Detailed description
Virtualenv installation [1]
Virtualenv is a tool to keep the dependencies required by different Python projects in separate places. The Virtualenv installation of TensorFlow will not override pre-existing version of the Python packages needed by TensorFlow.
step1: Install pip and Virtualenv
$ sudo apt-get install python-pip python-dev python-virtualenv
step2: Create a Virtualenv environment in the directory ~/tensorflow
$ virtualenv --system-site-packages ~/tensorflow
step3: Activate the environment
$ source ~/tensorflow/bin/activate
(tensorflow)$ # Your prompt should change
step4: Install TensorFlow just as you would for a regular Pip installation
Here we use Ubuntu/Linux 64-bit, GPU enabled, Python 2.7, and requires CUDA toolkit 7.5 and CuDNN v4
(tensorflow)$ export TF_BINARY_URL=https://storage.googleapis.com/tensorflow/linux/gpu/tensorflow-0.9.0-cp27-none-linux_x86_64.whl
step5: Install TensorFlow
(tensorflow)$ pip install --upgrade $TF_BINARY_URL
step6: Finisg installing
(tensorflow)$ deactivate
$ # Your prompt should change back
Install CUDA (GPUs on Linux)[2]
In order to build or run TensorFlow with GPU support, both NVIDIA's Cuda Toolkit (>= 7.0) and cuDNN (>= v2) need to be installed.
GPU card with NVidia Compute Capability >= 3.0 is needed, and our server is equipped with Quadro K2200.
step7: Download and install Cuda Toolkit
wget http://developer.download.nvidia.com/compute/cuda/7.5/Prod/local_installers/cuda_7.5.18_linux.run
sudo sh cuda_7.5.18_linux.run
step7: Download and install cuDNN
This step need to creat an account on Nvidia in order to download cuDNN file, the details already mentioned in Quick cheatsheet.
tar xvzf cudnn-7.0-linux-x64-v4.0-prod.tgz cuda/
sudo cp cudnn-7.5-linux-x64-v4/cudnn.h /usr/local/cuda/include
sudo cp cudnn-7.5-linux-x64-v4/libcudnn* /usr/local/cuda/lib64
sudo chmod a+r /usr/local/cuda/include/cudnn.h /usr/local/cuda/lib64/libcudnn*
To use TensorFlow later you will have to activate the Virtualenv environment again:
The basic usage can be find here [3]
$ source ~/tensorflow/bin/activate
(tensorflow)$ deactivate
(tensorflow)$ python -m tensorflow.models.image.mnist.convolutional
(tensorflow)$ deactivate