CloudXR Support for Pass Through V100

Hi CloudXR team,

We are setting up an on-premise VDI cluster to test out the cloudXR streaming capabilities.

However, we are unable to install cloudXR v3.0 on the server, received an error “No CloudXR capable GPU found”.

I will like to check with the CloudXR team on the support for a Tesla V100 running the standard pass through drivers (471.41) as we did not setup our V100 on a Grid vGPU solution.

Thanks

Hi

You can use the V100, but you must use the vGPU driver (not standard Tesla driver) to go with it which will require an RTX vWorkstation license.

If you don’t currently use vGPU, you can register for a free 90 day evaluation here: NVIDIA Enterprise Account Registration.

If you’re running VR with vGPU, then you’ll be able to run a maximum of 2x users from the V100 depending on the application type and your HMD choice. If you’re running AR, then you’ll be able to run more than that, again depending on your application and device type.

Regards

MG

Hi MG,

If I am spinning up my virtual desktops purely for CloudXR streaming purposes, RTX Workstation license is recommended? How about vApps or vPC licenses? From what I read, RTX workstation licenses support up to 8k resolution, is that the reason for the recommended license?

Also we do have plans to virtualize the same V100 GPU for AI workloads, will there be any problems if we apply seperate AI compute licenses for other vGPU profiles?

Cheers,
Alvin

Hi Alvin

vPC / vApps licenses will not unlock enough of the GPUs features / performance to be able to support that kind of workload. These licenses are for basic applications, and are not suited to XR where you really need the GPUs full performance and features.

RTX vWorkstation is the top tier license. It supports all workloads, including vCompute. So you can use your RTX vWorkstation license to support multiple Compute instances.

Regards

MG