I am looking to implement a radeon pro graphics cards in a work server (either a radeon pro duo or wx 7100). I am currently running xen with arch linux as dom0 and a few more linux's running as domu's. I also have a windows 7 domu with a r9 390 GPU pass through.
Here is my question: Is it possible to use the mxgpu capabilities of the radeon pro cards to run multiple guests that output through the same graphics card? For example, I would like to run 2 instances of windows vm's that split the resources of the radeon pro card, and both of the vm's display directly to monitors connected to the same radeon pro card.
I know that I can accomplish this by using vnc to connect to the windows vm's, but that would require 2 other computers that only connect to the vnc and because it is vnc, then there will be a bit of lag.
If it is not possible to output multiple vm's using the same radeon pro card, then is there a lag free (or reduced lag) method of remote desktop that can utilize the full capability of the radeon pro card?
My end goal is to create a virtual reality development desk, where one person used cad to create a 3d model and the other person puts the model into vr and tests it out. This would traditionally require 2 computers, one for each person. Since I already run a server using vm's, I figure why not combine everything into the same machine. The server is underutilized therefore it should have enough power to run multiple windows instances easily.
If you need any details about my system, let me know. I figure this is a more theoretical question, so my specific system was not important.
Thanks in advance,