1.安装CUDA Tookit 7.5

• go to the webset to Install CUDA Toolkit 7.5
• log out, CTRL + ALT + F1 enter tty
• kill X service by command:

sudo server lightdm stop

• run

sudo init 3 
sudo sh cuda_7.5.18_linux.run

Note when installing cuda_7.5.18, do not install openGL

2.安装cudnn v4

• 注册下载 cudnn-7.0-linux-x64-v4.0-prod.tgz
• run

 tar -xzvf cudnn-7.0-linux-x64-v4.0-prod.tgz
1 cd cuda/include
2 sudo cp *.h /usr/local/cuda-7.5/include/
3 cd ../lib64
4 sudo cp lib* /usr/local/cuda-7.5/lib64/ 
5 cd /usr/local/cuda-7.5/lib64/ 
6 sudo chmod +r libcudnn.so.4.0.7
7 sudo ln -sf libcudnn.so.4.0.7 libcudnn.so.4
8 sudo ln -sf libcudnn.so.4 libcudnn.so
9 sudo ldconfig

3.安装tensorflow

• install anncoda2

bash Anaconda2-4.1.1-Linux-x86_64.sh
加上ignore-installed

• 安装参考
  解决安装Tensorflow时的setup-tool错误
  tensorflow document

4.设置环境变量

vim ~/.bashrc
export LD_LIBRARY_PATH="$LD_LIBRARY_PATH:/usr/local/cuda-7.5/lib64"
export CUDA_HOME=/usr/local/cuda-7.5
//
export PATH=/usr/local/cuda-8.0/bin:$PATH
export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PATH

5.测试

>>> import tensorflow as tf
>>> hello = tf.constant('Hello, TensorFlow!')
>>> sess = tf.Session()
>>> print(sess.run(hello))
Hello, TensorFlow!
>>> a = tf.constant(10)
>>> b = tf.constant(32)
>>> print(sess.run(a + b))
42

参考文章
1.ubuntu 14.04 搭建GPU深度学习 tensorflow 平台

2. How to use CUDA and the GPU Version of Tensorflow for Deep Learning
   3.HOW TO INSTALL TENSORFLOW WITH GPU SUPPORT ON A MACHINE WITH UBUNTU 14.04 LTS

tensorflow0.10.0升级到tensorflow0.11.0

(1). 卸载tensorflow0.10.0

sudo pip uninstall tensorflow

(2). 卸载cuda7.5,cuda8.0不需要卸载

cd /usr/local/cuda/bin
sudo ./uninstall_cuda_7.5.pl
# 手动清除cuda-7.5文件夹
cd /usr/local/sudo rm -rf cuda-7.5
# 删除cuda7.5samplecd ~
sudo rm -rf NVIDIA_CUDA-7.5_Samples

(3). 安装cuda8.0, 已有的不需要安装

chmod +x cuda_8.0.44_linux.run
sudo ./cuda_8.0.44_linux.run
#除了安装显卡驱动的地方选no,其他地方默认就好

(4). 安装cudnn5.1, 官网下载,解压

sudo cp cuda/include/cudnn.h /usr/local/cuda/include/ 
sudo cp cuda/lib64/libcudnn* /usr/local/cuda/lib64/ 
sudo chmod a+r /usr/local/cuda/include/cudnn.h 
sudo chmod a+r /usr/local/cuda/lib64/libcudnn*

(5). 修改.bashrc文件, 将以前的7.5改成8.0,执行source .bashrc

参考：tensorflow0.10.0升级到tensorflow.011.0