APRIL 23, 2008 • What’s next for the PC as a gaming platform? That question is on the mind of many executives, especially when a publisher the size of Electronic Arts decides to cancel the PC version of Madden NFL 2009 – EA Sports primary franchise. So DFC turned to industry veteran Roy Taylor for his perspective.
Taylor has been a long-time advocate of the PC as the “fourth console system” and he is a founding board member of the recently formed PC Gaming Alliance (PCGA), a non-profit organization whose mission is to drive worldwide growth of PC gaming. Founding members of the PC Gaming Alliance include Acer, Activision, AMD, Dell, Epic, Intel, Microsoft and of course Nvidia. Taylor joined Nvidia in 2000 as the Vice President of Sales for Europe, Middle East, Africa and India. Since 2005, he has been the Vice President of Content Relations for Nvidia where he is responsible for the development and support of third party PC software as well as Nvidia support tools and publications.
DFC: What are some of the trends you are seeing in the market right now?
Roy: The easiest place to start is hardware and how that corresponds to consumers because the two are closely linked. In terms of hardware, let’s talk about the PC verses consoles and next generation consoles.
If you look at the current hardware, Xbox 360 and PlayStation 3, it is pretty good hardware, especially compared to anything that was in the past. The hardware makes the most sense if you have HDTV. So if you have a 720 1080p screen you are going to have a pretty good, graphically speaking gaming experience.
With HDTV you will have a good gaming experience. That might surprise readers to hear a PC guy saying that, but that is the fact. I think for a lot of gamers Gears of War was a great graphical experience on the console. And whilst Crysis was undeniably better graphically, Gears of War probably reflects graphics at a level that is ok for most people. A lot of people might aspire to the graphics of Cyrsis but feel it is out of reach right now.
So where am I going with this? Am I saying that graphics is done? Well, no it is not, but it is going to change. And this is where the hardware story gets really interesting.
Today’s graphics, in many instances, is really essentially nothing more than really pretty wall paper. So what I mean by that is that it does not profoundly affect the game play. And it doesn’t profoundly affect the game play because of scaling. There are some really interesting things taking place right now that will be changing that.
If you take a traditional game publisher’s view of PC gaming, you would absolutely limit your developer from doing anything that wouldn’t run on as low-end a spec as possible. Often times that means horrible Intel integrated graphics. The problem with that is that if you are trying to make a game that can scale back that far then you limit how far up you can take it.
I have this phrase that scalability is the enemy of experience and it is absolutely true because you strangle your maximum experience by having to scale back to your minimum experience.
The second thing is that, by trying to scale, you inherently inhibit the quality of game play graphics. If at the high end of the experience I can draw a bush that you can hide behind in a game, from a multiplayer perspective, that bush does not exist in the lower end of the experience and I have effectively broken the game play. I can’t have you hide behind a non-existent bush, I just can’t. Therefore, I’m not going to put the bush in. So now we have two bad effects and Intel integrated has a lot to answer for. The first negative is I’ve strangled my maximum experience by having to cater to the minimum experience. That says that scalability is not inherently elastic. Secondly, I limit what I can do in terms of graphics with game play by this tactic as shown by the “you can’t hide behind a bush that doesn’t exist” analogy.
So now how is hardware changing and what does this have to do with consoles? These issues are linked because from a publishers perspective today his number one focus is no longer the PC and PC scaling because he no longer develops games for the PC. Because of that he is fundamentally designing for the consoles, and because of that a good thing is happening: the minimum spec that is the fixed spec of the console is now dramatically better than the minimum spec for the Intel integrated PC. Now developers are building their game they are asking for the best possible console game. I’m not even talking about the best PC game, I’m just talking about something that beats Gears of War on PS3 or Xbox 360. From a developer’s point of view he is now focused on that mission. And the spec is fixed at a relatively good point.
Nvidia’s has been making an effort to explain that our installed base of GeForce users is about 189 million users, and of those, 69 million are GeForce 6600 or better.
That is a pretty good installed base to sell into. At the same time the minimum spec of an Xbox 360 or PS3 is a little bit ahead of a 6600. The result of that has been that the minimum graphics standard of all games is going up and that is a good thing. However, the maximum graphics experience is unfortunately capped by the consoles. So the consoles have done a good thing by bringing up the minimum spec but it is doing a potentially bad thing by capping the maximum spec.
So what does that mean for the developer? Well, he is not going to necessarily spend a great deal more money on art and time on the game to make the high end on the PC as good as it can be. So we have this plus and minus effect of the console on hardware.
