tensorflow-gpu.md

Step 1: Install Python 3.6

I use anaconda:

wget https://repo.anaconda.com/archive/Anaconda3-5.3.0-MacOSX-x86_64.sh
bash ./Anaconda3-5.3.0-MacOSX-x86_64.sh
conda create -n idp intelpython3_full python=3
source activate idp

Intel Python is still using Python 3.6.5 as the time of writing.

Step 2: Install Java8 and Bazel

brew cask install homebrew/cask-versions/java8
brew tap bazelbuild/tap
brew tap-pin bazelbuild/tap
brew install bazel

Step 3: Configure Tensorflow

git clone https://github.com/tensorflow/tensorflow.git
cd tensorflow
./configure

You can use default answers, with the exception of following configs:

Do you wish to build TensorFlow with CUDA support? [y/N]: y
CUDA support will be enabled for TensorFlow.

Please specify the CUDA SDK version you want to use. [Leave empty to default to CUDA 9.0]: 10.0

You should have downloaded CUDA Toolkit and cuDNN. After downloading cuDNN, you simply uncompress it to /user/local/cuda.

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.
Please note that each additional compute capability significantly increases your build time and binary size. [Default is: 3.5,7.0]: 6.1

Here you must specify the capability of your GPU. Follow the link in the prompt.