Nvidia Docker on WSL2
Published in
4 min readSep 13, 2022
In 2022 if we intend to build containers with GPU supports on Windows, the only solutions is WSL2 (Windows Subsystem). This tutorial guides step-by-step to install from scratch.
Hardware Environment
CPU: Intel(R) Xeon(R) W-2223 CPU @ 3.60GHz
RAM: 8.00 GB
GPU: NVIDIA Quadro P1000
Software Environment
Edition: Windows 11 Pro for Workstations
OS Build: 22000.795
BIOS: HP Sure Starter
Step by Step Instruction
1. Install Nvidia driver for specific GPU on Windows from Official Drivers | NVIDIA, and make sure your nvidia-smi works normally.
2. Install Docker Desktop for Windows from Docker Desktop WSL 2 backend, and enable WSL2 based engine in the Docker setting.
3. Enable Virtualization Technology on your BIOS. For example: HP PCs — Enable Virtualization Technology in the BIOS | HP® Customer Support
4. Install WSL and Linux distribution (here we use Ubuntu 20.04) in the Microsoft Store step by step: Install Ubuntu on WSL2 on Windows 11 with GUI support
5. Enable the Docker integration with WSL for the Ubuntu distribution in Docker setting.
6. Install CUDA on WSL2, Run the following commands by CUDA on WSL User Guide
# set default WSL engine to WSL2
C:\> wsl.exe --set-default-version 2# show all linux distribution
C:\> wsl.exe --install# start the Ubuntu system by WSL2
C:\> wsl.exe -d Ubuntu-20.04
In the Ubuntu shell:
# First, remove the old GPG key:
$ sudo apt-key del 7fa2af80# Install CUDA
$ wget https://developer.download.nvidia.com/compute/cuda/repos/wsl-ubuntu/x86_64/cuda-wsl-ubuntu.pin
$ sudo mv cuda-wsl-ubuntu.pin /etc/apt/preferences.d/cuda-repository-pin-600
$ wget https://developer.download.nvidia.com/compute/cuda/11.7.0/local_installers/cuda-repo-wsl-ubuntu-11-7-local_11.7.0-1_amd64.deb
$ sudo dpkg -i cuda-repo-wsl-ubuntu-11-7-local_11.7.0-1_amd64.deb
$ sudo apt-get update
$ sudo apt-get -y install cuda
7. Install Nvidia Docker on Ubuntu by INSTALLING DOCKER
# Add the package repositories:
$ 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# Download information from all configured sources
$ sudo apt-get update && sudo apt-get install -y nvidia-container-toolkit# Restart the Docker service
$ sudo systemctl restart docker
8. We are done here, let just check the GPU capability in the Ubuntu by WSL2.
# Run docker container to test GPU capability
$ sudo docker run --gpus all nvcr.io/nvidia/k8s/cuda-sample:nbody nbody -gpu -benchmark
