built with its default CUDA compute capabilities: 3.5 and 7.0." + ewarn "These may not be optimal for your GPU." + ewarn "" + ewarn "To configure caffe2 with the CUDA compute capability that is optimal for your GPU," + ewarn "set TORCH_CUDA_ARCH_LIST in your make.conf, and re-emerge caffe2." + ewarn "For example, to use CUDA capability 7.5 & 3.5, add: TORCH_CUDA_ARCH_LIST=7.5,3.5" + ewarn "For a Maxwell model GPU, an example value would be: TORCH_CUDA_ARCH_LIST=Maxwell" + ewarn "" + ewarn "You can look up your GPU's CUDA compute capability at https://developer.nvidia.com/cuda-gpus" + ewarn "or by running /opt/cuda/extras/demo_suite/deviceQuery | grep 'CUDA Capability'" + fi + + local mycmakeargs=( + -DBUILD_CUSTOM_PROTOBUF=OFF + -DBUILD_SHARED_LIBS=ON + + -DUSE_CCACHE=OFF + -DUSE_CUDA=$(usex cuda) + -DUSE_CUDNN=$(usex cuda) + -DUSE_FAST_NVCC=$(usex cuda) + -DTORCH_CUDA_ARCH_LIST="${TORCH_CUDA_ARCH_LIST:-3.5 7.0}" + -DBUILD_NVFUSER=OFF + -DUSE_DISTRIBUTED=$(usex distributed) + -DUSE_MPI=$(usex mpi) + -DUSE_FAKELOWP=OFF + -DUSE_FBGEMM=$(usex fbgemm) + -DUSE_FFMPEG=$(usex ffmpeg) + -DUSE_GFLAGS=ON + -DUSE_GLOG=ON + -DUSE_GLOO=$(usex gloo) + -DUSE_KINETO=OFF # TODO + -DUSE_LEVELDB=OFF + -DUSE_MAGMA=OFF # TODO: In GURU as sci-libs/magma + -DUSE_MKLDNN=OFF + -DUSE_NCCL=OFF # TODO: NVIDIA Collective Communication Library + -DUSE_NNPACK=$(usex nnpack) + -DUSE_QNNPACK=$(usex qnnpack) + -DUSE_XNNPACK=$(usex xnnpack) + -DUSE_SYSTEM_XNNPACK=$(usex xnnpack) + -DUSE_TENSORPIPE=$(usex tensorpipe) + -DUSE_PYTORCH_QNNPACK=OFF + -DUSE_NUMPY=$(usex numpy) + -DUSE_OPENCL=$(usex opencl) + -DUSE_OPENCV=$(usex opencv) + -DUSE_OPENMP=$(usex openmp) + -DUSE_ROCM=OFF # TODO + -DUSE_SYSTEM_CPUINFO=ON + -DUSE_SYSTEM_PYBIND11=ON + -DUSE_UCC=OFF + -DUSE_VALGRIND=OFF + -DPYBIND11_PYTHON_VERSION="${EPYTHON#python}" + -DPYTHON_EXECUTABLE="${PYTHON}" + -DUSE_ITT=OFF + -DBLAS=Eigen # avoid the use of MKL, if found on the system + -DUSE_SYSTEM_EIGEN_INSTALL=ON + -DUSE_SYSTEM_PTHREADPOOL=ON + -DUSE_SYSTEM_FXDIV=ON + -DUSE_SYSTEM_FP16=ON + -DUSE_SYSTEM_GLOO=ON + -DUSE_SYSTEM_ONNX=ON + -DUSE_SYSTEM_SLEEF=ON + + -Wno-dev + -DTORCH_INSTALL_LIB_DIR="${EPREFIX}"/usr/$(get_libdir) + -DLIBSHM_INSTALL_LIB_SUBDIR="${EPREFIX}"/usr/$(get_libdir) + ) + + if use cuda; then + addpredict "/dev/nvidiactl" # bug 867706 + addpredict "/dev/char" + + mycmakeargs+=( + -DCMAKE_CUDA_FLAGS="$(cuda_gccdir -f | tr -d \")" + ) + fi + cmake_src_configure +} + +src_install() { + cmake_src_install + + insinto "/var/lib/${PN}" + doins "${BUILD_DIR}"/CMakeCache.txt + + rm -rf python + mkdir -p python/torch/include || die + mv "${ED}"/usr/lib/python*/site-packages/caffe2 python/ || die + mv "${ED}"/usr/include/torch python/torch/include || die + cp torch/version.py python/torch/ || die + rm -rf "${ED}"/var/tmp || die + python_domodule python/caffe2 + python_domodule python/torch +} |