Nvidia CloudXR promises to beam high-end extended reality visuals directly into your headset from across the internet. This sounds amazing on paper, but is this the future of VR or just a niche technology that most of us can never benefit from?
While CloudXR may mainly be of interest to developers and those concerned with the technology behind XR applications, eventually, end users will have to take notice.
What is CloudXR?
CloudXR isn’t one particular thing but a platform that ties together several different technologies and services towards delivering a specific form of content remotely. This means there’s a hardware aspect to it, a software side and a policy and structure.
Nvidia wants to make it easy for VR content developers to leverage CloudXR, so they provide an API (Application Programming Interface) that makes it relatively simple for those writing VR software to leverage the power of CloudXR. Then there’s the network technology that delivers the content to the headset and, of course, Nvidia’s GPU technology.
CloudXR also relies on some very clever virtualization technology, which lets VR applications running in the cloud make maximum use of the hardware resources that are available.
Nvidia claims that CloudXR can deliver top-class XR content with no perceptible latency, so the person undergoing the experience wouldn’t even know that all the processing is happening somewhere far away in a
What Problem Does CloudXR Actually Solve?
Why bother making a platform like CloudXR? Isn’t it just easier to use a local computer or use a standalone VR system such as Oculus Quest? There are a couple of reasons why these solutions might not be ideal.
First of all, high-end computers that can offer the best VR experiences are expensive. Secondly, you need to be tethered to them. That can be via a cable, which is quite restrictive, but even wireless tethering to a powerful computer has range limits.
Those range limits might not be too much of an issue for VR but augmented reality applications won’t find much use when they can only be used within a few feet of a computer.
Standalone headsets like the Quest 2 are amazing feats of technology themselves. They represent the pinnacle of mobile CPU and GPU power, but even then, there are serious compromises when it comes to the complexity of simulations or the quality of visuals.
CloudXR has the potential to solve many of the shortcomings each of the VR and AR solutions have. In terms of cost, you no longer have to worry about hardware or maintenance. Range becomes irrelevant since anywhere with the right network infrastructure is an eligible zone.
Effectively, you now have access to more computing power than any one local computer can currently provide, and you only have to pay for the portion of that power you use for as long as you use it. That means XR experiences and applications that may not even be possible with current computing power.
The Technologies That Enable CloudXR
CloudXR is mainly possible thanks to big advances in cloud processing power and the advent of 5G networks. 5G is still very much in the early stages of its development, but when properly implemented, it offers incredible bandwidth and very low latency.
Both of these are key for XR technologies, but the second is perhaps the most important. Even directly connected by a high-bandwidth cable such as HDMI or USB 3, latency has been a problem in VR from day one. With even just a little too much of if, the immersion and comfort of VR can be shattered.
5G can, in theory, move data around with so little latency that it should rival a local direct connection. Secondly, NVidia’s RTX GPU technology incorporated hardware ray tracking and machine learning acceleration silicon.
Having enterprise-grade versions of RTX on tap opens up the sort of next-generation VR that simply hasn’t been seen yet.
The Challenges of CloudXR
This all sounds very impressive, and I’m sure Nvidia has had plenty of success during testing, but it’s hard not to be skeptical about this approach in practice.
Latency is a big worry, and VR, in particular, is incredibly latency-sensitive. It’s important to keep in mind that the latency here is bidirectional.
Sensor and input data from the headset has to be sent to the cloud, processed and then the sound and graphics that the user sees are sent back. That round trip has to happen in milliseconds.
If you’re producing 90 frames a second VR, that means a complete round trip of one 90th of a second for each frame to be produced.
VR hardware developers struggled for years to get wireless VR to work from one side of a room to the next, and now Nvidia wants to do it over hundreds or even thousands of miles. Given how unpredictable network performance is in the real world, I worry that CloudXR will have the same adoption problems as game streaming services such as Google Stadia.
A Promising Future
If, for the sake of argument, CloudXR technology works as advertised in the field and eventually trickles down from its current big-business target audience, it really could revolutionize XR as we use it from day to day.
If your XR headset doesn’t need to have any local processing power, other than to handle data and decode video and audio, it could be very slim indeed.
Imagine everyone within a company’s building, wearing slim and light AR headsets driven by a central cloud data center. It could also mean integrating all their data together and making them part of the same shared virtual environment.
While it’s not clear if CloudXR specifically is going to be the foundation of XR applications in the future, the technology Nvidia is pioneering here will undoubtedly be a part of it.