This page list useful resources about TensorFlow setup.
- See this post for my installation of GPU TensorFlow on Ubuntu Server 16.04 from source with Cuda 8
I browsed and referenced some of the posts – great appreciation for the authors for their sharing, and I list them below.
- Install GPU TensorFlow From Sources w/ Ubuntu 16.04 and Cuda 8.0 (September 8, 2016 by Justin)
- Install TensorFlow from source on Ubuntu 16.04 with GPU support (Oct 23, 2016 by Huidong Tian)
- Tensorflow v0.10 installed from scratch on Ubuntu 16.04, CUDA 8.0RC+Patch, cuDNN v5.1 with a 1080GTX (August 17, 2016 by Marceau)
- TensorFlow on a GTX 1080 by Mark Litwintschik (August 4, 2016)
- Install TensorFlow on Ubuntu 16.04 LTS (Oct 19, 2016 on Youtube)
- Setting up a Deep Learning Machine from Scratch by Sai Soundararaj (excellent installation notes)
- Dive Into TensorFlow, Part III: GTX 1080+Ubuntu16.04+CUDA8.0+cuDNN5.0+TensorFlow (Jul 18, 2016 on textminingonline.com)
- Getting Tensorflow to work with GPU NVidia GTX 1080 on Ubuntu 16.04 LTS (Jul 22, 2016 by abhay)
- Install TensorFlow with GPU support on a RedHat (supercluster) – (04 Sep 2016)
- NVidia GTX 1080 installation on Ubuntu 16.04 LTS (Jul 19, 2016 by )
- How can I install CuDNN on Ubuntu 16.04?
Step 0: Install cuda from the standard repositories. (See How can I install CUDA on Ubuntu 16.04?)
Step 1: Register an nvidia developer account and download cudnn here (about 80 MB)
Step 2: Check where your cuda installation is. For the installation from the repository it is /usr/lib/...
and /usr/include
. Otherwise, it will be /urs/local/cuda/
. You can check it with which nvcc
or ldconfig -p | grep cuda
Step 3: Copy the files:
$ cd folder/extracted/contents
$ sudo cp -P include/cudnn.h /usr/include
$ sudo cp -P lib64/libcudnn* /usr/lib/x86_64-linux-gnu/
$ sudo chmod a+r /usr/lib/x86_64-linux-gnu/libcudnn*
Adding -P
retains the symbolic links, i.e. sudo cp -P lib64/libcudnn* /usr/lib/x86_64-linux-gnu/
, and avoids the message: /sbin/ldconfig.real: /usr/lib/x86_64-linux-gnu/libcudnn.so.5 is not a symbolic link
– Max Gordon Jun 26 ’16 at 8:56
For Tensorflow to find everything, I had to copy include/cudnn.h
and the libraries in lib64/
to /usr/local/cuda-8.0/include
and /usr/local/cuda-8.0/lib64
(using CUDA 8.0, Ubuntu 14.04, Tensorflow 0.12.0rc0) – maybe this is helpful for somebody. – David Stutz Dec 9 ’16 at 12:16
- On the cloud (source: from this page)
If you don’t want to install TensorFlow, you can use TensorFlow over the web.
You can use Tensorflow over the web at https://cloud.sagemath.com/ Simply click on the link, create an account (or log in with your GitHub), and create a TensorFlow project.
You can also use Jupyter notebook to write TensorFlow programs.
======TensorFlow setup on RedHat supercomputer clusters
Install TensorFlow with GPU support on a RedHat (supercluster) (pdf)
Install TensorFlow on CentOS7 (pdf)
Installing TensorFlow on CentOS (pdf)