Currently CXR limits max 2 VR Users on one GPU, however ada lovelace architecture will have 3X video encoding performance of the previous generation as following. (NVIDIA’s New Ada Lovelace RTX GPU Arrives for Designers and Creators | NVIDIA Newsroom)
Can we have increased VR users on one GPU with ada lovelace RTX GPU?
Encoding is one potential bottleneck on the number of users that can stream VR (high resolution, high frame rate, stereo) from a single GPU and we expect the RTX Ada Lovelace architecture’s significant increase in encoding capacity and rendering capacity to allow an increase in XR users/GPU. But, there are a number of potential bottlenecks that will impact number of virtual machines and virtual GPUs that can stream XR from any given server. CPU/GPU communications, CPU performance, Rendering capacity, encoding capacity, etc… While we are excited to see the increases in streaming performance Ada Lovelace will offer, which we believe will be significant, those increases will be application specific, device specific (resolution/frame rate), and server specific!
Thanks for your explanation ;-)
As you mentioned, max VR users per one GPU can be varied according to conditions I hope NVIDIA offer detail guidelines with RTX Ada Lovelace architecture not just limit user number.