1、 Win10安装最新nvidia显卡驱动(最新版本N卡驱动已经集成wsl)
The NVIDIA Windows GeForce or Quadro production (x86) driver that NVIDIA offers comes with CUDA and DirectML support for WSL
https://www.nvidia.com/Download/index.aspx?lang=zh-cn

2、Win10安装WSL2
https://wslstorestorage.blob.core.windows.net/wslblob/wsl_update_x64.msi
# 启动管理员PowerShell
Start-Process powershell -Verb runAs
# 启用 wsl 低於18362 的版本不支持WSL 2
dism.exe /online /enable-feature /featurename:Microsoft-Windows-Subsystem-Linux /all /norestart
# 启用虚拟机
dism.exe /online /enable-feature /featurename:VirtualMachinePlatform /all /norestart
# 设置 wsl 默认版本
wsl --set-default-version 2
# 查看可用Linux发行版
wsl -l -o
# 目前已安装发行版及版本
wsl -l -v
# 安装发行版
wsl -d Ubuntu-22.04 --cd 存储目录
# 进入子系统
bash
# 查看是否有显卡驱动
nvidia-smi
# nvidia-smi无效的解决办法,不用重装nvdia驱动
# 普通用户及root配置环境变量
vim ~/.bashrc
# 末尾加入
export PATH=$PATH:/usr/lib/wsl/lib
source ~/.bashrc
3、根据nvidia-smi,选择安装对应版本的cuda,选择Linux -> x86_64 -> WSL-Ubuntu -> 2.0 -> runfile(local)
https://developer.nvidia.com/cuda-toolkit-archive
# 安装gcc Cuda 安装需要
apt install -y build-essential
# 下载和安装
wget https://developer.download.nvidia.com/compute/cuda/12.1.0/local_installers/cuda_12.1.0_530.30.02_linux.run
sudo sh cuda_12.1.0_530.30.02_linux.run
安装出错:
Unable to find the kernel source tree for the currently running kernel
解决方法:
# 查看内核版本
root@DESKTOP-990CIL0:~# uname -r
5.15.90.1-microsoft-standard-WSL2
# 下载安装对应内核头文件及源码(安装失败)
https://kernel.ubuntu.com/~kernel-ppa/mainline/v5.15.90/
cd /opt
wget https://kernel.ubuntu.com/~kernel-ppa/mainline/v5.15.90/amd64/linux-headers-5.15.90-051590-generic_5.15.90-051590.202301240242_amd64.deb &&
wget https://kernel.ubuntu.com/~kernel-ppa/mainline/v5.15.90/amd64/linux-headers-5.15.90-051590_5.15.90-051590.202301240242_all.deb &&
wget https://kernel.ubuntu.com/~kernel-ppa/mainline/v5.15.90/amd64/linux-image-unsigned-5.15.90-051590-generic_5.15.90-051590.202301240242_amd64.deb
sudo dpkg -i *.deb

# 下载安装对应内核头文件及源码(WSL2)
https://github.com/microsoft/WSL2-Linux-Kernel
cd /usr/src
wget https://github.com/microsoft/WSL2-Linux-Kernel/archive/refs/tags/linux-msft-wsl-5.15.90.1.tar.gz &&
tar -xzvf linux-msft-wsl-5.15.90.1.tar.gz
# 修改环境变量
vim ~/.bashrc
# 文件未追加
export PATH=/usr/local/cuda/bin${PATH:+:${PATH}}
export LD_LIBRARY_PATH=/usr/local/cuda/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}
# reload 环境变量配置
source ~/.bashrc
# 检查是否生效
nvcc -V
# 测试cuda
apt install -y git
cd /opt && git clone https://github.com/NVIDIA/cuda-samples.git
cd ./cuda-samples/Samples/1_Utilities/deviceQuery
make
./deviceQuery
# 输出Pass 则成功了
4、安装cuda对应的cudnn
https://developer.nvidia.com/rdp/cudnn-archive

Linux:
/mnt/XXX
tar -xf ./cudnn-linux-x86_64-8.8.0.121_cuda12-archive.tar.xz
cd cudnn-linux-x86_64-8.8.0.121_cuda12-archive
cp -P ./lib/* /usr/local/cuda-12.1/lib64/
cp -P ./include/* /usr/local/cuda-12.1/include/
# 查看cudnn版本
cat /usr/local/cuda/include/cudnn_version.h | grep CUDNN_MAJOR -A 2

5、安装nvidia-docker2
curl https://get.docker.com | sh
distribution=$(. /etc/os-release;echo $ID$VERSION_ID) \
&& curl -s -L https://nvidia.github.io/nvidia-docker/gpgkey | sudo apt-key add - \
&& curl -s -L https://nvidia.github.io/nvidia-docker/$distribution/nvidia-docker.list | sudo tee /etc/apt/sources.list.d/nvidia-docker.list \
&& curl -s -L https://nvidia.github.io/libnvidia-container/experimental/$distribution/libnvidia-container-experimental.list | sudo tee /etc/apt/sources.list.d/libnvidia-container-experimental.list
sudo apt update
sudo apt-get install nvidia-docker2
service docker start
# 搜索符合当前cuda版本的镜像
https://hub.docker.com/r/nvidia/cuda
# 拉取镜像
docker pull nvidia/cuda:12.0.1-devel-ubuntu22.04
# nvidia-docker 测试
sudo docker run -idt --name nvidia_docker_test --gpus all --shm-size 16G nvidia/cuda:12.0.1-devel-ubuntu22.04
sudo nvidia-docker start nvidia_docker_test
sudo nvidia-docker attach nvidia_docker_test
# 查看是否有显卡驱动
nvidia-smi
# 有则判定 nvidia-docker 已经成功安装和使用
exit