【转载】 https://blog.csdn.net/xundh/article/details/79139292

环境：Ubuntu

准备环境apt-get 更换源cd /etc/aptsudo apt-get install vim sudo vim sources.list

• 1
• 2
• 3
• 4
  

deb http://mirrors.ustc.edu.cn/ubuntu/ xenial main restricted universe multiversedeb http://mirrors.ustc.edu.cn/ubuntu/ xenial-security main restricted universe multiversedeb http://mirrors.ustc.edu.cn/ubuntu/ xenial-updates main restricted universe multiversedeb http://mirrors.ustc.edu.cn/ubuntu/ xenial-proposed main restricted universe multiversedeb http://mirrors.ustc.edu.cn/ubuntu/ xenial-backports main restricted universe multiversedeb-src http://mirrors.ustc.edu.cn/ubuntu/ xenial main restricted universe multiversedeb-src http://mirrors.ustc.edu.cn/ubuntu/ xenial-security main restricted universe multiversedeb-src http://mirrors.ustc.edu.cn/ubuntu/ xenial-updates main restricted universe multiversedeb-src http://mirrors.ustc.edu.cn/ubuntu/ xenial-proposed main restricted universe multiversedeb-src http://mirrors.ustc.edu.cn/ubuntu/ xenial-backports main restricted universe multiverse

• 1
• 2
• 3
• 4
• 5
• 6
• 7
• 8
• 9
• 10
• 11
  

sudo apt-get updatesudo apt-get upgrade

• 1
• 2
  

修改pip源:mkdir ~/.pipvi ~/.pip/pip.conf

• 1
• 2
  

[global]trusted-host =  pypi.douban.comindex-url = http://pypi.douban.com/simple

• 1
• 2
• 3
  

sudo apt-get install python-pipsudo apt-get install python-numpy swig python-dev python-wheel

• 1
• 2
  

安装Nvidia驱动

在系统设置->软件更新->附加驱动->选择nvidia最新驱动->应用更改

验证：输入命令：

nvidia-smi

• 1
  

显示NVIDIA-SMI结果。

下载并安装cuda8sudo sh cuda_8.0.61_375.26_linux.run

• 1
  

Install NVIDIA Accelerated Graphics Driver for Linux-x86_64 375.26?
注意是否安装显卡驱动选n。
安装位置：/usr/local/cuda-8.0
sample：/home/admin1

修改环境配置

sudo vim ~/.bashrc

• 1
  

加上

export PATH=/usr/local/cuda-8.0/bin:$PATH  export LD_LIBRARY_PATH=/usr/local/cuda-8.0/lib64:$LD_LIBRARY_PATH  

• 1
• 2
• 3
  

source ~/.bashrc

• 1
  

对gcc降版，降到5.3以下gcc --versionsudo apt-get install gcc-4.8sudo update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-4.8 100

• 1
• 2
• 3
  

接下来编译测试cuda的sample

cd /usr/local/cuda/samples/1_Utilities/deviceQuery sudo make./deviceQuery

• 1
• 2
• 3
  

正常的话会输出显卡型号信息。

下载安装cuDNN

https://developer.nvidia.com/rdp/cudnn-download，下载v6.0，安装：

sudo dpkg -i cuda-repo-ubuntu1604-8-0-rc_8.0.27-1_amd64​.deb

• 1
  

安装 tensorflowsudo apt-get install libcupti-devpip install tensorflow-gpu

• 1
• 2
  

上面已经设置了从douban获取软件件，另外国内清华镜像地址：https://mirrors.tuna.tsinghua.edu.cn/help/tensorflow/

**执行：**sudo ldconfig /usr/local/cuda/lib64或者export CUDA_HOME=/usr/local/cudaexport LD_LIBRARY_PATH=$CUDA_HOME/lib64:$LD_LIBRARY_PATH

• 1
• 2
• 3
• 4
• 5
  

否则会提示libcublas.so.8.0:cannot open shared object file:No such file or directory

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

奈斯