系统版本兼容要求
centos7.2 cuda9.0 cudnn7.4
centos7.5 cuda9.2 cudnn7.4
安装gcc
yum -y install gcc gcc-c++ kernel-devel
package manage-overview
https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#package-manager-overview
1、安装gpu显卡驱动
查看nvidia gpu信息
# nvidia-smi
2、安装nvidia检测
2.1添加ElRepo源
# rpm --import https://www.elrepo.org/RPM-GPG-KEY-elrepo.org
# rpm --import https://www.elrepo.org/RPM-GPG-KEY-elrepo.org
# rpm -Uvh https://www.elrepo.org/elrepo-release-7.0-3.el7.elrepo.noarch.rpm
2.2、安装显卡驱动检查
yum install nvidia-detect
2.3 运行
# nvidia-detect -v
Probing for supported NVIDIA devices...
[10de:15f8] NVIDIA Corporation Device 15f8
This device requires the current 410.78 NVIDIA driver kmod-nvidia
[10de:15f8] NVIDIA Corporation Device 15f8
This device requires the current 410.78 NVIDIA driver kmod-nvidia
[102b:0538] Matrox Electronics Systems Ltd. Device 0538
2.4、编辑grub文件
vim /etc/default/grub
在"GRUB_CMDLINE_LINUX"中添加
rd.driver.blacklist=nouveau nouveau.modeset=0
改完后的文件如下：
GRUB_TIMEOUT=5
GRUB_DISTRIBUTOR="$(sed 's, release .*$,,g' /etc/system-release)"
GRUB_DEFAULT=saved
GRUB_DISABLE_SUBMENU=true
GRUB_TERMINAL_OUTPUT="console"
GRUB_CMDLINE_LINUX="crashkernel=auto rd.lvm.lv=centos/root rd.lvm.lv=centos/swap rd.driver.blacklist=nouveau nouveau.modeset=0 rhgb quiet"
GRUB_DISABLE_RECOVERY="true"
随后生成配置
grub2-mkconfig -o /boot/grub2/grub.cfg
2.5、创建blacklist
vim /etc/modprobe.d/blacklist.conf
添加
blacklist nouveau
2.6、更新配置
mv /boot/initramfs-$(uname -r).img /boot/initramfs-$(uname -r)-nouveau.img
dracut /boot/initramfs-$(uname -r).img $(uname -r)
2.7、重启
reboot
2.8、确认禁用了nouveau
lsmod | grep nouveau
若无输出则禁用成功
3、安装cuda
cuda下载地址：
https://developer.nvidia.com/cuda-toolkit
# sh cuda_9.0.176_384.81_linux.run
如果出现you appear to be running an x server please exit x before installing
执行init 3 进入命令行模式，杀掉x server，然后再执行安装命令
===========
= Summary =
===========
Driver: Installed
Toolkit: Installed in /usr/local/cuda-9.0
Samples: Installed in /root, but missing recommended libraries
Please make sure that
- PATH includes /usr/local/cuda-9.0/bin
- LD_LIBRARY_PATH includes /usr/local/cuda-9.0/lib64, or, add /usr/local/cuda-9.0/lib64 to /etc/ld.so.conf and run ldconfig as root
To uninstall the CUDA Toolkit, run the uninstall script in /usr/local/cuda-9.0/bin
To uninstall the NVIDIA Driver, run nvidia-uninstall
Please see CUDA_Installation_Guide_Linux.pdf in /usr/local/cuda-9.0/doc/pdf for detailed information on setting up CUDA.