So where does that leave Nvidia? Well there are a couple of things that are really interesting. The first is, that whilst the PC is secondary in development to the console, the ability to make money on the PC is increasing. Let me give you another interesting tidbit. The average PC game sells a million copies. The average successful console game can sell anywhere between 3 and 5 million copies. While the gross profit to the developer and publisher are about equal for both, when a PC game can reach the 2 million level, it is dramatically more profitable for the developer and publisher.
DFC: So given all that, how do you go about convincing developers when they start cutting some of their PC development?
Roy: So this is the good bit of the interview. The first thing we have learned is that instead of competing with consoles, everything Nvidia does must support the consoles. Our game development tools include mental images, Shader Creation, FX Composer for in-game effects, NvPerf and AgPerf, which is a brand new tool you might not have heard about from our physics team. I’ll talk a lot about physics and AI in just a moment. All of those tools are being optimized to support games for consoles as well as PC. So we are making all of the code as portable as possible. The important thing about CUDA for gaming is the extent to which CUDA game development is more easily portable. So inherently, porting DirectX 10 to PS3 and Xbox 360 is not terribly complex, but it is not trivial either. The introduction of CUDA in the gaming space helps make porting of code more easy. So that is the first thing.
We are supporting development for PC and console. Now, at the same time we are also working extremely hard to market these additional effects that the PC can support to the GeForce owners who have proven not only to want it, but have a willingness to pay for it. We were very successful in helping make Lost Planet (from Capcom) a success on the PC last year. We also did this for BioShock, and we have in effect become the PC arm of the games world.
So when they say competition is always good, well for PC game development it is not. Developers are so busy working out how to develop for PS3 and Xbox 360 that the third one being PC is just fine. So the last thing these developers want is to have to develop a PC version 1, PC version 2 and PC version 3. Which is why Intel understands that the challenge for Larrabee is forcing publishers to develop to yet another platform, they just don’t want to do it.
So what the publishers want to do is to have the very best video game for PS3, the very best game for Xbox 360 and the very best game for the GeForce PC. Nvidia is helping to make that possible and we are finding it is very welcome.
Graphics are pretty good right now and most people are happy with it, but it does not mean graphics is done. And the reason graphics isn’t done is because of physics and AI. The big trend going on is the move to incorporate more and better physics into the games. And our acquisition of Ageia and PhysX give us the ability to accelerate physics on the GPU and game developers are really excited about it. Not gamers yet, because we haven’t even gone out to them yet.
The success of Havok proves why Nvidia and all of our success and growth to date is nothing compared to what is coming because the number of games that use physics through Havok is enormous. The number of games that use Ageia is enormous. If you can do pretty good physics on a simple CPU with 2 or 4 cores, what on earth are you going to be able to do with an Nvidia GPU, whether it is in a console or a PC, with hundreds of cores.
So where the scaling was bottlenecked previously, now it is unleashed. Because we have the GPU in the PS3, a very good one with the RSX chip, we have 90% market share with the PC, and the race is on for next-generation consoles.
There is only one hardware vendor that GPUs with hundreds of cores right now.
And physics is inherently parallel. The very nature of the way physics works is parallel and mathematically challenged.
DFC: After tracking tools for years, animation tools, middleware tools and now everything seems to be getting so consolidated there doesn’t’ seem to be much left. You’ve acquired companies like Mental Images and Ageia. The list of well established tool companies that Nvidia has acquired is extensive at this point. What is the next bubble of innovation in gaming tools and where is that bubble going to come from? Will it come from inside larger companies, like Nvidia?
Roy: You are absolutely right. The future of gaming is supporting the tools. We have heard that so clearly from the developer community and we have listened. What they want are really good tools and really good support. And we are gearing up for that. People say big companies don’t listen to their customers but we have been listening to the developers, it’s a religion inside Nvidia.
So let’s recap. Instead of fighting consoles we are going to help developers develop for consoles and PC. We believe that the PC will continue to adapt and move on faster than the console. We are already in the PS3 and we are in 90% of gaming PCs, and who knows where the future consoles are coming from. You can make whatever deductions from that you want to.
