# For the moment we only support the CPU and GPU backends of jaxlib. The TPU # backend will require some additional work. Those wheels are located here: # https://storage.googleapis.com/jax-releases/libtpu_releases.html. # For future reference, the easiest way to test the GPU backend is to run # NIX_PATH=.. nix-shell -p python3 python3Packages.jax "python3Packages.jaxlib-bin.override { cudaSupport = true; }" # export XLA_FLAGS=--xla_gpu_force_compilation_parallelism=1 # python -c "from jax.lib import xla_bridge; assert xla_bridge.get_backend().platform == 'gpu'" # python -c "from jax import random; random.PRNGKey(0)" # python -c "from jax import random; x = random.normal(random.PRNGKey(0), (100, 100)); x @ x" # There's no convenient way to test the GPU backend in the derivation since the # nix build environment blocks access to the GPU. See also: # * https://github.com/google/jax/issues/971#issuecomment-508216439 # * https://github.com/google/jax/issues/5723#issuecomment-913038780 { absl-py , addOpenGLRunpath , autoPatchelfHook , buildPythonPackage , config , cudnn ? cudaPackages.cudnn , fetchurl , flatbuffers , isPy39 , lib , python , scipy , stdenv # Options: , cudaSupport ? config.cudaSupport or false , cudaPackages ? {} }: let inherit (cudaPackages) cudatoolkit cudnn; in assert cudaSupport -> lib.versionAtLeast cudatoolkit.version "11.1"; assert cudaSupport -> lib.versionAtLeast cudnn.version "8.2"; let version = "0.4.4"; pythonVersion = python.pythonVersion; # Find new releases at https://storage.googleapis.com/jax-releases/jax_releases.html. # When upgrading, you can get these hashes from prefetch.sh. See # https://github.com/google/jax/issues/12879 as to why this specific URL is # the correct index. cpuSrcs = { "x86_64-linux" = fetchurl { url = "https://storage.googleapis.com/jax-releases/nocuda/jaxlib-${version}-cp310-cp310-manylinux2014_x86_64.whl"; hash = "sha256-4VT909AB+ti5HzQvsaZWNY6MS/GItlVEFH9qeZnUuKQ="; }; "aarch64-darwin" = fetchurl { url = "https://storage.googleapis.com/jax-releases/mac/jaxlib-${version}-cp310-cp310-macosx_11_0_arm64.whl"; hash = "sha256-wuOmoCeTldslSa0MommQeTe+RYKhUMam1ZXrgSov+8U="; }; "x86_64-darwin" = fetchurl { url = "https://storage.googleapis.com/jax-releases/mac/jaxlib-${version}-cp310-cp310-macosx_10_14_x86_64.whl"; hash = "sha256-arfiTw8yafJwjRwJhKby2O7y3+4ksh3PjaKW9JgJ1ok="; }; }; gpuSrc = fetchurl { url = "https://storage.googleapis.com/jax-releases/cuda11/jaxlib-${version}+cuda11.cudnn82-cp310-cp310-manylinux2014_x86_64.whl"; hash = "sha256-bJ62DdzuPSV311ZI2R/LJQ3fOkDibtz2+8wDKw31FLk="; }; in buildPythonPackage rec { pname = "jaxlib"; inherit version; format = "wheel"; # At the time of writing (2022-10-19), there are releases for <=3.10. # Supporting all of them is a pain, so we focus on 3.10, the current nixpkgs # python version. disabled = !(pythonVersion == "3.10"); # See https://discourse.nixos.org/t/ofborg-does-not-respect-meta-platforms/27019/6. src = if !cudaSupport then ( cpuSrcs."${stdenv.hostPlatform.system}" or (throw "jaxlib-bin is not supported on ${stdenv.hostPlatform.system}") ) else gpuSrc; # Prebuilt wheels are dynamically linked against things that nix can't find. # Run `autoPatchelfHook` to automagically fix them. nativeBuildInputs = lib.optionals cudaSupport [ autoPatchelfHook addOpenGLRunpath ]; # Dynamic link dependencies buildInputs = [ stdenv.cc.cc ]; # jaxlib contains shared libraries that open other shared libraries via dlopen # and these implicit dependencies are not recognized by ldd or # autoPatchelfHook. That means we need to sneak them into rpath. This step # must be done after autoPatchelfHook and the automatic stripping of # artifacts. autoPatchelfHook runs in postFixup and auto-stripping runs in the # patchPhase. Dependencies: # * libcudart.so.11.0 -> cudatoolkit_11.lib # * libcublas.so.11 -> cudatoolkit_11 # * libcuda.so.1 -> opengl driver in /run/opengl-driver/lib preInstallCheck = lib.optional cudaSupport '' shopt -s globstar addOpenGLRunpath $out/**/*.so for file in $out/**/*.so; do rpath=$(patchelf --print-rpath $file) # For some reason `makeLibraryPath` on `cudatoolkit_11` maps to # /lib which is different from /lib. patchelf --set-rpath "$rpath:${cudatoolkit}/lib:${lib.makeLibraryPath [ cudatoolkit.lib cudnn ]}" $file done ''; propagatedBuildInputs = [ absl-py flatbuffers scipy ]; # Note that cudatoolkit is snecessary since jaxlib looks for "ptxas" in $PATH. # See https://github.com/NixOS/nixpkgs/pull/164176#discussion_r828801621 for # more info. postInstall = lib.optional cudaSupport '' mkdir -p $out/bin ln -s ${cudatoolkit}/bin/ptxas $out/bin/ptxas ''; pythonImportsCheck = [ "jaxlib" ]; meta = with lib; { description = "XLA library for JAX"; homepage = "https://github.com/google/jax"; sourceProvenance = with sourceTypes; [ binaryNativeCode ]; license = licenses.asl20; maintainers = with maintainers; [ samuela ]; platforms = [ "aarch64-darwin" "x86_64-linux" "x86_64-darwin" ]; }; }