GPU-P Assistance for insano's

Hoping I came to the right place. I have a technical question that for the first time has stumped me in over 15 years.

I multibox 7x Diablo 2 Resurrected clients with Hyper-V clients, using GPU-Passthrough. I have been doing this on 2x GTX 1080’s ran in SLI for months, but while streaming at the same time, the cards were no longer able to keep up with the demand vs the quality I wanted.

I upgraded to 2x RTX 3090’s (non sli as of yet).

After I get to between 2-4 VM’s open and games launched, the game that’s running on my PHYSICAL machine freezes with no recovery (not the vm’s). An even bigger problem is, if I end the task, it BSOD’s the physical machine (Video_Scheduler_Internal_Error). Even more curious is, while the physical machine’s game is frozen, all the VM’s games continue to operate just fine (including d2r), and I can continue to use both my physical computer and the vm’s indefinitely, including games / etc, UNTIL I close the frozen instance of D2R. Then I BSOD. I think this last part is the most confusing to me.


-I have tried removing 1 card and just using 1 card as well
-I reconfigured my existing VM’s as well as created totally new ones.
-I tried monitoring for power fluctuation AND heat issues - no evidence of issues
-I’ve ran furmark to stress test - Table Stable
-I’ve done a full purge of all old Nvidia drivers to both physical and VM environments
-I’ve reran the scripts to allocate GPU partitions
-I’ve tried 3 different driver versions for the 30xx series cards
-I’ve tried lowering my graphics settings of the physical machine back down to identical to what I had it on while running on the 1080’s successfully.
-I’ve disabled DLSS
-I’ve tried installing a single 3090 and a 1080 as a secondary, and only passing through the previously working 1080, but still end up dead in the water.
-I’ve tried BOTH the 3090’s as individuals to rule out the possibility of a bad card.

I’m missing something. Please help.

Tech Specs;

Ryzen 9 3900x
MSI X370 Gaming Plus Mobo (Yes I know this brings limitations with 8x config 2nd pcie slot)
64 Gig DDR4
EVGA RTX 3090 x2
EVGA 1300W Gold Rated PSU

The ONLY thing I changed, was the stupid cards. Every other piece of hardware is known to function doing this for over a month.

-Script used for partition;

$vm = “D2 MBox1”

Add-VMGpuPartitionAdapter -VMName $vm
Set-VMGpuPartitionAdapter -VMName $vm -MinPartitionVRAM 80000000 -MaxPartitionVRAM 100000000 -OptimalPartitionVRAM 100000000 -MinPartitionEncode 80000000 -MaxPartitionEncode 100000000 -OptimalPartitionEncode 100000000 -MinPartitionDecode 80000000 -MaxPartitionDecode 100000000 -OptimalPartitionDecode 100000000 -MinPartitionCompute 80000000 -MaxPartitionCompute 100000000 -OptimalPartitionCompute 100000000

Set-VM -GuestControlledCacheTypes $true -VMName $vm
Set-VM -LowMemoryMappedIoSpace 1Gb -VMName $vm
Set-VM –HighMemoryMappedIoSpace 32GB –VMName $vm

I’ll literally buy someone lunch if they can help get me back going with this. First time posting here so not sure what to expect.

Day 3 of diagnosis and I think I finally fixed it. Think it had to do with resolution settings of the boxes being too small versus the screen sizes. I can no longer use 800x600 on the vboxes which stinks, but even after upping it to 720p across all the boxes, it’s still less than half the strain on the cards compared to 1080’s, so I’m flying now.

Thanks to anybody who may have thought about it.