You can be sure that we are in strong position. So if we have the tools to support and we are now embracing the console support as well as the PC support, we become very helpful and valuable. I have heard it asked, “Well how can you do that if the tools are free? How can they be as good as the ones you pay for?” If it makes them feel better we can charge them. And frankly speaking if it means we can employ a bigger support staff, we’ll do it. Mental Images does still charge, and for Ageia’s PhysX support we still charge for the Xbox 360 support. We do charge for PS3 support but if a developer is working to make the PC experience great we’ll often give them the licensing for free. We also have a great relationship with Epic and Epic has great PhysX in the Unreal 3 Engine and that will continue for future generations. Of course Epic is all about supporting console as well as PC so they are thrilled with us and they are really excited with what we are doing.
The next big wave for PC gaming is physics. So developers are thinking, “I have a parallel math problem and I have this wonderful installed base of massively parallel GPUs that I can run physics on.”
That leads to AI, because physics without AI is inherently restrictive. What I mean by that is, say we are going into a street and we blow up the buildings and create a lot of rubble. Now I can no longer go down the street, it’s blocked from all the rubble I’ve created. You need artificial intelligence to know that you can’t drive over the rubble. Furthermore you need artificial intelligence to know you have to take a second route. You start to see how artificial intelligence is completely linked to physics.
DFC: You are working closely with Natural Motion?
Roy: Yes, we are working with Natural Motion. They in particular create artificial intelligence for bodies. It works in that if a character in the game pushes someone they don’t simply just fall to the ground, instead they will grab something as they fall down, or push back. This is AI for bodies. It turns out because they are so closely linked to physics, AI is parallel as well.
Remember I told you graphics wasn’t dead. Every time you blow something up into 10,000 pieces you have to draw all of the 10,000 pieces, and render them and shade them and shadow them. So graphics as it stands today is certainly not done, but for many people it is good enough. But as we add physics into the game play we bring back an increased demand for graphics. So having a powerful GPU in a machine with hundreds of cores that can accelerate hardware-accelerated graphics and hardware-accelerated physics and the AI, and be able to do all that through an API like DirectX and/or CUDA is really exciting. That is why I’m working 12 to 14 hours a day right now.
DFC: Who are you working with right now? Are you the primary liaison between the developers, tool vendors and Nvidia?
Roy: Yes, for the middleware companies and the developers, along with our developer technology and evangelist teams. And the burgeoning, wonderful, new developers in places like China and India. We are astounded by the talent of some of these guys. The game world today is dominated today by some really wonderful characters like Ken Levine, Tim Sweeney and John Carmack but they are going to have to make some room for some wonderful talent coming out of India, China, Australia and Eastern Europe.
DFC: As far as game genres go how do you see this all playing out?
Roy: We are going to see genre blending, we are already seeing that even in the last two years and we are going to see more of that. With the power that we’ve got we are going to see RPGs, FPS shooters and MMOs blending. I don’t know if you are aware of The Agency, but it is effectively an MMO, FPS. It’s a cartoon-style game where you are effectively in an online world as a secret agent. Its cool, its fun and we are definitely going to see more of that. So when does an RPG stop being an RPG and start becoming an MMO? We are going to see physics in all genres, all of them.
DFC: As far as the bottlenecks for developers, does it come down to having the tools?
Roy: Absolutely, it’s the tools. There is no developer in the world who wouldn’t want to put great physics into his game. We are at an installed base of 70 million GeForce GPU chips that can run physics and AI and the RSX chip inside the PS3, and there is a competitor’s GPU chip inside the Xbox 360, being able to accelerate physics and AI on the GPU is a wonderful thing. But developers need the tools to do it. Just to hand code is a nightmare, so they need to tools to enable those effects quickly, and they also need support.
That is the mantra for the video game industry right now. If I were writing your article I would call it “Support and Tools” because that is really what we need right now.
A good example of one of our tools is PerfHUD which is used by almost all game developers. We just got results back from a survey we did to developers that showed they got speedup on their game code of on average 37% and in some cases up to 400%.
DFC: What is Nvidia’s strategy for tools acquisitions? How are acquisitions made?
Roy: There is a content group and I am a part of this content group. I am one leader in the group and there are several other leaders. We come together after a show such as GDC and discuss what the developers have told us they need. From there we make a decision if it is a tool developers want and then we determine if it is something we should make or buy.
And yes I am out scouting around always to see if there are any interesting tools we should know about.
DFC: So are there any intriguing new tools you’ve seen lately?
Roy: Well, yes, but unfortunately I have a horrible feeling Intel might read your article, so I better not say. On that theme, the only other thing I would like to mention is there are a couple of things that bug me, How come when my character goes in the water they never get wet? Or why does snow never accumulate or shift? These are things that will be addressed in the future.
