How many revolutions should we expect in the next seven years?

TG Daily: I'm hearing more and more from CPU manufacturers that they rely upon their partners in the graphics business to supply graphical performance improvements. If that's the case, is it fair to call it an "Intel PC" or an "AMD PC" anymore, from a gaming perspective, or should we start calling this the "Nvidia PC with the Intel CPU?"

During the next seven years, Nvidia's Derek Perez explained to us, Microsoft is likely to continue to change its DirectX APIs every three years. That's when the revolutions happen; during the three-year period of DirectX 10's reign, Nvidia's product lines will all scale upward every six months, for a total of six revisions per revolution. Meanwhile, game platforms will stay fixed, from a technological evolutionary standpoint, for the next five years; but that relative stagnation will serve as a market force that pushes console prices down.

DP: I'm alright with that. Seems logical. If you think about it, none of the pixels get to the CPU. Literally everything you see on your screen, every pixel, every detail, goes through the graphics processor.

TG Daily: So the whole "multicore" thing may be a little overrated?

DP: No, I don't think so. You still need CPU performance for video playback. You're still going to need the cores. Two cores, plus the GPU, put together, really make a powerful platform.

TG Daily: Along the way, along those six-month stairsteps, I would expect there to be at least one or two revolutions, things that change the formula, not so that you're constantly scaling up, but something that is demonstrably different, that changes the picture.

DP: [There tends to be a new API every three years. In terms of features] DX9 is locked. You can't get too many more new features in a locked API. You've got an API that stays fixed for three years, then you have performance that you continue to push. The API will stay flat, it'll come out with three generations; the performance will take leaps and bounds, depending on tuning architecture, micron process, SLI, quad-SLI...16-SLI, 32-SLI...

TG Daily: There was one game developer I talked to earlier this week who said that, in rendering scenes these days, today's technology is based on rasterization; but as performance improves, we're going to move away from rasterization and more towards ray-tracing, and we'll start to see an era where ray tracing is built into the hardware. Did he know what he was talking about?

DP: It could happen. There are so many different theories about what you could do with hardware in the next few years. It's all based on the API. You start with the API; that's your baseline function. Maybe, maybe not. We'll see.

TG Daily: This is ostensibly a gaming conference, and gaming is perhaps the one industry that pushes the boundaries of what is possible graphically on a PC. But there are other segments...Are all the other market segments just benefiting from the trickle-down theory of what gaming produces?

DP: No. Things like video. Video's really, really big right now. Everybody's watching movies on their PCs...We've invested a lot of R&D in our networking, our southbridge and northbridge, our MCP technology, RAID technology. We look at it as more about an experience, not just about gaming. [And] not so much for gamers, but about enthusiasts.

TG Daily: A couple of years ago at an Nvidia investors' event, it was said that the company was going to be concentrating on three determined tiers in the graphics performance market: an upper-level tier, then a mainstream, followed by a general consumer tier. Since that time, some of your partners and competitors have tried to define the market differently. Microsoft made bold attempt to say, "Maybe there's five tiers here..." There have been some others in the market who have said that there are really just two - your economy buyer and your high-end buyer.

DP: I don't think it matters, to be real honest, whether it's three or ten. People have price points that they want to buy at, people that are upgrading. They know what they want, they know how much they want to spend, and we have a product for them. Whether you categorize it as enthusiast, performance, mainstream, value, entry-level - whatever you call the categories, it truly doesn't matter. At the end of the day for the consumer, it's about having great products at price points that consumers want.

TG Daily: Well, you have to admit that the premium tier that you have now - that the entire graphics market has now - is much higher-priced, even adjusted for inflation, than it has been in the last several years. There's a belief now, more than ever, that there is an upper-tier cream of the crop, the purchaser who wants absolutely everything, even if only for the next six months. Does the feedback you're getting validate that, and is it absolutely true that high performance will always trickle down in a two-year curve?

DP: You could look at companies like Alienware, Falcon, things like Dell's doing with their XPS line, and that right there validates that there is a want and a need for enthusiast-level, high-performance gaming PCs in the marketplace. Are we comfortable with where things are headed? Yeah. Do we like the fact that we're able to bring technology into the high end, and then leverage that technology to bring it to the [right] price point? Yes. And the feedback is great. I may not be in the market for a $699 graphics card, but $299, I know I have a GeForce product that performs great and has a feature set of a high-end GPU, then I'm very happy with it.

TG Daily: Thank you for this interview.