Pick a card, any card
Nvidia's Tom Petersen on Steam boxes, Titan and how mobile will catch up to console
"You buy a graphics card today, it's going to be awesome for three years, but there's going to be something better in nine months."
In that one sentence, Nvidia's Tom Petersen has neatly summarised the eternal dilemma of the PC gamer: even if you're completely happy with your rig's performance, you can't quite rest easy if you know it could be better. You'd think that, as director of technical marketing, Petersen would be a prime candidate for the espousal of the bleeding edge of gaming hardware, no matter the cost, but his attitude is much more pragmatic than that.
"I think it's true that people should not expect that if they buy something today, that it's going to be the top for forever. It's probably not going to be the top for forever, but it's going to play games for a long time.
"Technology innovation is a rapid, sort of semi-'eat your own children' kind of market, where you're basically developing something that's going to replace what you've done in the past," Petersen muses. "The idea, though, is that the benefits we're delivering are significant from generation to generation.
The latest generation is Nvidia's 'Maxwell' architecture - a range of cards which is focused on producing more output per watt, reducing noise, heat and battery drain, making them perfect for small form-factor machines, tablets and mobiles. Between Maxwell and the recent Tegra chipsets, Nvidia seems to be making significant advances in portable processing power, so I ask Petersen the divisive question: can we ever expect parity between portables and mains-powered gaming systems?
"It's probably not two years away before a mobile device is giving you the game experience of the equivalent console"
"When a mobile hits a console performance? You can kind of graph it," he replies. "It's hard to predict, but the difference between a mobile device and a console is that a console is plugged in, a mobile device is not. We're going to have to make some pretty specific steps to be able to do on a battery what a console can do plugged in, but I don't think it's impossible. It's just going to be difficult.
"The other way I think about it is, at what point does a mobile platform get good enough and the technology be good enough so that you can't distinguish the experience versus the console? I feel like that is very close. It's probably not two years away before a mobile device is giving you the game experience of the equivalent console. At that point, that's when this whole question gets more real for, I think, console guys.
"We've got mobile coming up from the bottom. Desktop PC's and Mobile PC's are pressuring from the high end. How do consoles remain relevant? It's not clear."
That opinion seems to be reflected in Nvidia's current portfolio - both the PS4 and the Xbox One are powered by GPUs from rival manufacturer AMD. With the new machines establishing their user bases more quickly than most expected, does Petersen feel that Nvidia should have been involved?
"I would look at the market as an option for Nvidia, but it's not a uniquely special market. At the end of the day, we're going to decide what to invest in based on what our expected return is. In the current round, it's very difficult for us to imagine making a reasonable margin as the console provider. This last round, we didn't see that as our most profitable avenue for investment.
"We've been investing in things like desktop GPU's. We've been investing in GRID. We invested in Shield and Tegra. All of those are multi-billion dollar investments. We believe they have a larger probability for a good return for our investors. If you compare that to the console market. The console business is actually pretty difficult.
"I can't predict how console remains competitive in the face of increasing mobile competence. PC prices are continuing to put pressure on that. The whole business model for console is challenged. It's under a lot of stress because of Free to Play and because of micro transactions.
"I feel like consoles have a little bit of adaptation to happen to make sense. Right now, I just scratch my head a little bit, when it seems pretty obvious that they're in the middle. They're in the middle between a massive platform, which is mobile and a massive platform, which is PC.
"I think consoles need to reinvent themselves, just like PC's do and just like mobile does"
"How do you build a console to harvest revenue off of the software in a model where a lot more of that revenue is going directly to the game guys? It's a tough nut. It's a tough nut. It's not impossible, but I think consoles need to reinvent themselves, just like PC's do and just like mobile does. I believe in the last earnings from AMD, they even mentioned that their margin is going to be a little bit depressed because of the contribution from console."
It's a diplomatically phrased way of dismissing the current console generation as interesting but not business-viable, essentially - perhaps exactly the way you'd expect an executive to talk about a market which has just become the exclusive territory of his company's main, if not only, competitor.
In fact, it strikes me that the tide of business has flowed back and forth quite evenly between AMD and Nvidia - that despite Nvidia's bigger market cap, the two co-exist successfully in a market of their own. We chat for a while about what Petersen calls a "stable duopoly".
