[wmf]

The use case is allowing the host system and VM(s) to access the same GPU at the same time.

[jandrese]

You have a Linux box but you want to play a game and it doesn't work properly under Proton, so you spin up a Windows VM to play it instead.

The host still wants access to the GPU to do stuff like compositing windows and H.265 encode/decode.

[skykooler]

And outputting anything to the screen in general. Usually, your monitor(s) are plugged into the ports on the GPU.

[ur-whale]

Yeah, I got that from the technical explanation.

What's the practical use case, as in, when would I need this?

[EDIT]: To maybe ask a better way: will this practically help me train my DNN faster?

Or if I'm a cloud vendor, will this allow me to deploy cheaper GPU for my users?

I guess I'm asking about the economic value of the hack.

[Sebb767]

> To maybe ask a better way: will this practically help me train my DNN faster?

Probably not. It will only help you if you previously needed to train it on a CPU because you were in a VM, but this seems unlikely. It will not speed up your existing GPU in any way compared to simply using it bare-metal right now.

> Or if I'm a cloud vendor, will this allow me to deploy cheaper GPU for my users?

Yes. This ports a feature from the XXXX$-range of GPUs to the XXX$-range of GPUs. Since the performance of those is similar or nearly similar, you can save a lot of money this way. It will also make the entry costs to the market lower (i.e. now a hypervisor could be sub-1k$, if you go for cheap parts).

On the other hand, a business selling GPU time to customer will probably not want to rely on a hack (especially since there's a good chance it's violating NVidias license), so unless you're building your on HW, your bill will probably not drop. But if you're an ML startup or a hobbyist, you can now cheap out on/actually afford this kind of setup.

[lovedswain]

Running certain ML models in VMs

Running CUDA in VMs

Running transcoders in VMs

Running <anything that needs a GPU> in VMs

[ur-whale]

This is the exact same information you posted above.

Please see my edit.