Getting up and running with TensorFlow

Installation

• Get CUDA 7.5 from https://developer.nvidia.com/cuda-downloads

sudo apt install nvidia-cuda-toolkit

• Get cuDNN v4 from https://developer.nvidia.com/cudnn -- this will require signing up to the developer program. Use locate cuda.h and locate libcuda.so to find the right place to copy these. The following works on Ubuntu 16.04.

tar xvzf cudnn-7.*
sudo cp cuda/include/cudnn.h /usr/include/cuda
sudo cp cuda/lib64/libcudnn* /usr/lib/x86_64-linux-gnu/
sudo chmod a+r /usr/include/cudnn.h /usr/lib/x86_64-linux-gnu/libcudnn*

• Install using conda

conda create -n tensorflow-gpu python=3.5 anaconda
source activate tensorflow-gpu
pip install jupyter pandas tables matplotlib
export TF_BINARY_URL=https://storage.googleapis.com/tensorflow/linux/gpu/tensorflow-0.10.0rc0-cp35-cp35m-linux_x86_64.whl
pip install --ignore-installed --upgrade ${TF_BINARY_URL}

• Test that the install has been successful

python -c "import tensorflow"

• Install bazel: see here

echo "deb [arch=amd64] http://storage.googleapis.com/bazel-apt stable jdk1.8" | sudo tee /etc/apt/sources.list.d/bazel.list
curl https://storage.googleapis.com/bazel-apt/doc/apt-key.pub.gpg | sudo apt-key add -
sudo apt update
sudo apt install bazel swig