|
|
Nvidia: Larrabee is a reaction to CUDA
Nvidia responds to Pat Gelsinger’s comments about CUDA being just a ‘footnote’ in computing history
Intel may have put the wind up the graphics business with the development of its Larrabee graphics chip, but Nvidia reckons that Larrabee is just a reaction to what Nvidia has already achieved with its GPGPU CUDA technology. What’s more, the comments from Intel’s Pat Gelsinger earlier this month have also stirred up a debate about the future of multi-core programming.
Nvidia’s general manager of its GPU computing group, Andy Keane, told Custom PC that the high level of interest in CUDA 'is causing Larrabee. Larrabee’s the reaction.’ He then added that ‘these comments from Gelsinger; if we were not making a lot of headway do you think he’d even give us a moment’s notice? No. It’s because he sees a lot of this activity. The strategy is to try to position it [CUDA] as something scary and unique, and it’s really not; it’s something that’s very accessible.’
Gelsinger said that CUDA would end up in the ‘interesting footnotes in the history of computing annals – they had great promise and there were a few applications that were able to take advantage of them.’ He then added that ‘generally an evolutionary compatible computing model, such as we’re proposing with Larrabee, we expect will be the right answer long term.’
However, Nvidia says that Gelsinger’s comments were misleading. ‘We use common languages,’ says Keane, ‘and this is where the Gelsinger information is totally misinformed, because it [CUDA] is standard C. It is actually the open 64 compiler which was originally designed for the Itanium – that’s our compiler. We’re actually using a CPU compiler, but we’ve given it a set of rules that basically say “if you write your program this way, it will scale across a few cores, or hundreds of cores.”’
‘Industry standard languages always live,’ added Keane, ‘that’s kind of the misinformation from Gelsinger. We’re just C, and CUDA’s just a set of rules around C.’ Keane was also keen to point out that Intel was also behind the times when Anwar Ghuloum, a principal engineer with Intel's Microprocessor Technology Lab, said: ‘developers should start thinking about tens, hundreds, and thousands of cores now.’
Reddit | Technorati
the more realistic you make something look, the more obvious the flaws are... like movement and lip sync, proportion and scale, etc, the more annoying that you cant do anything you want due to coding limitations
Comment by NewParadigm at 10:54pm 23rd July 2008
What they said above..................... lol.
Comment by Lightning_Pete at 11:49am 23rd July 2008
The model I'm currently working on takes 2 hours to render one frame on my office quad core Xeon and it still doesn't look good. Trying to make something look realistic is way too demanding on developers. If I could make the city of Cardiff look cartoony I would: instead of ray-tracing, if I used a very basic lighting system with a very unique art style it would look much better than it does now, without much effort and being much faster to run. If you want the lighting to be pin-point accurate then everything else has to be pin-point accurate as well! Super boring to do and it defeats the purpose of expressing yourself. Ray-tracing is a challenge not because of the technique or technology but because of the consequences on the rest of model development.
Comment by Zhaoman at 11:40pm 22nd July 2008
Photo realism isn't just unachievable, actually it probably is, to a certain level, you can use methods like bump mapping to get around the detail requirements on models... but they again have limitations on what can be achieved. I think its more that photo-realism isn't really desirable in a computer games, and thats before we even touch on the subject of psychology/sociology... what would you think of a person who could happily shoot a perfectly real looking person, in a perfectly real looking setting. Lets take an example... how many people play EVE? Would you really want to play that with ray traced scenes? ... think about it before answering...
Comment by NewParadigm at 7:07pm 22nd July 2008
I completely agree that ray tracing is not the way forward for graphics. In my humble opinion games are a very sophisticated form of art. I have been working on a 3D model recently on 3DS Max and to achieve a believable world using ray tracing requires photo realistic models, which just isn't achievable in terms of development time or on the hardware of consumer PCs. A unique art style would achieve a much better result than ray-traced photo realism. Popular games that I think have a very unique art style yet doesn't have photo realistic visuals demonstrate this: Halo series, Metal Gear series, Ninja Gaiden series, Final Fantasy series to list but a few. Many people would consider the graphics of these games to be at the pinnacle of human achievement, but they are nothing compared to the detail required for good looking ray-tracing. These games still have to possess a very different style to get the impact across, and that is the case with all forms of art. Photo realism just isn't achievable and that's why ray-tracing just isn't realistic.
Comment by Zhaoman at 5:06pm 22nd July 2008
Then to quote your earlier title "I hope not"...there is such a massive amount of research going on at the moment into optimizing software for parallel processing...I personally don't think its hype...with the results of that research and the hardware hopefully all coming together shortly there are going to be massive improvements.
Comment by technogiant at 3:28pm 22nd July 2008
Larrabee is going to be a massive exercise in hype, and that its actual performance and impact on the industry is going to be pretty mundane. Dont get me wrong, I hope I't isn't, the industry could do with some shaking up and excitement.
Comment by NewParadigm at 2:47pm 22nd July 2008
I agree there's only so far you can take the graphics side of things especially when other aspects of gaming are weak...perhaps the future will be Nvidia/ATi rasterized graphics leaving larrabee doing vastly improved physics and AI computations...either way it's going to be an exciting couple of years.
