一、驱动安装
1.驱动安装
方法一：使用官方 NVIDIA 仓库（推荐）
sudo add-apt-repository ppa:graphics-drivers/ppa
sudo apt update
ubuntu-drivers devices
sudo ubuntu-drivers autoinstall
sudo reboot
方法二：使用标准 Ubuntu 仓库
sudo apt update
apt search nvidia-driver
sudo apt install nvidia-driver-535
方法三：使用 NVIDIA 官方 .run 文件
wget https://us.download.nvidia.com/XFree86/Linux-x86_64/535.154.05/NVIDIA-Linux-x86_64-535.154.05.run
chmod +x NVIDIA-Linux-x86_64-535.154.05.run
sudo systemctl isolate multi-user.target
sudo ./NVIDIA-Linux-x86_64-535.154.05.run
2.验证安装
nvidia-smi
3.常用管理命令
# 查看已安装驱动
dpkg -l | grep nvidia-driver
# 查看可用驱动版本
ubuntu-drivers list
4.重新安装驱动
sudo apt remove --purge nvidia-*
sudo apt autoremove
sudo apt install nvidia-driver-535
二、nvidia-docker安装
1.删除错误的仓库配置
sudo rm /etc/apt/sources.list.d/nvidia-docker.list
sudo apt-key del 0
2.获取正确的系统版本信息
distribution=$(. /etc/os-release; echo $ID$VERSION_ID)
3.添加 NVIDIA 容器仓库的 GPG 密钥
curl -fsSL https://nvidia.github.io/libnvidia-container/gpgkey | sudo gpg --dearmor -o /usr/share/keyrings/nvidia-container-toolkit-keyring.gpg
4.添加正确的仓库配置
curl -s -L "https://nvidia.github.io/libnvidia-container/$distribution/libnvidia-container.list" | \
sed "s#deb https://#deb [signed-by=/usr/share/keyrings/nvidia-container-toolkit-keyring.gpg] https://#g" | \
sudo tee /etc/apt/sources.list.d/nvidia-container-toolkit.list
5.进行安装
sudo apt update
sudo apt install -y nvidia-container-toolkit
6.添加运行时 /etc/docker/daemon.json，主要是 runtimes 代码块
{
"exec-opts": ["native.cgroupdriver=systemd"],
"graph": "/data/docker_storage",
"log-driver": "json-file",
"log-opts": {
"max-size": "100m"
},
"storage-driver": "overlay2",
"storage-opts": [
"overlay2.override_kernel_check=true"
],
"insecure-registries": ["172.16.4.17:8090", "152.136.254.160:8090"],
"registry-mirrors": [
"https://dockerpull.com",
"https://dockerproxy.cn",
"https://docker.anyhub.us.kg",
"https://dockerhub.jobcher.com",
"https://dockerhub.icu",
"https://docker.hpcloud.cloud",
"https://docker.1panel.live"
],
"live-restore": true,
"runtimes": {
"nvidia": {
"path": "/usr/bin/nvidia-container-runtime",
"runtimeArgs": []
}
},
"default-runtime": "nvidia"
}7.容器调用gpu测试
docker run --rm --gpus all nvidia/cuda:11.0-base nvidia-smi