Is mxnet-mkl(Linux) on pip built with GNU OpenMP or Intel OpenMP

1

Hi MxNet Users,

I’m wondering what feature flag was enabled with mxnet-mkl pre-built binary on Pip. Specifically is it built using GNU OpenMP OR Intel OpenMP?

Thanks,
Peter

2

Hi @chenliu0831

I believe it is built using Intel OpenMP. The pip package comes with libiomp5.so.

libiomp5.so  libmklml_intel.so  libmxnet.so

ldd libmxnet.so

	linux-vdso.so.1 =>  (0x00007ffef7398000)
	libcudart.so.9.0 => /usr/local/cuda/lib64/libcudart.so.9.0 (0x00007f7a05bea000)
	libcublas.so.9.0 => /usr/local/cuda/lib64/libcublas.so.9.0 (0x00007f7a01f6d000)
	libcurand.so.9.0 => /usr/local/cuda/lib64/libcurand.so.9.0 (0x00007f79fe009000)
	libcusolver.so.9.0 => /usr/local/cuda/lib64/libcusolver.so.9.0 (0x00007f79f940e000)
	libmklml_intel.so => /home/ubuntu/anaconda3/lib/python3.6/site-packages/mxnet/./libmklml_intel.so (0x00007f79f17ab000)
	libiomp5.so => /home/ubuntu/anaconda3/lib/python3.6/site-packages/mxnet/./libiomp5.so (0x00007f79f13cf000)
	librt.so.1 => /lib/x86_64-linux-gnu/librt.so.1 (0x00007f79f11c7000)
	libmkldnn.so.0 => /home/ubuntu/anaconda3/lib/python3.6/site-packages/mxnet/./libmkldnn.so.0 (0x00007f79f07a5000)
	libdl.so.2 => /lib/x86_64-linux-gnu/libdl.so.2 (0x00007f79f05a1000)
	libgfortran.so.3 => /home/ubuntu/anaconda3/lib/python3.6/site-packages/mxnet/./libgfortran.so.3 (0x00007f79f027a000)
	libcufft.so.9.0 => /usr/local/cuda/lib64/libcufft.so.9.0 (0x00007f79e81d9000)
	libcuda.so.1 => /usr/lib/x86_64-linux-gnu/libcuda.so.1 (0x00007f79e70d7000)
	libnvrtc.so.9.0 => /usr/local/cuda/lib64/libnvrtc.so.9.0 (0x00007f79e588c000)
	libstdc++.so.6 => /usr/lib/x86_64-linux-gnu/libstdc++.so.6 (0x00007f79e550a000)
	libm.so.6 => /lib/x86_64-linux-gnu/libm.so.6 (0x00007f79e5201000)
	libgcc_s.so.1 => /lib/x86_64-linux-gnu/libgcc_s.so.1 (0x00007f79e4feb000)
	libpthread.so.0 => /lib/x86_64-linux-gnu/libpthread.so.0 (0x00007f79e4dce000)
	libc.so.6 => /lib/x86_64-linux-gnu/libc.so.6 (0x00007f79e4a04000)
	/lib64/ld-linux-x86-64.so.2 (0x00007f7a32074000)
	libquadmath.so.0 => /home/ubuntu/anaconda3/lib/python3.6/site-packages/mxnet/./libquadmath.so.0 (0x00007f79e47c4000)
	libnvidia-fatbinaryloader.so.410.104 => /usr/lib/x86_64-linux-gnu/libnvidia-fatbinaryloader.so.410.104 (0x00007f79e4577000)