Lets say I have a PC running Windows 10 natively with two GPUs in it - a Quadro P2200 and a GeForce RTX2070 Super, each connected to a separate monitor. Further, lets say that I want to use the Quadro for work tasks and the GeForce card for gaming.
My question is simple - is this actually possible, and if so, how does it actually work? I am currently putting together a PC to do this where the work and play GPUs are in separate VMs with VFIO etc, but it turns out I have no idea if its possible to combine the two VMs into one (i.e. have one VM with two GPUs passed through). So lets begin with the non-VM case described above.
How does Windows “know” to use the Quadro for work tasks and the GeForce for play? Does it depend on which monitor (and thus GPU) the application is run on? What if the gaming card was from AMD? Is Windows able to handle having two different GPU driver stacks installed at the same time? Does anyone have experience doing something like this?
I think it’s going to depend on the app/game. I think games use the GPU the monitor is attached to. For work stuff like rendering, I think the “work” apps let you pick what devices to render on. I have Blender and DAZ Studio and they let you pick the render devices.
So by this logic, you need to be sure to start applications on the correct monitor - that is, the one connected to the appropriate GPU. How exactly? Do you end up with a separate, stand-alone task bar per monitor (i.e. not a single mirrored task bar), and its a simple case of just using the right task bar to launch the application? I do not have two GPUs in my native Windows box so am unable to answer this question for myself.
I would be extremely surprised if you get a picture at all (on both screens) when plugging them into completely different cards like that. Not even sure that works with SLI even.
Ha…
Apparently, that does actually work. You can even play games on the card that is not plugged in as long as the game lets you select the card. Interesting.