Links
Comment on page

Windows

Everything you need to know about deploying Supervisely agent on Windows WSL
Poster

Deploy Supervisely agent with GPU on Windows WSL

This tutorial explains how to deploy the Supervisely agent with GPU support on Windows Subsystem for Linux (WSL). Whether you prefer following a video tutorial or a text-based guide, we've got you covered. In this guide, we'll walk you through the process step by step.
If you're a visual learner and prefer to watch along, check out our comprehensive video guide on deploying the Supervisely agent with GPU on Windows WSL. This guide will take you through each step visually, making it easier to follow along and set up the environment correctly.
Machine Specs used in video guide:
  • Operating System: Microsoft Windows 10 Enterprise (10.0.19045 Build 19045)
  • GPU: NVIDIA GeForce RTX 4090
  • GPU Driver Version: 536.67
  • UBUNTU: 22.04.2 LTS
  • Docker Desktop Version: 4.20.1 (110738)
Video guide
If you prefer written instructions and a more detailed breakdown, here's how you can deploy the Supervisely agent with GPU on Windows WSL:
Machine Specs used in text guide:
  • Operating System: Microsoft Windows 11 Pro (10.0.22621 Build 22621)
  • GPU: NVIDIA GeForce RTX 3080 Ti (Laptop)
  • GPU Driver Version: 536.67
  • UBUNTU: 22.04.2 LTS
  • Docker Desktop Version: 4.1.1 (69879)

Table of Contents

Prerequisites

  • Windows 10 Home, Pro or Enterprise (64-bit edition). Version 1903 or higher, with Build 18362 or higher.
or
  • Windows 11 Home, Pro or Enterprise (64-bit edition).
and

How to install

Step 1. Turn on WSL

Use windows search to find "Turn Windows features on or off" and open it.
Turn Windows features on or off
Scroll down and locate "Windows Subsystem for Linux", check the box and restart your computer. If the box is already checked proceed to the next step.
Windows Subsystem for Linux

Step 2. Install Windows Terminal

Open Microsoft Store and find Windows Terminal and press Get.
Windows Terminal

Step 3. Install Ubuntu

Open Microsoft Store and find Ubuntu 22.04.2 and press Get.
Ubuntu 22.04.2

Step 4. Install NVIDIA GPU Driver

Go to NVIDIA site and download the latest driver for your GPU.
Fill the form and press Search.
NVIDIA Search
Press Download button and install the driver.
NVIDIA Download

Step 5. Docker Desktop

Download Docker Desktop and install it.
If you have problems running Docker Desktop, check out the possible problems when running Docker Desktop just below.
Possible problems when running Docker Desktop
Docker Desktop - WSL Kernel version too low
​
Docker Desktop WSL Kernel version too low
​
Open Windows Terminal and run the following command:
wsl --update
Docker Desktop Windows Hypervision is not present
​
Docker Desktop Windows Hypervisor is not present
​
Restart you computer and go to BIOS settings and enable Virtualization.
Docker Desktop Resources - You don't have any WSL 2 distros installed
​
Docker Desktop Resources - You don't have any WSL 2 distros installed
​
In this case you need to update your WSL distro to version 2.
Open Windows Terminal and run the following commands:
  1. 1.
    Get name of your WSL distro
wsl -l -v
Output:
NAME STATE VERSION
* Ubuntu-22.04 Running 1
docker-desktop-data Running 2
docker-desktop Running 2
  1. 2.
    Update your WSL distribution to version 2
wsl --set-version Ubuntu-22.04 2
Output:
Conversion in progress, this may take a few minutes.
The operation completed successfully.
  1. 3.
    Set default WSL version to 2
wsl --set-default-version 2
Output:
The operation completed successfully.
Open Docker Desktop and go to Settings -> Resources -> WSL integration. Check "Enable integration with my default WSL distro" and "Ubuntu 22.04" and press Apply & Restart as shown below.
Docker Desktop Resources
Open Docker Desktop and go to Settings -> Docker engine and add runtime to the docker config file as shown below and press Apply & Restart:
{
"default-runtime": "nvidia",
"runtimes": {
"nvidia": {
"path": "nvidia-container-runtime",
"runtimeArgs": []
}
}
}
Or you can copy and paste merged config file from here:
{
"builder": {
"gc": {
"defaultKeepStorage": "20GB",
"enabled": true
}
},
"experimental": false,
"features": {
"buildkit": true
},
"runtimes": {
"nvidia": {
"path": "/usr/bin/nvidia-container-runtime",
"runtimeArgs": []
}
}
}
Docker engine

Step 6. Install NVIDIA Container Toolkit

Open Ubuntu terminal via Windows terminal
Ubuntu Windows Terminal
Install NVIDIA Container Toolkit repository for your distribution by running the following command:
distribution=$(. /etc/os-release;echo $ID$VERSION_ID) \
&& curl -fsSL https://nvidia.github.io/libnvidia-container/gpgkey | sudo gpg --dearmor -o /usr/share/keyrings/nvidia-container-toolkit-keyring.gpg \
&& 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
Update the APT repository cache and install the nvidia-container-toolkit package:
sudo apt-get update
sudo apt-get install -y nvidia-container-toolkit
sudo nvidia-ctk runtime configure --runtime=docker
Restart Docker Desktop.
Enter the following command to verify that the installation was successful:
sudo docker run --rm --runtime=nvidia --gpus all nvidia/cuda:12.2.0-runtime-ubuntu22.04 nvidia-smi
If you have problems running this container, try upgrading CUDA Toolkit to latest version.
After docker image is pulled, you will see nvidia-smi output.
NVIDIA SMI

Step 7. Deploy Supervisely Agent

Deploy Supervisely Agent with GPU support on Windows WSL.
Open Supervisely instance and go to the Start -> Team Cluster page and press "Add" button
Add Agent
Select "Supervisely agent".
Select Agent
In the modal window go to "advanced settings" and check "Use nvidia runtime" option to enable GPU support.
Agent Settings
Copy the instructions command and run it in the Ubuntu terminal.
Agent Instructions
After the agent docker image is pulled you will see this message in the terminal. It means that agent is successfully deployed.
Agent Instructions
Go to the Team Cluster page and open your agent, check that agent is running. That's it! Now you can run GPU tasks on your Windows machine.
Agent Running
Last modified 3mo ago