ubuntu安装nvidia驱动 - Leonardo

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