homebrew-core/Formula/mkl-dnn.rb

37 lines
1.1 KiB
Ruby

class MklDnn < Formula
desc "Intel Math Kernel Library for Deep Neural Networks"
homepage "https://01.org/mkl-dnn"
url "https://github.com/intel/mkl-dnn/archive/v1.2.1.tar.gz"
sha256 "c69544783c453ab3fbf14c7a5b9a512561267690c9fc3e7fc3470f04756e0ab3"
bottle do
cellar :any
sha256 "8365520f2ee8245beacd14c63e8c8cf66d90f502a1ec2fe15771810de1e0fe36" => :catalina
sha256 "6e46c0a7840a076ff108f2d6befc99fa4cb6fbe2b9b44b1592f68966fedaa85a" => :mojave
sha256 "2ed24a849cfeb4b0a2c5d16376ac1bce5847eb4fd61e32bc18faca487ea32f23" => :high_sierra
end
depends_on "cmake" => :build
depends_on "doxygen" => :build
def install
system "cmake", ".", *std_cmake_args
system "make"
system "make", "doc"
system "make", "install"
end
test do
(testpath/"test.c").write <<~EOS
#include <mkldnn.h>
int main() {
mkldnn_engine_t engine;
mkldnn_status_t status = mkldnn_engine_create(&engine, mkldnn_cpu, 0);
return !(status == mkldnn_success);
}
EOS
system ENV.cc, "-L#{lib}", "-lmkldnn", "test.c", "-o", "test"
system "./test"
end
end