I love WSL1. With WSL2 and the move to Hyper-V, I just decided to drop WSL and manage my own VM manually.

I use the remote development tools on VSCode that power WSL to make everything feel like it's running on the host directly, just like WSL2.

This way I don't have any confusion born from it pretending not to be a VM. No issues with network port mappings, dns stuff, who owns what binary and how does it execute? When is WSL2 running and when is it not? etc etc

Ultimately, I would rather run Linux on my desktop as a daily driver but the desktop experience is not quite there yet for me to live (at least) 8 hours a day in. Gnome 40 looks great though, can't wait too see what the future holds.

My dream is running Linux as a daily driver where the UI is more polished than MacOS. Windows will run in a VM and will be used for API calls for playing games (I wonder if we will ever see some kind of GPU sharing for virtualisation).

> I wonder if we will ever see some kind of GPU sharing for virtualisation

Hasn't this been solved for some time now? Or am I misunderstanding what you want? I remember seeing a few articles about using this for gaming on Windows VMs on Linux hosts. I haven't jumped in though so maybe it's not what I think it is.

https://wiki.gentoo.org/wiki/GPU_passthrough_with_libvirt_qe...

I think it's also definitely in use for compute stuff on VMs in professional environments.

https://docs.microsoft.com/en-us/windows-server/virtualizati...

Edit: Ah, maybe sharing a GPU between a host and VMs isn't possible at the moment. Only pass-through and maybe sharing between the VMs?

Nope, Nvidia will never let you pass PCIe virtual functions to VMs. They'd rather charge 7x the price for data center GPUs with that capability.

Again: don't buy Nvidia. This is one of the many reasons why you should consider avoiding this brand (for GPUs). I even gave them the benefit of the doubt with their Nvidia Shield (bought two Pro 2020's) only to have them shove me up ads through my throat via a software update.

And for PCIe VFs for virtualized GPUs, AMD does _not_ provide it either outside of their datacenter line.

Note that if you just passthrough the whole GPU, that works out of the box on NVIDIA cards.

(btw, for some NV cards, there is: https://github.com/DualCoder/vgpu_unlock to crack that protection)