NVIDIA Enables GPU Passthrough for Virtual Machines on Consumer-Grade GeForce GPUs

[AleksandarK]

Editor's note: This is not a part of April Fools.

NVIDIA has separated professional users and regular gamers with the company's graphics card offering. There is a GeForce lineup of GPUs, which represents a gaming-oriented version and its main task is to simply play games, display graphics, and run some basic CUDA accelerated software. However, what would happen if you were to start experimenting with your GPU? For example, if you are running Linux, and you want to spin a virtual machine with Windows on it for gaming, you could just use your integrated GPU as the GeForce card doesn't allow for virtual GPU passthrough. For these purposes, NVIDIA has its professional graphics card lineup like Quadro and Tesla.

However, this specific feature is about to arrive in even the GeForce lineup offerings. NVIDIA has announced that the company is finally bringing the basic virtual machine passthrough functions to the gaming GPUs. While the feature is representing a step in the right direction, it is still limited. For example, the GeForce GPU passthrough supports only one virtual machine and SR-IOV feature is still not supported on GeForce. "If you want to enable multiple virtual machines to have direct access to a single GPU or want the GPU to be able to assign virtual functions to multiple virtual machines, you will need to use NVIDIA Tesla, Quadro, or RTX enterprise GPUs.", says the NVIDIA FAQ. GeForce virtualization, which is now still in beta, is supported on R465 or higher drivers.

The full content from NVIDIA's website is written below.

NVIDIA has enabled GPU passthrough beta support for a Windows virtual machine on GeForce GPUs. What does this mean?
With virtualization enabled, GeForce customers on a Linux host PC can now enable GeForce GPU passthrough on a virtual Windows guest OS. There are a few GeForce use cases where this functionality is beneficial such as:

• GeForce customers wanting to run a Linux host and be able to launch a Windows virtual machine (VM) to play games
• Game developers wanting to test code in both Windows and Linux on one machine

What class of virtualization is supported on GeForce GPUs?
GeForce GPU passthrough supports 1 virtual machine. SR-IOV is not supported on GeForce. If you want to enable multiple virtual machines to have direct access to a single GPU or want the GPU to be able to assign virtual functions to multiple virtual machines, you will need to use NVIDIA Tesla, Quadro, or RTX enterprise GPUs.

Which driver is GeForce virtualization (beta) supported on?
GeForce virtualization (beta) is supported on R465 or higher drivers.

Which GeForce GPUs and Windows OSes support virtualization?
The feature is enabled on all GeForce/TITAN GPUs supported in the R465 driver (Kepler and later for Desktop; Maxwell and later for Notebook) for Windows 10.

Do you need to have more than one GPU installed or can you leverage the same GPU being used by the host OS for virtualization?
One GPU is required for the Linux host OS and one GPU is required for the Windows virtual machine.

View at TechPowerUp Main Site

 

[aktpu]

Is this april fools? Must be

 

[Camm]

I got excited thinking they enabled SR-IOV.

Nope

But at least its officially supported now rather than the previous work arounds.

 

[Shou Miko]

Great will this mean we won't be able to purchase a high-end Nvidia GPU again because oh no wait the server market also require them now.....

 

[stimpy88]

So to be clear, this is only for Linux?

If your using VMWare in Windows, it won't be hardware accelerated - direct to the GPU still?

 

[ixi]

April, april, nice one ngreedia.

 

[1d10t]

SR-IOV on consumer card or just another April's fool? AMD already turn me down me when they decided not to enable this feature on Vega.

edited

Pass-through on VDI, move along people nothing to see here.

 

[AleksandarK]

Not a part of April Fools. SR-IOV still not there unfortunately.

 

[hasherr]

"RTX enterprise GPUs." - which ones are those?

 

[silentbogo]

"RTX enterprise GPUs." - which ones are those?

Quadro RTX 6000/8000. AFAIK 4000 doesn't support vGPUs either, which sucks cause I have a few reasonably priced local options....

 

[Xajel]

Source: nvidia.custhelp.com
Date: 1 April.

 

[Haile Selassie]

That's one of the worst April Fools tricks ever. It's not funny, it is just mean.

 

[Raendor]

Has nothing to do with April. The news were out there a few days ago already. I wonder if this can be used to run something like virtualbox with Win7 for older games?

 

[tygrus]

Some Hypervisors + host OS + VM OS combinations allow PCIe / dGPU to be dedicated to a single VM. How is this different?

 

[TumbleGeorge]

Quadro RTX 6000/8000. AFAIK 4000 doesn't support vGPUs either, which sucks cause I have a few reasonably priced local options....

Where is my RTX A8000. Nvidia make 8000 model in older series which is last with "Quadro" in the name. But in last series cut max number to misery 6000 Where is my true flagship of last series?

 

[Solaris17]

what? MS has removed remoteFX you can only do it via DDA and you can already do it via DDA if the board supports it.

I'm not sure what this accomplishes. Things like proxmox andmaybe? vmware, already allow GPU passthough of consumer cards.

 

[Chomiq]

Source: nvidia.custhelp.com
Date: 1 April.

This was announced yesterday.

 

[Papahyooie]

This is good news, even if it's not SR-IOV. Previously to use nvidia cards under hypervisors like Proxmox and Unraid, you had to do some voodoo to get the drivers to load. Presumably now, it will be plug and play.

 

[TheoneandonlyMrK]

I bet the hardware/software lock that limits this to one virtual machine isn't hacked(lol not) any time soon, unlike their For the Gamer's mining lock.