I want to assign GPU to VMs created with Hyper-V in my Windows Server 2022 using GPU pool , but vm start fail

刘先生 5 Reputation points
2025-01-03T08:19:29.18+00:00

I want to assign GPUs to VMs created with Hyper-V in my Windows Server 2022 using GPU pool .

I have NVIDIA GPU Tesla M60 plugged into Windows Server 2022, and create GPU Pool .

  • Host Server : Windows Server 2022
  • NVIDIA GPU : Tesla M60
  • VM : using Windows 2019
  • RemoteFX will not be used.

After I selected a pool from the GPU pool and assigned it to a highly available virtual machine, the virtual machine failed to start, with a prompt indicating that multiple devices had been allocated to the virtual machine.
However, if I assigned this pool to a non-highly available virtual machine on the host, it could start normally. May I ask if there are any special settings for highly available allocation?

Thank you.

Hyper-V
Hyper-V
A Windows technology providing a hypervisor-based virtualization solution enabling customers to consolidate workloads onto a single server.
2,787 questions
{count} votes

2 answers

Sort by: Most helpful
  1. Anthony LAVABRE 1,225 Reputation points
    2025-01-04T17:43:06.8366667+00:00

    Hi,

    The issue may be caused by a GPU allocation conflict between highly available (HA) virtual machines. You can check the exclusive GPU allocation, HA compatibility or otherwise event logs.

    You can check this link too :

    https://learn.microsoft.com/en-us/windows-server/virtualization/hyper-v/gpu-partitioning?pivots=windows-server

    https://learn.microsoft.com/en-us/windows-server/virtualization/hyper-v/deploy/use-gpu-with-clustered-vm?tabs=windows-admin-center&pivots=windows-server

    Sincerely,

    0 comments No comments

  2. Hania Lian 20,271 Reputation points Microsoft Vendor
    2025-01-06T12:05:46.8033333+00:00

    Hello,

    When assigning GPUs to a highly available VM, there are some additional considerations and configurations that need to be addressed to ensure that the HA feature does not interfere with the GPU allocation.

    Here are a few things you might consider:

    1、Ensure that the HA configuration allows for the specific allocation of hardware resources like GPUs. Sometimes, HA settings may prevent certain types of resource allocations to maintain failover readiness.

    2、Make sure that the NVIDIA drivers and software components installed on the host server are compatible with the version of Windows Server 2022 you’re using and support the specific GPU model (Tesla M60). Additionally, check if there are any specific requirements or recommendations by NVIDIA for setting up GPU passthrough in a highly available environment.

    3、Verify that the GPU passthrough settings are correctly configured both on the host and within the VM. This includes ensuring that the correct devices are being passed through and that the VM is configured to use these devices properly.

    4、Review the VM’s settings to confirm that it is set up to receive and utilize the GPU resources. This includes checking the device assignment settings and ensuring that the VM is not configured in a way that conflicts with the HA requirements.

    Best Regards,

    Hania Lian

    ============================================

    If the Answer is helpful, please click "Accept Answer" and upvote it.

     

    0 comments No comments

Your answer

Answers can be marked as Accepted Answers by the question author, which helps users to know the answer solved the author's problem.