"I wouldn't call it convivial," he chuckles when I ask him about their relationship. "It's definitely an intense competition. I don't feel like it's personal...I don't think AMD is a bad company. I think that they're actually doing a lot of good work. I think it's just sometimes that they're in a tough position because we're a larger company now and we're dedicating a larger percentage of our revenue towards GPU design.
"It's very, very difficult to be the number two guy. The real risk, I think, is falling into this consistent number two. It's difficult to make the investment level remain competitive there. They really need to be number one once in a while."
AMD's latest innovation is Mantle - an API which reduces the 'translation' time between games and graphics drivers to make them run more smoothly. It's a neat idea which strikes at the core of PC development, the diversity of hardware which is at once the platform's biggest strength and obstacle. It's a notion that clearly intrigues Petersen, and he's openly interested in a lot of Mantle's objectives.
"From what we understand, it's all about creating a less overhead intensive way to access directly the hard drive. Which makes a lot of sense. It's primarily a way to improve the performance and eliminate layers.
"The problem is, of course, that doesn't work very well unless you're delivering the same visual quality as before. As far as we know, Mantle is not designed to change the look of a game. It's supposed to basically make the same look, only run faster. That's what we think is a great idea. Hopefully, we'll see that in more places.
"The big challenge for Mantle is it's hard to benchmark because it's not using any industry standard., but from our perspective, the idea of increased performance by thinning the amount of software layers between the hardware and a game is a great idea. I view Mantle as probably a nice idea in terms of thinning the layers. Whether it's a competitive response to what we're doing or not, I'm not sure. I will also say that the biggest challenge of any new interface or new spec like that, is adoption. My guess is it's going to be somewhat difficult to push for the smaller share vendor to drive a new standard that's not adopted by all GPU's. However, I would say some technology like Mantle, that was used by all GPU's, is probably in everybody's best interest."
"We have big enough business where we can afford to deliver a product for people that are not going to buy a lot of them, but they're going to buy the best gaming device you can have"
Nvidia's own flagship GPU brand is Titan - monstrously powerful cards which are formidable enough on their own, but can also be strapped together to produce frankly ridiculous levels of processing capability. These cards aren't cheap, though, and anyone who can afford six of them in a rig is more likely to be controlling satellites than Lara Croft. The range has always felt more like shop-window tech than mass-market - bragging rights for the extreme gamer. Petersen agrees, with some caveats.
"I don't view Titan as technology that's going to filter down per se," he tells me. "I view it as...it's a love letter to PC gaming. We have big enough business where we can afford to deliver a product for people that are not going to buy a lot of them, but they're going to buy the best gaming device you can have. Since we have a huge business, to me it's more about building the brand.
"We don't really need to make a ton of money on Titan, because it's almost like that's one component of our entire stack. To be truthful, our strategy is not to take Titan technology and bring it down, it's to continue to innovate that ultimate enthusiast, discerning buyer, and to incubate that as sort of a discerning, differentiated design. It's not limited, by cost, really.
"For example, you can have a whisper quiet card that's the highest performance graphics card in the world. That's never been the case before. This is the real test. A couple of years ago, Jen-Hsun (Huang, Nvidia founder) asked us what card to we have in our PC's at home. Nobody told him that we had our highest end graphics card, because at the time, our highest end graphics card was too hot and was too loud. Almost all of us preferred something down the stack. Now, that's not the case. Every single one of us, of the leaders in Nvidia, would want a Titan. There's just nothing about it that's not delightful.
"I would almost say it's more like a leadership statement of this is what PC gaming can do. You can have multiple monitors running at 25 by 14 or 4K in the future and you can get this amazing experience. We're incubating the industry now. There's no ... Intel is not incubating gaming. When we're out there showing Titan, showing multi-monitor gaming, showing all of our gamings technology, it's about helping people understand where's the industry going.
"A product like Titan helps us communicate that."
Another industry shift that Nvidia has jumped into, perhaps less predictably, is streaming. The Shield, launched last year, is primarily an Android-based handheld, using Nvidia's Tegra 4 chip to power a five inch tablet screen to play Android titles. On top of that, though, it also allows PC games to be streamed from a PC running an Nvidia GPU. It's not directly in competition with services like OnLive or PlayStation Now, but there's a clear statement of intent about what Nvidia sees as its remit in the near future.
