This is pretty interesting. I read through some of the details to see if this could also help with running local LLMs/rendering as I've had Plasma freeze while trying to draw some animation or blur when the VRAM is filled to the brim, but it doesn't seem like that's the case since it is made to prioritize and protect the memory of the foreground application (as the focused window). So it should work great for games but might actually make things worse if your VRAM load is in a daemon or background process