APU drivers for GPU inside VM--does it work?

I guess if the news hasn’t already swept the internet by storm, then it isn’t possible to use an APU (Ryzen 2200G) in a virtual machine while utilizing it’s built in GPU in that VM. What might be preventing this capability, and what advancements would be made to enable this transition in capabilities of the virtual machine / drivers?

Ah, I just remembered, a GPU cannot share it’s resources with more than one system–would the following be possible:

Run the operating system with a dedicated GPU, and passthrough the onboard vega GPU to a VM?

I doubt it would work. You won’t be able to separate the graphics for passthough unless they are in their own IOMMU group, which is unlikely as the graphics reside on the CPU. The less ideal solution would be to use the APU for your graphics natively, and pass the dGPU through instead.