Install WSL2 - Run Linux on Windows with GPU Support

Requirements

This is a new process that is much simpler then before but you need Windows 11 or Windows 10 with the latest updates (at least the 2004 update).

Also make sure your hardware supports virtualization.

Run msinfo32 at the command line, you should see something like this:

Setup WSL2

wsl --install

You should get something like this:

After you reboot you can start the setup process for Ubuntu

You can now start using Ubuntu!

If you want to list the WSL2 distributions currently installed on your machine:

wsl --list --verbose

GPU support / CUDA

If you have a NVDIA GPU at your Linux distro command line type

nvidia-smi

You should see your card listed with the driver number.

If you need install CUDA and Cudnn libs necessary for tensorflow to work you simply can install CUDA the usual way For example CUDA 10 The same command lines apply except that you skip the driver installation part

Access linux files

In File explorer address bar type:

\\wsl.localhost

You should have access to your linux distribution’s entire filesystem