On Demand Webinar: Limitless Capabilities of NVIDIA CloudXR 2.0

Learn how NVIDIA CloudXR can be used to deliver limitless virtual and augmented reality over networks (including 5G) to low cost, low-powered headsets and devices

The recent webinar shares how NVIDIA CloudXR can be used to deliver limitless virtual and augmented reality over networks (including 5G) to low cost, low-powered headsets and devices — while maintaining the high-quality experience traditionally reserved for high-end headsets that are plugged into high-performance computers.

At the end of this webinar, we hosted a Q&A session with our guest speakers from The Grid Factory, co-founder and CTO Ben Jones and Applications Specialist Tom Murray. The Grid Factory is a UK-based immersive technology integrator.

Below are the answers to the top questions asked during the webinar.

Q: Does CloudXR stream the data at the native resolution and refresh rate of the headset? If so, how well does it handle high refresh rates?

The Grid Factory: Yes, CloudXR delivers data at the resolution of the HMD. CloudXR has been tested on the Valve Index at 144Hz, and as long as the environment is designed appropriately, the experience is as good as a local one.

Q: What is the size of the GPU cluster recommended for CloudXR to support multiple users?

The Grid Factory: The size of the cluster will depend on how many CloudXR sessions a single GPU can support. For virtual reality, if using the RTX 6000, RTX 8000 or NVIDIA A40, then up to two concurrent CloudXR sessions may be supported, depending on the application requirements.

If using augmented reality, then this number may increase due to the change in application requirements and client optics. To be able to support multiple users on the same GPU, the environment must be running NVIDIA virtual GPU software, which allows the virtualization of GPU resources.

Q: Are there any CAD packages supported (Creo, SOLIDWORKS, NX, Inventor etc)?

The Grid Factory: We have tested CloudXR with Autodesk, Siemens and SOLIDWORKS products with great success. CloudXR does work with these products and other OpenVR design applications.

Q: How would the CloudXR solution paired with Oculus Quest 2 compare to a tethered VR experience (such as a Valve Index) in terms of latency?

The Grid Factory: If the Wi-Fi 6 network is performing correctly, there should be no discernible difference other than those specified by the manufacturers in terms of hardware specifications. However, due to the Quest 2 now being able to access the same hardware resources (such as the CPU and GPU) as the Valve Index, the performance differences should be minimal, with the Quest 2 having the benefit of better mobility over the Valve Index or any other tethered HMD.

Q: Do you need a 5G-capable device to use the VR?

The Grid Factory: No, currently there aren’t any 5G-capable XR headsets on general sale (except perhaps the NReal, but this capability is only available when connected to a 5G phone; that is a separate purchase to the headset, and is only available in South Korea and Japan).

CloudXR will work across Wi-Fi 5 networks, but there are performance considerations. The minimum bandwidth required for a CloudXR experience in a headset is 50mbp/s, but we would suggest that a larger bandwidth than this is necessary, and this doesn’t take into consideration other devices that may be accessing the internet through the Wi-Fi network.

Q: What computing power are we talking about on the AWS side?

The Grid Factory: The P instances (V100), and the G instances (T4) have been tested with CloudXR.

Q: Does the client need a GPU on the client side?

The Grid Factory: If you have a tethered headset such as a Vive, then yes, although it doesn’t need to be very powerful as it needs to decode the stream from the CloudXR server.

If you are in a mobile headset such as Quest 2 or Pico Neo 2, then you do not need a GPU. The point of CloudXR is that all the GPU rendering and encoding is done on the server side, rather than the client side. This means that the experience is standard across a wide variety of devices. And because of this, the battery life of all-in-one or all-in-two headsets is improved when users are in the application, as the grunt work is being done outside the headset. 

To learn more, visit the CloudXR page where there are plenty videos, blog posts, webinars, and more to help you get started. 

And don’t miss out on the latest AR and VR news at the GPU Technology Conference (GTC), which starts April 12, 2021. Register now for free access and check out the keynote by NVIDIA CEO and founder Jensen Huang, and other XR sessions available at GTC.

Leave a Reply

Your email address will not be published. Required fields are marked *