Hi NVIDIA Support,
We are evaluating NVIDIA vGPU 20.2 for GPU-P on Microsoft Hyper-V.
Environment:
Server: Dell Precision 7960 Tower
Host OS: Windows Server 2025 Datacenter
Hypervisor: Microsoft Hyper-V
GPU: NVIDIA RTX 6000 Ada Generation 48 GB
GPU Device ID: PCI\VEN_10DE&DEV_26B1
GPU Subsystem ID: SUBSYS_16A11028 (Dell)
Current working NVIDIA driver: 597.06
NVIDIA vGPU evaluation version: 20.2
vGPU Windows Server Host Driver: 596.84
The vGPU 20.2 host package contains nvgridswhostserver.inf. However, the RTX 6000 Ada entry in this INF specifies:
PCI\VEN_10DE&DEV_26B1&SUBSYS_16A110DE
Our Dell OEM RTX 6000 Ada reports:
PCI\VEN_10DE&DEV_26B1&SUBSYS_16A11028
Therefore, Windows Server does not consider the vGPU 596.84 host driver a matching driver for our GPU. The package can be added to the Driver Store successfully, but it does not bind to the GPU.
With the existing 597.06 driver, Get-VMHostPartitionableGpu detects the GPU and reports ValidPartitionCounts: {32}, but nvidia-smi -q reports:
Virtualization Mode: None
Host VGPU Mode: N/A
Could you please confirm:
Is Dell OEM RTX 6000 Ada SUBSYS_16A11028 supported for GPU-P with NVIDIA vGPU 20.2 on Windows Server 2025 Hyper-V?
If supported, is there a Dell/OEM-specific vGPU host driver containing this subsystem ID?
Is the generic nvgridswhostserver.inf package intended only for NVIDIA reference-board subsystem 16A110DE?
Is any GPU firmware/VBIOS or Dell-specific configuration required before enabling GPU-P?
Thank you.