ubuntu16.04 LTS+GeForce GTX 1070+TensorFlow(GPU版)安装

软件:ubuntu16.04 LTS+GeForce GTX 1070+TensorFlow

1、准备阶段

电脑配置: CPU: intel 酷睿 i7-7700k、显卡: GTX1070、内存：16G、硬盘：256G SSD + 2TB机械硬盘

软件： ubuntu16.04+cuda8+cudnn8+Anaconda2+tensorflow

2、安装NVIDIA GTX1070显卡驱动 （驱动也可以事先下载后再装）

如果ubuntu上曾经安装过驱动，则首先删除之前的驱动，然后再进行下面的操作。如果没有安装过，则直接进入下面的步骤。

1）sudo add-apt-repository ppa:graphics-drivers/ppa

2）sudo apt-get update

3）sudo apt-get install nvidia-384

4）sudo apt-get install mesa-common-dev

出现问题：“E: 无法定位软件包 mesa-commmon-dev”
解决办法：我当初在输入mesa后按Tab键自动补全就好了（很奇怪：为啥直接输入的名字却识别不了）。如果这样也不行的话，可能需要更新镜像源了。

出现问题2：E: 无法获得锁 /var/lib/dpkg/lock - open (11: 资源暂时不可用)

E: 无法锁定管理目录(/var/lib/dpkg/)，是否有其他进程正占用它？

解决方法：sudo rm /var/cache/apt/archives/lock

sudo rm /var/lib/dpkg/lock

5）sudo apt-get install freeglut3-dev

注：当然也可以先下载显卡驱动，然后再装驱动。

6）测试显卡驱动是否安装成功

命令：nvidia-smi

出现问题：NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running.

解决方法：重启系统就好了（原因位置），重新输入命令后出现下面的结果，则表面cuda安装成功。

3、安装cuda

下载链接：https://developer.nvidia.com/cuda-downloads

官方安装教程：http://docs.nvidia.com/cuda/cuda-quick-start-guide/index.html

1）sudo sh cuda_8.0.44_linux.run

注：下面这句一定要写'n',不然前面装的驱动不是白装了
Install NVIDIA Accelerated Graphics Driver for Linux-x86_64 384.81?

-------------------------------------------------------------
Do you accept the previously read EULA?
accept/decline/quit: accept

Install NVIDIA Accelerated Graphics Driver for Linux-x86_64 384.81?
(y)es/(n)o/(q)uit: n

Install the CUDA 8.0 Toolkit?
(y)es/(n)o/(q)uit: y

Enter Toolkit Location
 [ default is /usr/local/cuda-8.0 ]:

Do you want to install a symbolic link at /usr/local/cuda?
(y)es/(n)o/(q)uit: y

Install the CUDA 8.0 Samples?
(y)es/(n)o/(q)uit: y

Enter CUDA Samples Location
 [ default is /home/xw ]:

Installing the CUDA Toolkit in /usr/local/cuda-8.0 ...
Missing recommended library: libXi.so
Missing recommended library: libXmu.so

Installing the CUDA Samples in /home/xw ...
Copying samples to /home/xw/NVIDIA_CUDA-8.0_Samples now...
Finished copying samples.

===========
= Summary =
===========

Driver:   Not Selected
Toolkit:  Installed in /usr/local/cuda-8.0
Samples:  Installed in /home/xw, but missing recommended libraries

Please make sure that
 -   PATH includes /usr/local/cuda-8.0/bin
 -   LD_LIBRARY_PATH includes /usr/local/cuda-8.0/lib64, or, add /usr/local/cuda-8.0/lib64 to /etc/ld.so.conf and run ldconfig as root

To uninstall the CUDA Toolkit, run the uninstall script in /usr/local/cuda-8.0/bin

Please see CUDA_Installation_Guide_Linux.pdf in /usr/local/cuda-8.0/doc/pdf for detailed information on setting up CUDA.

***WARNING: Incomplete installation! This installation did not install the CUDA Driver. A driver of version at least 384.00 is required for CUDA 8.0 functionality to work.
To install the driver using this installer, run the following command, replacing <CudaInstaller> with the name of this run file:
    sudo <CudaInstaller>.run -silent -driver

Logfile is /tmp/cuda_install_4749.log

2）配置环境变量

2.1）命令：vim ~/.bashrc

添加以下内容：

export PATH=/usr/local/cuda-8.0/bin:$PATH
export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PATH
export LD_LIBRARY_PATH=/usr/lib/x86_64-linux-gnu:$LD_LIBRARY_PATH
export LD_LIBRARY_PATH=/lib/x86_64-linux-gnu:$LD_LIBRARY_PATH

2.2）命令：source ~/.bashrc

3）编译实例，验证cuda是否安装成功

$ cd NVIDIA_CUDA-8.0_Samples/

$ make

$ cd bin/x86_64/linux/release/

$ ./deviceQuery

4、安装cuDNN(cudnn7)

从Tar文件安装

$ tar -zxvf cudnn-8.0-linux-x64-v6.0.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*

5、安装Anaconda2

 $ bash Anaconda2-5.1.0-Linux-x86_64.sh
进入以下界面:

Do you approve the license terms? [yes|no]

>>>  yes

...

Do you wish the installer to prepend the Anaconda2 install location
to PATH in your /home/xw/.bashrc ? [yes|no]
[no] >>> yes

$ python

6、在Anconda2中安装tensorflow

$ conda create -n tensorflow

$ source activate tensorflow

激活tensorflow: source activate tensorflow

关闭tensorflow: source deactivate tensorflow

安装tensorflow:

(tensorflow) xw@xw-System-Product-Name:~$ pip install --ignore-installed --upgrade https://storage.googleapis.com/tensorflow/linux/gpu/tensorflow_gpu-1.6.0-cp36-cp36m-linux_x86_64.whl

上面会发生超时现象,不予推荐。

使用科大镜像：

(tensorflow) xw@xw-System-Product-Name:~$ pip install tensorflow-gpu -i https://pypi.mirrors.ustc.edu.cn/simple

测试tensorflow是否安装成功

$ python

>>> import tensorflow

出现下面的问题：

ImportError: libcublas.so.8.0: cannot open shared object file: No such file or directory

上面的问题说明tensorflow-gpu版本高了，我这个装了最新的tensorflow-gpu 1.6的，所以我改成安装1.4

(tensorflow) xw@xw-System-Product-Name:~$ pip install tensorflow-gpu==1.4 -i https://pypi.mirrors.ustc.edu.cn/simple

如果还是不行，就再降低tensorflow-gpu的版本。

再次测试后：

卸载命令：

卸载cuda：$ sudo /usr/local/cuda-9.0/bin/uninstall_cuda_9.0.pl

卸载驱动程序：

第1种方式： $ sudo /usr/bin/nvidia-uninstall

第2种方式：$ sudo apt-get remove --purge nvidia-*    （推荐）

卸载tensorflow：

第1种：sudo rm -rf tensorflow/

第2种：conda remove -n tensorflow --all

第3种：pip uninstall tensorflow-gpu （推荐）

参考文献：http://blog.csdn.net/swiftfake/article/details/79429539

官方安装文档：https://www.tensorflow.org/install/install_linux