Comment by technogiant at 2:11pm 22nd July 2008
Once larabee gains market penetration as it will because it will provide good rasterized graphics and accelerate so many other everyday tasks then developers will start to make ray traced games which is what larrabee was designed for...then it will overtake rasterized graphics...my prediction...in 5 - 10 years all games will be ray traced on some x86 larrabee type system
Comment by technogiant at 12:26pm 22nd July 2008
rather see effort being put in to making other areas of games more impressive and immersive, things such as physics, AI, dialogue, mood/atmosphere, sound. I think too much focus has been put on the graphical element of games to the detriment of gameplay.
Comment by NewParadigm at 1:08pm 22nd July 2008
having played with ray tracing a fair bit doing architectural renderings I honestly have to say that I hope ray tracing does not become a major game tool. In most cases you have to fiddle and cheat to get a decent dramatic lighting effect anyway. If ray tracing becomes standard I think we will be pretty dissapointed with how mundane and rubbish games will start to look. Its the same thing for painting miniature figures... real world lighting solutions just dont look impressive. Also unless you go in top the utmost detail with your medels, and then apply a very processor intensive, high quality lighting solution you get pretty flat pictures. For the amount of processing it takes I just dont believe ray tracing is worth it.
Comment by NewParadigm at 1:08pm 22nd July 2008
Once larabee gains market penetration as it will because it will provide good rasterized graphics and accelerate so many other everyday tasks then developers will start to make ray traced games which is what larrabee was designed for...then it will overtake rasterized graphics...my prediction...in 5 - 10 years all games will be ray traced on some x86 larrabee type system
Comment by technogiant at 12:26pm 22nd July 2008
LoL...just love the toilet humor...I don't think that initially Larrabee is going to be able to compete with Nvidia/ATi in the rasterized graphics department...It will be a mid range graphics card and only does rasterization because it has to.
Comment by technogiant at 12:25pm 22nd July 2008
LoL...just love the toilet humor...I don't think that initially Larrabee is going to be able to compete with Nvidia/ATi in the rasterized graphics department...It will be a mid range graphics card and only does rasterization because it has to.
Comment by technogiant at 12:25pm 22nd July 2008
how often do you need to do that crap compared to the number of times you just need to pee?
Comment by NewParadigm at 11:39am 22nd July 2008
Using the same toilet analogy...you can do more with a full cubicle...Have you ever tried crapping in a urinal?
Comment by technogiant at 10:46am 22nd July 2008
is that a stream processor is a urinal and an x86 processor is a full cubicle with a Krapper style unit installed, in the same area there can be 2-3x as many urinals as full toilets. Now consider why the queue for the ladies is always longer than the queue for the gents!
Comment by Initialised at 2:08am 22nd July 2008
Too busy re-packaging failed gamecube games as Wii games! They were c**p then, they're c**p now. Shame the lowest common denominator is such a cash cow, it means the envelope doesn't get pushed as far as it could.
Comment by gavomatic57 at 7:00pm 21st July 2008
thinking aboot multiple cores, the problem is with the clarity of what is being said, games developers can't develop for hundreds of cores for CPU/GPGPU because then the GPU/shader cores wouldn't be able to be fully utilised for graphics. But developers of 3D software, video editing programs etc can and already are because while a program renders a 3D frame or a video sequence, generally the GPU/shaders sit redundant. And if you look at nvidia's website they already have things like Gelato, PowerDraft, MAXtreme etc which use GPGPU/shaders to help with CPU-intensive tasks. Also it's one thing to 'think aboot' developing for thousands of cores, but we're not really there and working out exactly HOW to fully utilise all those cores efficiently, especially for things like games which require a balance of GPU, shader & CPU processing without getting the whole thing bogged down is another thing altogether, whether it be on LarryMosquito, GTXR3i or ATchoo.
Comment by EdArch at 5:36pm 21st July 2008
Larrabee sounds quite disgusting, what an excuse to muscle in on a market. I hope ATi and nVidia show Intel something about making GPUs: another round of 8800GTXs or 9800 Pros anyone?
Comment by Zhaoman at 5:28pm 21st July 2008
Interesting. I have to agree with Intel on this view, developers should be thinking about Thousands of Cores. Problem with game developers at the moment, they are too damned lazy to implement this technology!
Comment by C7ouD at 4:39pm 21st July 2008
Make a Comment
- One top four motherboard maker will die, says MSI
- AMD plans supercomputer with 1,000 GPUs
- Nvidia announces GeForce GTX 285
- Creative takes wraps off Zii processor architecture
- Nvidia announces stereoscopic 3D glasses
- BFG unveils hand built Phobos gaming PCs
- UPDATED: LiveJournal announces mass layoffs
- Top 10 PC games chart
- Seagate Barracuda 7200.12 gets 500GB on single platter
- Creative reveals new Zii processor
These are the most popular news articles over the past seven days.
- Phenom II overclocked to 6GHz
- GeIL unveils Cyclone RAM cooler
- Rumour control: GeForce GTX 260 GX2 in January
- Nvidia cuts price of 216 SP GeForce GTX 260
- VIDEO: PC version of Mirror’s Edge features hardware PhysX
- Get Half-Life for less than a dollar on Steam
- IBM plans to build virtual human brain
- RivaTuner 2.20 adds SLI and CrossFire support
- VIDEO: ArmA II trailer released
Mobile Broadband
Fastest, cheapest 3G mobile broadband dongles from 3, Vodafone, T-Mobile and Orange
from just £10/month