Logfile is /tmp/cuda_install_7874.log
验证CUDA 9.0 是否安装成功
终端输入：
nvcc -V
可以看到cuda的版本信息
接着尝试运行一下cuda中自带的例子：
cd /usr/local/cuda-9.0/samples/1_Utilities/deviceQuery
make
./deviceQuery
可以看到输出成功
deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 10.0, CUDA Runtime Version = 10.0, NumDevs = 2
Result = PASS
卸载
To uninstall the CUDA Toolkit, run the uninstall script in /usr/local/cuda-9.0/bin
To uninstall the NVIDIA Driver, run nvidia-uninstall
3、安装cudnnv7
https://docs.nvidia.com/deeplearning/sdk/cudnn-install/index.html
下载完成以后将其解压到Cuda的目录当中，依次执行如下命令：
tar -xzvf 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*
运行一个小Demo即可.
如果安装了 例程和用户指南 这个包的话,我们可以找到位于 /usr/src/cudnn_samples_v7的mnistCUDNN这个小例子.
拷贝到 你的home/yourdir 任意文件夹下
$cp -r /usr/src/cudnn_samples_v7/ $HOME
进入 mnistCUDNN
$ cd $HOME/cudnn_samples_v7/mnistCUDNN
编译
$make clean && make
运行
$ ./mnistCUDNN
如果安装成功了,你会看到这样结果
Test passed!
其实还可以cmake 一下你的caffe/build,也能很快测试是否安装成功
13.安装gpu版的TensorFlow(先配置加速器)
$ sudo pip install tensorflow-gpu
root用户在根目录下新建.pip目录，在目录中创建文件pip.conf(/root/.pip/pip.conf)，配置内容如下,这里使用的清华源，还是挺快的：
[global]
index-url=https://pypi.tuna.tsinghua.edu.cn/simple
配置完成，无需任何操作，直接通过pip install即可安装任何想要的工具，再次来对比一下(输入pip install tensorflow之后立马截图就已经是如下图所示的效果)。
14.测试TensorFlow
走过前面的沟沟坎坎，终于到了测试这一步了，是不是很happy。
[root@gpuserver ~]# python
Python 2.7.5 (default, Nov 20 2015, 02:00:19)
[GCC 4.8.5 20150623 (Red Hat 4.8.5-4)] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> import tensorflow as tf
>>> hello = tf.constant('Hello, TensorFlow!')
>>> sess = tf.Session()
2018-12-12 17:10:51.572488: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 AVX512F FMA
>>> sess = tf.Session()
>>> print(sess.run(hello))
Hello, TensorFlow!
>>>
如果你可以正确的运行上面这个小的例子，那么恭喜你，gpu版的TensorFlow安装成功了，还等什么，赶紧造起来吧！
centos7.2安装pip
yum install -y epel-release
yum install -y python-pip
6、安装kernel-devel
yum -y install kernel-devel
centos7.2配置图形化界面启动
# systemctl get-default
multi-user.target
# systemctl set-default graphical.target
附录：
1、cuda安装过程记录
Installing the NVIDIA display driver...
Installing the CUDA Toolkit in /usr/local/cuda-10.0 ...
Missing recommended library: libGLU.so
Missing recommended library: libX11.so
Missing recommended library: libXi.so
Missing recommended library: libXmu.so
Installing the CUDA Samples in /root ...
Copying samples to /root/NVIDIA_CUDA-10.0_Samples now...
Finished copying samples.
===========
= Summary =
===========
Driver: Installed
Toolkit: Installed in /usr/local/cuda-10.0
Samples: Installed in /root, but missing recommended libraries
Please make sure that
- PATH includes /usr/local/cuda-10.0/bin
- LD_LIBRARY_PATH includes /usr/local/cuda-10.0/lib64, or, add /usr/local/cuda-10.0/lib64 to /etc/ld.so.conf and run ldconfig as root
To uninstall the CUDA Toolkit, run the uninstall script in /usr/local/cuda-10.0/bin
To uninstall the NVIDIA Driver, run nvidia-uninstall
Please see CUDA_Installation_Guide_Linux.pdf in /usr/local/cuda-10.0/doc/pdf for detailed information on setting up CUDA.
Logfile is /tmp/cuda_install_16878.log