To handle VR graphics, gaming PCs have to be 7 times more powerful

The folks at Nvidia, the world’s largest maker of stand-alone graphics chips, are good at numbers. So I talked to them recently about their expectations for virtual reality gaming. They showed me some demos of the Oculus Rift and HTC Vive virtual reality headsets, powered by a PC using Nvidia’s GeForce GTX Titan X graphics chip. The demos were just a taste of the many VR experiences we’ll see at the 2016 International CES, the big tech trade show in Las Vegas next week.

And they said that it takes graphics processors that are about seven times more powerful to run VR, compared to a standard PC game. By next year, when the first major PC-based VR headsets ship, there will be about 13 million PCs in the market that will be powerful enough to run VR — in the right way. Nvidia says it can extend that number to 25 million if the VR game makers use Nvidia’s Gameworks VR software, which makes the VR processing more efficient.

That shows that the market for VR won’t be gigantic in 2016, at least on the PC. Sony is also launching PlayStation VR on the PS4 video game console. Sony’s installed base of 30 million PS4 units means that its market potential will be bigger than the PC’s. But Nvidia notes that VR on the PC will be more demanding in terms of graphics processing required.

We recently interviewed Jason Paul, general manager of Nvidia’s Shield, gaming, and VR business. Paul said Nvidia is in contact with more than 600 companies that are working on VR projects. That’s a huge VR ecosystem, and Nvidia wants to be one of the leaders that makes it all happen.

Here’s an edited transcript of our conversation.

GamesBeat: We saw that there were 234 companies that one of our services was tracking.

Jason Paul: We have more than 600 companies in our tracker now. There’s probably one or two thousand at this point, globally.

GamesBeat: Is that projects or actual startups?

Paul: Startups and companies. It includes Fortune 500 companies that are doing VR as well. But there’s a massive amount of interest.

GamesBeat: I’m starting to get a handle on the larger picture. Seattle, the bay area, and Los Angeles seem to have the concentrations. New York as well. But it’s happening everywhere.

Paul: A lot of European-based companies. We’re seeing a bit of activity in China as well, a number of headset guys building headsets.

GamesBeat: I’m waiting to sort it all out before I take meetings with everyone. But we’ll be seeing a lot of them at CES. Virtuix was apparently ready a year ago.

Paul: I have one on order. Still waiting. But apparently they’re shipping their first units to the Kickstarter.

GamesBeat: They were building them on the assumption that the headsets would be out by now.

Paul: A few people are probably waiting on Oculus and HTC, yeah. One or two. But it’s getting close. Our perspective on VR is that it’s a new computing paradigm, like PC or mobile cloud. There will be a lot of energy around games, but it’s going to be much bigger than games. The apps and experiences that come out are going to change the way we do business, the way we interact socially with people, how we enjoy and consume entertainment.

Things like Tilt Brush, Everest VR as an educational and experiential tool, some of the news and storytelling projects, like gaining the perspective of Syrian refugees. We’re looking at VR with gaming at the core of it, but also in our professional spaces as well, other social and consumer apps.

GamesBeat: I feel like if someone gets one of these strong art tools out there, it could be great.

Paul: Have you see the Oculus art app, Medium? They have one with their touch controllers. It’ll be really interesting to see people go crazy with sculpture and new forms of art.

We see the potential of VR as very large, but we also see a big challenge as far as the computing power that’s required. If you look at your typical PC gaming experience, 90 percent of the gamers out there play at 1080p. For a smooth experience you don’t want to go below 30fps. Compare that to VR where the displays are about 2K, but you have to render closer to 3K, and you don’t want to go below 90fps. It’s about a sevenfold increase in raw performance to render for VR versus traditional PC gaming. You have to do that in less than 20 milliseconds from head rotation to what shows up on your display.

GamesBeat: What is Nvidia doing in VR?

Paul: Nvidia is doing three core things to solve this problem. First, we’re building fast GPUs, and we’re building them specifically architected for VR. Our Maxwell architecture has specific capabilities and features that make it very fast for VR. We have some technology that increases performance by up to 50 percent for VR applications.

Software-wise we’re making sure the out-of-box experience for customers is perfect. We want the first VR experience everyone gets when the headsets come out to be a good one – no stutter, no lag. Our GeForce Experience software and our Game Ready drivers are core to making sure that experience is delivered.

We’re working with all of the VR ecosystem through an SDK we call Gameworks VR on the consumer side and Designworks VR on the professional side. That helps headset manufacturers get lower latency and plug-and-play compatibility, and it helps developers get better performance out of their apps.

We’re engaged in this across all of our businesses. GeForce for consumer and gaming PCs, Quadra for professional visualization — theme parks, entertainment centers, location-based installations of VR — and Tegra on the mobile side. On the mobile side there’s a couple of headsets. A company called Game Face Labs down in L.A. has integrated a Tegra processor into its VR headset. Another company in Silicon Valley called Atheer Labs is building an AR headset with Tegra in a hip pack.

Our GPUs are being used on the graphics side to power the headsets, but they’re also being used on the capture side to stitch 360 video. Companies like Video Stitch are building GPU-accelerated stitching platforms to take images from three to 24 cameras and piece it together into 360 panoramas using GPUs. And there’s a bunch of activity going on in tracking and machine learning around GPUs.