1.离线安装nvidia驱动

  • NVIDIA驱动需要GCC等C/C++开发环境```shell

    gcc -v

    如果系统显示没有找到GCC指令，或没有显示GCC版本信息，则需要安装C/C++环境 ```

  • 安装gcc```shell

    yum -y install gcc gcc-c++ kernel-devel

    ```

  • 查看GPU型号```shell

    lspci | grep -i vga

    lspci | grep -i nvidia

lspci -v -s 00:0f.0(上一步得到)

  2. - 禁用Nouveau驱动```shell
  3. # vim /etc/modprobe.d/nouveau.conf
  4.   blacklist nouveau
  5.   options nouveau modeset=0
  6. # cp /boot/initramfs-$(uname -r).img /boot/initramfs-$(uname -r).img.bak
  7. # dracut /boot/initramfs-$(uname -r).img $(uname -r)
  • 改为终端模式启动```shell

    systemctl set-default multi-user.target

    reboot

lsmod | grep nouveau

输出为空即为OK


- 下载对应型号的离线安装包<br />[nvidia-官网](https://www.nvidia.cn/Download/index.aspx?lang=cn)
- 安装<br />必须指定kernel source path，否则会报错；kernel的版本和系统内核有关，可能会有差别```shell
# ./NVIDIA-Linux-x86_64-418.87.00.run --kernel-source-path=/usr/src/kernels/3.10.0-957.el7.x86_64  -k $(uname -r)

参考：https://www.cnblogs.com/YSPXIZHEN/p/11466145.html

2.离线安装cuda

  • 下载离线包：https://developer.nvidia.com/cuda-downloads
  • 安装：```shell

    ./cuda_….run

    ```
    多等待一会，会跳出对话框，让你勾选要安装的软件，然后Install即可
  • 安装成功后需要配置PATH：```shell

    vim /etc/profile

    (末尾添加对应cuda路径，安装成功后会给你对应信息) export PATH=/usr/local/cuda/bin:$PATH export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PATH

    source /etc/profile

    ```

3.离线安装docker(19.03.6)

在 http://rpm.pbone.net/index.php3?stat=3&limit=1&srodzaj=1&dl=40&search=container-selinux&field[]=1&field[]=2 下载container-selinux….rpm

  • 安装
    将8个依赖复制到/root/docker
    将docker-ce、docker-ce-cli、containerd-io和container-selinux复制到/root/docker/rpm```shell

    rpm -ivh /root/docker/*.rpm

    rpm -ivh /root/docker/rpm/container-selinux

    rpm -ivh /root/docker/rpm/container-io

    rpm -ivh /root/docker/rpm/docker-ce-cli

    rpm -ivh /root/docker/rpm/docker-ce

    ```

  • 启动```shell

    systemctl start docker

docker info



参考：[https://blog.csdn.net/u013058742/article/details/80075633](https://blog.csdn.net/u013058742/article/details/80075633)

<a name="7a90073f"></a>
### 4.离线安装nvidia-docker2

前提，已经安装了docker

- 下载离线包：```shell
# distribution=$(. /etc/os-release;echo $ID$VERSION_ID)
# curl -s -L https://nvidia.github.io/nvidia-docker/$distribution/nvidia-docker.repo | sudo tee /etc/yum.repos.d/nvidia-docker.repo

# yum install --downloadonly nvidia-docker2 --downloaddir=/tmp/nvidia


得到五个安装包：```shell libnvidia-container1-1.3.0-1.x86_64.rpm libnvidia-container-tools-1.3.0-1.x86_64.rpm nvidia-container-runtime-3.4.0-1.x86_64.rpm nvidia-container-toolkit-1.3.0-2.x86_64.rpm nvidia-docker2-2.5.0-1.noarch.rpm


- 安装```shell
# rpm -ivh libnvidia-container1-1.3.0-1.x86_64.rpm libnvidia-container-tools-1.3.0-1.x86_64.rpm

# rpm -ivh nvidia-container-runtime-3.4.0-1.x86_64.rpm nvidia-container-toolkit-1.3.0-2.x86_64.rpm nvidia-docker2-2.5.0-1.noarch.rpm

参考：https://codeleading.com/article/99064052095/