ubuntu下tensorflow的GPU版本安装，包括Anaconda，cuda，cudnn

准备软件:

基于python2.7：

• Anaconda3-5.1.0-Linux-x86_64.sh
• tensorflow-1.12.0-cp27-none-linux_x86_64.whl
• tensorflow_gpu-1.12.0-cp27-none-linux_x86_64.whl
• cuda-repo-ubuntu1604-9-0-local_9.0.176-1_amd64.deb
• cudnn-9.0-linux-x64-v7.4.1.5.tgz

链接打包下载：

 

基于python3.6：

• Anaconda3-5.1.0-Linux-x86_64.sh
• tensorflow-1.12.0-cp36-cp36m-linux_x86_64.whl
• tensorflow_gpu-1.12.0-cp36-cp36m-linux_x86_64.whl
• cuda-repo-ubuntu1604-9-0-local_9.0.176-1_amd64.deb
• cudnn-9.0-linux-x64-v7.4.1.5.tgz

链接打包下载：

 

 

Anaconda3安装：

bash Anaconda3-5.1.0-Linux-x86_64.sh

conda create py36 python=3.6

mkdir -p ~/dev/env/py36TFGPU/

touch ~/dev/env/py36TFGPU/readME.txt > "This is the home of my tensorflow by GPU"

 

 tensorflow_gpu安装：

sudo dpkg -i tensorflow_gpu-1.12.0-cp36-cp36m-linux_x86_64.whl

sudo apt-get install -f

import的时候出现异常libcudnn.so.7: cannot open shared object file: No such file or directory 
 

export PATH=/usr/local/cuda-9.0/bin${PATH:+:${PATH}}
export LD_LIBRARY_PATH=/usr/local/cuda/lib64:/usr/local/cuda-9.0/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}

验证：

    import tensorflow as tf
    hello = tf.constant("Hello, TensorFlow!")
    sess = tf.Session()
    print(sess.run(hello))

输出结果应该是hello tensorflow，并且附带上GPU相关信息。

 

cuda

sudo dpkg -i cuda-repo-ubuntu1604-9-0-local_9.0.176-1_amd64.deb

sudo apt-get install -f

sudo apt-key add /var/cuda-repo-9-0-local/7fa2af80.pub

sudo apt-get update

sudo apt-get install cuda

 

cudnn

tar zxvf  cudnn-9.0-linux-x64-v7.4.1.5.tgz

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 /usr/local/cuda/lib64/libcudnn*

 

验证cuda的安装

nvcc -V