• 自己离线环境下安装cuda10.0和cudnn7.6.5的过程

安装顺序：

1. 安装gcc
2. 显卡驱动
3. cuda和cudnn

安装gcc

sudo yum install gcc gcc-c++
sudo yum install make
sudo yum install kernel-devel
sudo yum install elfutils-libelf-devel   # 安装显卡驱动bug building kernel modules 需要安装 
# 在终端中输入： 
gcc --version 
# 结果显示：
gcc (GCC) 4.8.5 20150623 (Red Hat 4.8.5-39)

安装显卡驱动

1. 英伟达官网下载驱动 如下图选择对应的显卡型号驱动
   

chmod +x NVIDIA-Linux-x86_64-410.129-diagnostic.run
sudo sh NVIDIA-Linux-x86_64-410.129-diagnostic.run
# 安装提示下一步就ok 安装完成后
nvidia-smi
# 输出如下
Tue Aug 25 17:58:50 2020
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 418.87.00    Driver Version: 430.50       CUDA Version: 10.1     |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|===============================+======================+======================|
|   0  GeForce RTX 208...  Off  | 00000000:1A:00.0 Off |                  N/A |
| 13%   39C    P8     9W / 257W |    805MiB / 11019MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+
|   1  GeForce RTX 208...  Off  | 00000000:1B:00.0 Off |                  N/A |
| 19%   49C    P2    67W / 257W |   7791MiB / 11019MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+
|   2  GeForce RTX 208...  Off  | 00000000:3D:00.0 Off |                  N/A |
| 13%   32C    P8    15W / 257W |   9105MiB / 11019MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+
|   3  GeForce RTX 208...  Off  | 00000000:3E:00.0 Off |                  N/A |
| 19%   41C    P8    20W / 257W |     10MiB / 11019MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+
|   4  GeForce RTX 208...  Off  | 00000000:88:00.0 Off |                  N/A |
| 17%   38C    P8     6W / 257W |     10MiB / 11019MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+
|   5  GeForce RTX 208...  Off  | 00000000:89:00.0 Off |                  N/A |
| 20%   41C    P8     8W / 257W |     10MiB / 11019MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+
|   6  GeForce RTX 208...  Off  | 00000000:B1:00.0 Off |                  N/A |
| 13%   39C    P8     7W / 257W |     10MiB / 11019MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+
|   7  GeForce RTX 208...  Off  | 00000000:B2:00.0 Off |                  N/A |
| 24%   46C    P8     9W / 257W |   8835MiB / 11019MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+

• 安装的驱动430.50，对应需要安装的CUDA 版本是10.1，截图是另外一台机器截图

安装CUDA

1. 英伟达官网下载CUDA 下载对应的cuda版本 如下图

chmod +x cuda_10.1.105_418.39_linux.run  # 下载的CUDA 文件
sudo sh cuda_10.1.105_418.39_linux.run

# 由于驱动是好的不需要重装，安装时把Driver前面的X去掉
# 按着提示来装
Do you accept the previously read EULA?
accept/decline/quit: accept

# 因为第一步已经安装驱动，故这里选 N
Install NVIDIA Accelerated Graphics Driver for Linux-x86_64 367.48?
(y)es/(n)o/(q)uit: n

Install the CUDA 10.1 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

2. 安装完以后要修改环境变量 ， 如下

# added by cuda 10.0 installer
sudo vim ~/.bashrc
# 最后面添加如下两行 cuda安装文件夹要对应自己安装的版本
export PATH="/usr/local/cuda-10.0/bin:$PATH"'
export LD_LIBRARY_PATH="/usr/local/cuda-10.0/lib64:$LD_LIBRARY_PATH"'
# 保存退出
source ~/.bashrc
nvcc --version //查看是否安装成功 或者
nvcc -V
# 输出如下
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2018 NVIDIA Corporation
Built on Sat_Aug_25_21:08:01_CDT_2018
Cuda compilation tools, release 10.0, V10.0.130

安装CUDNN

1. 英伟达官网下载CUDNN 下载对应的cuda版本 如下图,， 需要注册登录，需要填写一个问卷

# 解压cudnn文件
tar -xvf cudnn-10.0-linux-x64-v7.6.5.32.tgz
sudo cp cuda/include/cudnn.h /usr/local/cuda-10.0/include/
sudo cp cuda/lib64/libcudnn* /usr/local/cuda-10.0/lib64/
# 为所有用户设置读取权限
sudo chmod a+r /usr/local/cuda-10.0/include/cudnn.h /usr/local/cuda-10.0/lib64/libcudnn*
#  查看cudnn版本，前三个数字表示版本号7.6.5
cat /usr/local/cuda/include/cudnn.h | grep CUDNN_MAJOR -A 2
# 输出如下
#define CUDNN_MAJOR 7
#define CUDNN_MINOR 6
#define CUDNN_PATCHLEVEL 5
--
#define CUDNN_VERSION (CUDNN_MAJOR * 1000 + CUDNN_MINOR * 100 + CUDNN_PATCHLEVEL)

#include "driver_types.h"
# 如果是cudnn 8版本以上的使用下面命令
cat /usr/local/cuda/include/cudnn_version.h | grep CUDNN_MAJOR -A 2