Is it possible to use a remote V100 in a server for general display/OpenGL tasks in Windows Server?

Professional Graphics and Rendering System Management Virtual GPU
#1

Hi there.

We’ve got a few V100’s in a bunch of new Dell R740 series hosts.

One of our people would like to run Windows Server 2016 on one of these hosts and use strictly Windows based OpenGL + Direct3D accelerated applications on these hosts. They can’t run in Linux.

My question is - is this actually possible with the V100 card using RDP? Or TurboVNC? Can the V100 actually accelerate general/standard GUI’s and “full desktop experience” in a Windows environment - just as a Quadro or consumer grade card, can?

Thanks.

#2

Hi,

By default, when Windows NIVIDIA Driver is installed, Tesla GPUs are in TCC mode. To use a V100 for general display/OpenGL task in Windows Server, you would need to put the GPU in WDDM mode, and that is only possible if you have a GRID license.

Thanks,
Ryan Park

#3

Can you change a V100 on a cloud vendor’s instance from TCC to WDDM mode? For example, on AWS’ P3 instance with a V100, can you launch one of these instance and then switch from TCC to WDDM mode. How do you change from TCC to WDDM? Is there a utility? How do you get a grid license?

Thanks,

#4

I’m curious about this too. The V100s from the cloud providers (Amazon/Azure) will be one of the few ways to experiment with the new ray tracing parts of the Gameworks tools announced at GDC. It would be great to be able to use cloud V100s for this, but I guess they need to be able to run DirectX apps.

Is there an easy way to do this? How do you get a license to do this, and does it cost?

Many thanks for any help.