|
|
John Carmack talks ray-tracing, PC gaming
Author: Tim SmalleyPublished: 14th March 2008
Carmack believes CrossFire and SLI are good for the industry, unlike Sweeney.
John Carmack, one of the most respected games developers in the industry, has spoken out about the recent debate surrounding PC games development and also talked about the possibility of ray-tracing in games in a recent interview with PC Perspective.
"From a developer stand point the uncomfortable truth is that the console capabilities really dominate the development decisions today. If you look at current titles and how they’ve done on the console, you know, high end action GPU based things, the consoles are so the dominate factor that it’s difficult to set things up so that you can do much to leverage the really extreme high end desktop settings," said Carmack.
He also shared his thoughts on the move to multi-GPU graphics subsystems – and his opinions differed greatly from Tim Sweeney's. "I've always been a big proponent of these high end boutique systems – way back from the early days of 3dfx I always thought it was a real feather in their cap early on that they could pay more money and have a bigger system and have it double up and just go faster," Carmack said.
"I think it’s a really good option and certainly companies like Nvidia and AMD are throwing all the resources they possible can at making the newer, next-generation cards. But to be able to have this ability to just pay more money and get more performance out of a current generation is really useful thing to have," he added.
Carmack also talked about ray-tracing during the interview, offering some interesting insights into what might happen when Intel enters the market with its many-core Larrabee architecture – one thing that Intel has been pushing has been ray-tracing.
"I’m not really bullish on [ray-tracing in a classical sense] taking over for primary rendering tasks which is essentially what Intel is pushing," said Carmack. "There are large advantages to rasterization from a performance standpoint and many of the things that they argue as far as using efficient culling technologies to be able to avoid referencing a lot of geometry, those are really bogus arguments because you could do similar things with occlusion queries and conditional renders with rasterization. Head to head rasterization is just a vastly more efficient use of whatever transistors you have available.
"But, I do think that there is a very strong possibility as we move towards next generation technologies for a ray tracing architecture that uses a specific data structure; rather than just taking triangles like everybody uses and tracing rays against them and being really, really expensive [from a performance perspective]," Carmack continued.
Carmack said he's been pitching his ideas to the industry's key hardware vendors and he believes that Intel's push towards a conventional ray tracer "is unlikely to win out."
Finally, he also talked a bit about Nvidia's acquisition of Ageia. Carmack has been very public on his feelings about hardware-accelerated physics—specifically referring to Ageia—and it seems that his opinion hasn't changed much. "That was one of those things where it was a stupid plan from the start and I really hope Nvidia didn't pay too much because I found the whole thing disingenuous. Many people from the very beginning said their entire business strategy was to be acquired because it should have been obvious to everybody that the market for an add-in physics card was just not there."
In the rest of the interview, there's a lot more insightful commentary on where the games development and hardware industry is going – it really is well worth the read, even if some of it does get quite deep.
You can of course discuss his thoughts on the industry's direction in the forums.
Related
Comments (27)
Compare over 250 mobile phones &
52,000 deals!
Compare over 100 broadband & mobile broadband deals online!
- Site Links:
- About
- Email Editor
- Newsletter
- Privacy
- RSS
- Score Guide
- Our Other Websites:
- Auto Express
- Channel Pro
- Computer Buyer
- Computer Shopper
- Den of Geek
All content © bit-tech.net 2000-2009
© Copyright Dennis Publishing Limited licensed by Felden
But then again, I am a Doom fanboy.
Edit: I just read the original article. Is it just me or does Carmack's brain seem to be permenantly about 3 years ahead of everyone elses?
Who isn't?! :? :)
And now we wait for his next tech demo with bated breath.
Heh, he's one of the pioneers of the industry as far as games development goes - he's designing the technology of tomorrow, today. If you ever had chance to speak to any of the hardware designers, you'd get the same - it doesn't matter who because they all have their vision of the future. Some visions are interesting, some are completely whacko and others are closer to reality. :)
Its a shame his recent games havent been 3 years ahead of everyone else. In many people's eyes, ID's last few games have been major let downs (Doom3 and Quake4 in particular) Both were not lacking in graphics, but instead gameplay.
So whilst ID and John C have the best technological minds of their field.. it doesnt appear to be translating to games with innovative and new gameplay.
To cite back to the days of Doom 1 is a mistake. They may have been original then.. but not now.
Saying all that, I find his comments very interesting and can see a time when 8 and 16 core CPU's will actually be used for graphics, physics, and game play. Whether thats Ray Tracing or otherwise.
