GPU hardware detected but unable to start (error code 10)

Hi all,

Starting to bash my head against the wall on this one.

I’m running ESX6 and Cisco UCS C240 M4 with 2 K1s.

I can get things working with PCI passthrough, but I want to use vGPU.

To the best of my knowledge I’ve done all that is required - but when I build a test win7 machine (trying to keep it simple) I can’t seem to install the nvidia driver as it says GPU hardware not found. In device manager the item is there, albeit with a ! and and error saying it can’t be started (error 10).

I have installed GPU manager in the hypervisor (nvidia-smi) and it is showing use by this machine (as in the memory is allocated wrt the vGPU profile selected)

I’m sure I’ve had this working before and thought it was really simple to do. Something has changed and I can’t figure it out!.

If you look at the properties of the VM, tehre is a PCI device 0 that says unsupported (this says K1 Grid when using PCI passthrough).

I’m not a Vmware guru, so it’s possible I’ve missed something in the logs…

Any thoughts would be greatly appreciated.

Hi there,

We’d need a bit more info to check through where the issue may be happening but as a good place to start, do you have the deployment guide? If not, I would grab a copy from here (you need to register but the download is free): http://www.nvidia.com/object/grid-vgpu-deployment-guide.html

Dont forget that you need the right driver for the deployment mode, so if you are going to run vGPU you need the ‘NVIDIA GRID VGPU’ driver from our website for vSphere 6.0.

Hope that helps for now, shout back if not and we’ll see if we can find the issue.

Many thanks,
Sarah

This is a common error when the auxiliary power cable is either missing, or not fitted correctly.

As Sarah has mentioned, you need to use the correct driver for vGPU, so ensure you’ve downloaded the vGPU bundle which contains the vGPU manager .vib and the driver which is installed into the VM.

You should also ensure that if you’ve previously configured for vDGA, you remove the passthrough allocation as whilst this is in place, the GPU’s are not available for vGPU.

My video below takes you through the process of deploying vGPU enabled desktops on ESX with Horion, but it does assume a basic level of knowledge on vSphere and Horizon.

https://www.youtube.com/watch?v=7Ybq0hPvYHk