Now, I'm sure many of us are aware of NVIDIA's "bug" that prevents the vGPU pass-through of standard GeForce cards; and, despite being a well-discussed topic in the community beyond just these forums, there is little information on how to solve that issue for a broad range of environments. This is, technically, a niche issue; but, there are a considerable number of companies out there who don't have the budget to go purchase GRID GPUs. high-end Quadros, or Teslas.
My question is... has anyone had any success with "hacking" this under XCP and/or does the XCP team have any out of the box ideas that might "workaround" this limitation similar to the way KVM did so? While I understand that much of this is limited to the type of virtualization Xen performs in regards to VMs, one can hope that someone comes up with a creative way to get around it. I know for sure that there is a driver version before 330 something where they weren't performing this machine check in the NVIDIA driver; but, that automatically eliminates several models of GPUs as candidates.
I know for a fact that both the 1030 and 1060 will not pass-through in vGPU mode to Windows VMs. I'm still messing around with a couple Linux VMs to try and circumvent it. I am also looking into doing manual PCI pass-through and device hide through GRUB (reportedly has had spotty success) instead of using the interface.
On a related, but, secondary, note - to those who have run into this issue and invested in alternative GPUs - what are you using and how successful has it been?