Tim Sweeny DICE 2012 Talk

[Mr Bean]

Video link

The Truth

There are few people in the gaming industry that you simply must pay attention to when they speak. One of them is John Carmack, founder of id Software and a friend of the site, creator of Doom. Another is Epic Games' Tim Sweeney, another pioneer in the field of computer graphics that brought us the magic of Unreal before bringing the rest of the gaming industry the Unreal Engine.

At DICE 2012, a trade show for game developers to demo their wares and learn from each other, Sweeney gave a talk on the future of computing hardware and its future. (You can see the source of my information and slides here at Gamespot.) Many pundits, media and even developers have brought up the idea that the next console generation that we know is coming will be the last - we will have reached the point in our computing capacity that gamers and designers will be comfortable with the quality and realism provided. Forever.

Read the rest of the article here
Surfing the web came to this article of commentary on Tim Sweeny's DICE talk and the talk itself. I have to say I found both fascinating because it goes mathematically into what we will need in GPU power to achieve photo-realism. Not as in Holodeck, but as in the ability to render a scene on a monitor at a speed at which details would be indistinguishable from real objects and fast and detailed enough our eyes could not detect any lag or shifting.

TL:DR Tm Sweeney is a terrible public speaker who spends half an hour explaining that we need roughly 2000% of today's CPU/GPU power to achieve that goal. Also he does maths.

[Starglider]

TL:DR Tm Sweeney is a terrible public speaker who spends half an hour explaining that we need roughly 2000% of today's CPU/GPU power to achieve that goal. Also he does maths.

This is obviously wrong as it should be apparent even to non-IT people that movie render farms use much more than 20 times as much computing power per frame than a gaming PC. What Time Sweeney actually said was that 2000 times current single GPU compute power is needed for complete photorealism. I would agree on the order of magnitude, but this basically assumes conventional triangle mesh or spline geometry, rendering approaches for high detail deformable materials and realistic fluids and vapors can add a lot of compute usage, then of course there's the rapidly growing physics compute requirements of realistic environments. Though I'd note that if we could track eye position accurately enough to only render at full-res for the fovea we'd need a lot less brute force for the render stage.

[CaptHawkeye]

GDC used to be better than even E3 for showing relevant happenings in the game industry but it's really not doing that anymore. The one game I was happy to see gameplay for was the MMOFPS Planetside 2 which looks pretty fun. Other than that it was full of the usual vague promises, pointless tech demos, and keynote conferences delivered by industry "personalities" who haven't actually made a game since the 90s.

EDIT:

This is obviously wrong as it should be apparent even to non-IT people that movie render farms use much more than 20 times as much computing power per frame than a gaming PC. What Time Sweeney actually said was that 2000 times current single GPU compute power is needed for complete photorealism. I would agree on the order of magnitude, but this basically assumes conventional triangle mesh or spline geometry, rendering approaches for high detail deformable materials and realistic fluids and vapors can add a lot of compute usage, then of course there's the rapidly growing physics compute requirements of realistic environments. Though I'd note that if we could track eye position accurately enough to only render at full-res for the fovea we'd need a lot less brute force for the render stage.

The problem is that these guys seem to be disconnected from the market they're serving. They're overwhelmed with technology and progress but can't seem the understand that the progress needs to be disseminated to consumers or developers or it serves no purpose.

That and like you say the industry is still obsessed with achieving performance through raw brute force and their comes a point of negative returns on that. I know it's still tempting to take that route because it's proven and cheap, but it also takes the game industry nowhere fast.

[Tolya]

I am actually amazed at the fact that anyone listens to anyone trying to predict future. C'mon, Carmack and Sweeney aren't seers. They made good software back in the day. They cannot and do not predict how the future is going to pan out.

This is one of those "nah, who will want a passenger airplane since you can cross Atlantic so comfortably on a ship?" moments.

[Mr Bean]

I am actually amazed at the fact that anyone listens to anyone trying to predict future. C'mon, Carmack and Sweeney aren't seers. They made good software back in the day. They cannot and do not predict how the future is going to pan out.

This is one of those "nah, who will want a passenger airplane since you can cross Atlantic so comfortably on a ship?" moments.

Did you read or listen to any part of his talk? He gave a thirty minute speech where he used math and the scientific method to guess when we hit the ceiling on graphics. He's not predicting the future he's calculating and while calculating he's not predicting because he does not claim to know several things like if we will hit any unknown roadblocks.