"I think, and I believe we think, at Nvidia, that streaming of games is an incredibly disruptive technology that's going to make high quality gaming available to very low power handhelds," Petersen says, in the emphatic tones of a man engaging on a favourite subject. "As an example, if you can stream the best graphically enhanced video game to a very low power hand held device, I think that's going to be awesome.
"I think, right now with our game stream technology, we're right at the point where that becomes delightful"
"As long as it delivers the promise of giving you interactive, real-time, stutter free, latency-low gaming. That's what we're trying to do. We're trying to make that experience awesome. We know there's a bunch of technology that's required and we've been working on it for about five years. I think, right now with our game stream technology, we're right at the point where that becomes delightful.
"It's going to take a little bit more, but it's really, really good right now and it's about to become just great. Once that happens, I think it's going to be a change in consumers' minds. Where, before you started streaming movies, you always thought you needed to have a HDDVD or a Blu-Ray to get that quality experience because you couldn't possibly stream that. Well, all that's changed now.
"I don't remember the last time I bought a DVD. You stream movies because it gives you that incredible experience. Once gaming reaches that point, it's going to be the same. You're going to be streaming your games because it's just a better experience than, in many cases, playing it the old way.
"Shield is a particularly good device for streaming games, especially in the house using our game stream technology. I think the Shield device is a final product. It's a great Android gaming platform. It's great for streaming, but we've also learned a tremendous amount. Since we think streaming is the future, this was the great learning device. We figured out, 'Hey, people that buy Shield, almost always are GeForce customers.
"If you're a GeForce customer, you buy a SHIELD because you want to use it for streaming. Maybe that task is slightly different from Android gaming. As we start to understand a little bit better who our customers are, we'll probably be building products that are more tightly tuned towards specific uses."
Speaking of that, and the ever-increasing pressure on consoles from smaller form-factor PCs, our conversation turns to another PC-centric company taking a step out side its apparent comfort zone: Valve. With the first wave of Steam machines now specced up and ready to hit the living room, Nvidia's cards are playing a prominent part - with the Maxwell technology seemingly a perfect fit. Is this a trend we can expect to see Nvidia backing more directly?
"We're definitely not on the sidelines," Petersen allows. "I think the Steam initiative and the Steam Box in general, is a neat idea. I think we should go back and look at the real strength of a PC. That comes from this hyper competition that results in a lower cost and a lot of hardware variation that different people can pick based on their performance preference, right? That's all good. It gets consumers a better platform and a more personalized platform.
"I think the Steam initiative and the Steam Box in general, is a neat idea. I think we should go back and look at the real strength of a PC"
"The negative is it's like, different levels of performance make it hard to content developers to target. We've come out with a technology called GeForce Experience. I don't know if you're familiar with it?"
I am. It's an app which sits in your system tray, if you sport an Nvidia card, and scans your PC for games, allowing you to automatically optimise your machine's performance by tweaking their settings based on the power of your rig. It's the sort of thing that Valve's Steam service has been crying out for.
"That's our answer. We're deploying the power of the Cloud to test thousands and thousands of configurations. As a user, you plug in the best hardware that you want and you say, 'Optimize.' We will automatically set the settings so you get a great experience. It seems to me like, again, since I'm not privy to Steam's strategy, I can guess, that since they're doing the OS and they're still allowing variation on the hardware size, it's not a closed platform from a hardware perspective. My guess is they're going to borrow some of the ideas that we pioneered with GeForce Experience. I don't know that, but if I were Steam, building a console style experience, I would expect for them to try to simplify this a little bit.
"The easiest way to simplify, I think, is through the technologies that we've pioneered with GeForce Experience. If you look at a game like Skyrim, there are, I forgot the right number, but it's like, a million combinations of settings and it's probably more. Users that are not dramatically in to the PC hardware, into graphics, they hardly know what the names mean, let alone what the combinations of settings are. I feel like that, in general, is both a strength and weakness of PC gaming.
"It's a strength, obviously because you can get customized, perfect performance. It's a weakness because it's too complicated. I think Steam Box should take a crack, hopefully, at simplifying that."