Tesla M10 on Dell R740 Server 2016 Standard

NVIDIA Virtual GPU Forums Virtual GPU NVIDIA Virtual GPU Drivers
#1

Hi,

We have a Dell R740 with two Tesla M10 cards. The Nvidia Grid (version 6.2) installs without issue except the reboot to complete the install freezes on the windows symbol.

We have have patched windows to the latest level, tried an older version of vgpu driver and still the same issue.

Prior to rebooting to complete the install the M10s appear in the device manager.

Any ideas?

Thanks

#2

Hi

Which Hypervisor are you using?

If you’re watching the VM through the Hypervisors VM Console, then the VMs normally freeze at that point once the GPU driver has been installed, this is normal behavior. Make sure you have RDP enabled on the VM so you can connect remotely after the GPU driver has been installed.

Regards

Ben

#3

Hi,

This is the installation of the grid software on the host OS and the post install reboot to complete the install.

It freezes just before reaching the windows login.

Ive done a clean install of Server 2016 with no hypervision role setup and still the same issue

Thanks

#4

PS we will be using HyperV direct attach once we have solved this problem

I have tried change all Memory I/O options (56TB, 12TB and 512GB) options as well

#5

Hi

Sorry, I’ve never used Hyper-V, so no idea what’s required. Any of the other supported platforms, then I could help, but not Hyper-V I’m affraid.

Regards

Ben

#6

Hi Ben,

We are just trying to get the nvidia vGPU drive installed on a bare bones 2016 Server. The driver simply will not install.

Thanks

Jon

#7

Hi

If this is your Hyper-V Host, and you’re going to be running 8 VMs on it using Passthrough / DDA, then why are you installing the NVIDIA driver on it?

Again, sorry, I don’t use Hyper-V, so don’t know if Hyper-V requires it. But with the other Hypervisors (XenServer, vSphere, KVM), you only need to install the driver in the Hypervisor if you’re going to be using vGPU. If you’re going to be using Passthrough / DDA, then the driver has nothing to do, so there’s no point installing it. For Passthrough / DDA, the driver should only be installed in the VMs with GPUs attached, not on the Hypervisor.

Does that help at all?

Regards

Ben

#8

Hello all,

Was there ever a solution to the problem JonP? I am having the exact same problem