查看GPU信息

命令

lspci | grep -i nvidia
sudo lshw -numeric -C display

输出

06:00.0 VGA compatible controller: NVIDIA Corporation GK110B [GeForce GTX TITAN Z] (rev a1)
06:00.1 Audio device: NVIDIA Corporation GK110 HDMI Audio (rev a1)
07:00.0 3D controller: NVIDIA Corporation GK110B [GeForce GTX TITAN Z] (rev a1)

禁用 nouveau 驱动

操作

sudo vim /etc/modprobe.d/blacklist-nouveau.conf
# 输入以下内容
blacklist nouveau
options nouveau modset=0

执行

sudo update-initramfs -u

重启电脑

sudo lspci | grep nouveau

如果没有内容，则禁用成功。

安装cuda

到官网 下载对应版本的cuda安装包，我下载的是Ubuntu14.04,cuda8.0,runfile(local)版

执行以下命令

wget https://developer.nvidia.com/compute/cuda/8.0/Prod2/local_installers/cuda_8.0.61_375.26_linux-run
sudo /etc/init.d/lightdm stop   # 关闭X-server
sudo init 3
chmod 777 cuda_8.0.61_375.26_linux-run
sudo ./cuda_8.0.61_375.26_linux-run
sudo vim ~/.bashrc
# 添加以下三行内容
export CUDA_HOME=/usr/local/cuda-8.0
export PATH=/usr/local/cuda-8.0/bin:$PATH
export LD_LIBRARY_PATH=/usr/local/cuda-8.0/lib64:$LD_LIBRARY_PATH
source ~/.bashrc         #使配置的环境变量生效 
sudo vim /etc/ld.so.conf.d/cuda-8-0.conf
# 添加以下二行内容
/usr/local/cuda-8.0/lib64
/usr/local/cuda-8.0/extras/CUPTI/lib64
sudo ldconfig
nvidia-smi              #出现GPU相关信息表示驱动安装成功
nvcc --version          #查看cuda版本信息
sudo service lightdm start 

验证

cd /usr/local/cuda-8.0/samples/1_Utilities/deviceQuery  
sudo make  
sudo ./deviceQuery

保证编译没有error，执行deviceQuery输出信息的最后类似以下：

deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 8.0, CUDA Runtime Version = 8.0, NumDevs = 2, Device0 = GeForce GTX TITAN Z, Device1 = GeForce GTX TITAN Z
Result = PASS

安装cuDNN

为Deep Learning框架设计的一套GPU计算加速方案

下载地址,选择cuDNN v5.1 Library for Linux

执行以下命令

tar -zxvf cudnn-8.0-linux-x64-v5.1.tgz 
cd cuda
sudo cp lib64/* /usr/local/cuda-8.0/lib64/
sudo cp include/cudnn.h /usr/local/cuda-8.0/include/
sudo chmod a+r /usr/local/cuda/include/cudnn.h /usr/local/cuda-8.0/lib64/libcudnn*
sudo ln -sf libcudnn.so.5.1.10 libcudnn.so.5  
sudo ln -sf libcudnn.so.5 libcudnn.so  
sudo ldconfig  

安装tensorflow

Install pip

sudo apt-get update
sudo apt-get install python3
sudo apt-get install python3-pip python3-dev
sudo pip3 install --upgrade pip

Install TensorFlow

sudo pip3 install tensorflow         #CPU
sudo pip3 install tensorflow-gpu     #GPU

Validate your installation

• 代码

#!/usr/bin/python3.5

import tensorflow as tf
print('Loaded TF version', tf.__version__)
hello = tf.constant('Hello, TensorFlow!')
sess = tf.Session()
print(sess.run(hello))

• 运行结果

Loaded TF version 1.2.1
b'Hello, TensorFlow!'

bug

ImportError: libcudnn.so.5: cannot open shared object file: No such file or directory

cd /usr/local/cuda-8.0/lib64
sudo ln -s libcudnn.so.6.0.21 libcudnn.so.5

Not uninstalling pip at /usr/lib/python2.7/dist-packages, owned by OS

用root权限执行以下脚本

import pip
from subprocess import call

for dist in pip.get_installed_distributions():
    call("pip install --upgrade " + dist.project_name, shell=True)

如果还是不行， sudo apt-get remove python-pip3

卸载

sudo apt-get remove --purge nvidia-*        #卸载驱动
sudo pip3 uninstall tensorflow