Since Google Tensorflow claimed that tensorflow-gpu no longer supports 
MAC OSX since 1.2.0 due to the OpenMP issue of clang of Apple,
I built this unoffcial tensorflow-gpu for MAC OSX so that
Hackintosh users or Mac users with eGPU can run tensorflow-gpu with CUDA.

1. Must use python3 instead of python2

2. Must be installed on MAC OSX 10.6+

3. Must install Nvidia GPU drivers

4. Must install Nvidia CUDA toolkit 9.1 (if not, you need to re-compile by yourself)

5. Must install Nvidia CUDA cudnn 7.0 (if not, you need to re-compile by yourself)

6. Must set up cuda environment (make sure 'nvcc -V' shows the cuda version '9.1')

7.CUDA compute capability is in [3.0,5.0] (if not, you need to re-compile to get a better performance)

nvcc -V

brew install coreutils

brew install llvm

brew install cliutils/apple/libomp

git clone https://github.com/zylo117/tensorflow-gpu-macosx

./configure
  #Please specify the location of python.: Accept the default option
    #Please input the desired Python library path to use.:  Accept the default option
    #Do you wish to build TensorFlow with Google Cloud Platform support? [Y/n]: n
    #Do you wish to build TensorFlow with Hadoop File System support? [Y/n]: n
    #Do you wish to build TensorFlow with Amazon S3 File System support? [Y/n]: n
    #Do you wish to build TensorFlow with XLA JIT support? [y/N]: n
    #Do you wish to build TensorFlow with GDR support? [y/N]: n
    #Do you wish to build TensorFlow with VERBS support? [y/N]: n
    #Do you wish to build TensorFlow with OpenCL SYCL support? [y/N]: n
    #Do you wish to build TensorFlow with CUDA support? [y/N]: y
    #Please specify the CUDA SDK version you want to use, e.g. 7.0.: 9.1
    #Please specify the location where CUDA 9.1 toolkit is installed.: Accept the default option
    #Please specify the cuDNN version you want to use.: 7
    #Please specify the location where cuDNN 7 library is installed.: Accept the default option
    ##Please specify a list of comma-separated Cuda compute capabilities you want to build with.
    ##You can find the compute capability of your device at: https://developer.nvidia.com/cuda-gpus. (GTX10X0: 6.1, GTX9X0: 5.2)
    #Please note that each additional compute capability significantly increases your build time and binary size.: 6.1,5.2,5.0,3.0
    #Do you want to use clang as CUDA compiler? [y/N]: n
    #Please specify which gcc should be used by nvcc as the host compiler.: Accept the default option
    #Do you wish to build TensorFlow with MPI support? [y/N]: n
    #Please specify optimization flags to use during compilation when bazel option "--config=opt" is specified:  Accept the default option
    #Would you like to interactively configure ./WORKSPACE for Android builds? [y/N]: n

export CUDA_HOME=/usr/local/cuda
export DYLD_LIBRARY_PATH=/usr/local/cuda/lib:/usr/local/cuda/extras/CUPTI/lib
export LD_LIBRARY_PATH=$DYLD_LIBRARY_PATH
export PATH=$DYLD_LIBRARY_PATH:$PATH

# bazel clean --expunge # Use this if you failed to compile before.

bazel build --config=cuda --config=opt --cxxopt="-D_GLIBCXX_USE_CXX11_ABI=0" --action_env PATH --action_env LD_LIBRARY_PATH --action_env DYLD_LIBRARY_PATH //tensorflow/tools/pip_package:build_pip_package

gcc -c -fPIC ./nccl_patched/nccl_ops.cc -o ./nccl_patched/_nccl_ops.o

gcc ./nccl_patched/_nccl_ops.o -shared -o ./nccl_patched/_nccl_ops.so

mv ./bazel-bin/tensorflow/contrib/nccl/python/ops/_nccl_ops.so ./bazel-bin/tensorflow/contrib/nccl/python/ops/_nccl_ops.so.bk

cp ./nccl_patched/_nccl_ops.so ./bazel-bin/tensorflow/contrib/nccl/python/ops/

./bazel-bin/tensorflow/tools/pip_package/build_pip_package/ ./tmp/tensorflow_pkg

This version will not supports multi-cpu/machine training.

It's not like you are gonna need it anyway.

**Either compile from src through pypi (not recommended)**
    pip3 install tensorflow-gpu-macosx
**or install from relese (the easiest way)**
    pip3 install *.whl