Well, of the new Doom3 engine -- I'm not.
I really didn't find it fun, running around in the dark with a torch on my machine gun. Got pretty boring. Walk in the dark, boo! Zombie jumps out. Walk in the dark, nothing there, walk 4 feet, boo! Imp suddenly behind you, repeat. :(
I did love the original Doom, if that helps :D
That could be one of the reasons why you didn't enjoy the game, there's no torch on the machine gun, so you must have cheated by using one of the duct tape mods. That would do a good job of ruining the atmosphere of the game.
I still think Doom 3 is one of the best games out there and I go back to it time and time again.
it didnt have any spectacularly ground breaking gameplay elements but it was a bloody good FPS
Q4 machine gun is one of my favourite FPS weapons i think, so very satisfying :D
Did I ? I thought there was a torch you hooked on the gun? Oh well, if it took cheating to actually "see" any of the baddies in the game then to me that's a bit poor in the design department. You'd've thought that with such a new, space-aged engine they'd want to show it off, not pull the curtains and turn off the lights :/
Anyhow, if the atmos was "run around in the dark" then great... that was achieved.. Thing is, I could do that in my flat and have the same experience.. :)
Wolfenstein was far more fun. At least I could see things...
It only scared me in that it had loads of "jumpy" bits, as in, turn the corner and briiiiiing! <sound of scarey effect when something jumps out>, something jumps out at me which to me isn't exactly the same.
Now F.E.A.R., with creepygirl crawling around under desks and in wierd Exorcist gymnastic positions and strobe effects, that sh1t freaks me out and I did get proper spooked in that. Few skidded undies in that one, I don't mind admitting!
Maybe you monitor has a horrible contrast ratio. I could see just fine.
Once I got to the stealth ninja's though I was happy :D
Yeah they were cool. Was difficult not to hit the
cheat buttonslow-mo on those dudes.You're right about other developers really making use of id's engines in a way they've struggled to do since Q3 really, but I'm afraid your mistaken about Duke 3D, the Build engine wasn't based on anything id had done. It is true how much of an influence the various Quake engines have had on the gaming world. Even the Source engine still has bits of the original HL engine in, which itself was a highly modified version of the Quake 1 engine (not the Q2 engine as many people believe).
The problem with the D3 engine is that it's way ahead of its time. It was the first engine to utilse 100% per-pixel lighting, which was really impressive, but it meant there was zero ambient light in the game, which not only is unrealistic, but is also a recepie for a very dark game. IIRC, even UE3 needs the old fashioned light-maps as well as per-pixel lighting to achieve ambient light, so in that respect it's actually taking a step backwards from the D3 engine. Personally I'm looking forward to the unique texturing in id tech 5, and having an engine that isn't as dark as D3, even if it did suit that particular game perfectly (which wasn't a bad game, really, scared the shite out of me on many occasions)
Edit: Funky diagrame of the Id tech family tree:
http://upload.wikimedia.org/wikipedia/commons/thumb/6/63/Quake_-_family_tree.svg/691px-Quake_-_family_tree.svg.png
However they did sort the engine issues with Q4 which was actually a good laugh and opened the engine up for more wide open enviroments.
the only thing i hate in carmacks engines is the "shiney" effect all the way thru on everything.
ETQW uses a massively updated D3 engine, they put HUGE amounts of effort into reintroducing light-maps in a meaningful way to enable decent outdoor lighting. The vanilla D3 engine's attempts at light-maps which, whilst they do exist, are pitiful. The problem is that every pixel in the game has a rather complicated set of lighting processes to work through when rendering, exponentially factorerd up when multiple lights hit the same pixel. Having more than about 3 light sources hit any one pixel completely kills the frame rate. Also, in reality, a single light source bounces light off surfaces, so any single point would be hit numerous times by a single light source. That's why if you shine a light at a wall, the rest of the room is lit up to an extent. D3 can't do that, if you shone a really bright light directly at a wall, the rest of the room would stay pitch black, as it simply lacks the horse power to calculate all the bounces. Light-maps can, as this is all worked out before hand and 'baked' onto the map, but it makes it static.
Ray-tracing would solve this because the rendering is not worked out from the light source and then bounced around the room, but instead from the viewer themselves, so every pixel is only rendered once (as you only see it once per frame), as I understand it.
the imps spawned to slow to make them the least scary not to metion i could see there portals and get my shotgun ready for them as they "spawned" in. and thats even on the harder difficulitys.