What is Digg?Sponsored by Travelzoo
Take Advantage of Ridiculously Low Holiday Airfares view!
travelzoo.com - Flights $52 and up for Thanksgiving, Christmas & New Year. But move on it now.
66 Comments
- Woknblues, on 10/12/2007, -1/+28in other news, Microsft said that Apple is a doody head.
- TheGalacticFork, on 10/12/2007, -0/+12"Anything you can do, I can do better!"
-"I can do anything better than you!"
"No you can't!"
-"Yes I can!"
"No you can't!"
-"Yes I can!" - psylence, on 10/12/2007, -0/+11The 1980's called, they want their math co-processor back.
- Mr_Lyle, on 10/12/2007, -0/+10ATI and NVIDIA have, by far, the greatest big dick contest ever between 2 companies. The only way they can be toppled is if Bill Gates and Steve Jobs went at it on Celebrity Boxing.
- automagnus, on 10/12/2007, -0/+10Did you even read the article?
The whole point of it is that they claim a physics card isn't needed - Choppy, on 10/12/2007, -0/+10Next thing we know, ATI and NVidia will claim that their GPUs make breasts look bigger and bouncier.
But seriously though, is there any proof of their claim? Simply saying that the hardware is capable without showing evidence or benchmarks is not going to make me believe them. - SpringBoy, on 10/12/2007, -0/+10Realtime ray-tracing running off a CPU - 12 years ago? Are you sure?
The OpenRT ( http://www.openrt.de ) project has previosuly shown that software solutions run dramatically slower than hardware solutions. It's simply common sense: a general-purpose CPU CANNOT provide the same amount of power in any given application as can dedicated hardware. They had a Pentium 4 3.0GHz running Quake3 with RT much slower than a custom made GPU running at around 300MHz.
Graphics are obviously the best example. GPUs run at around 1/6th the clock speed of high-end CPUs and yet provide dramatically faster solutions to what they do - because their hardware is completely designed to do those tasks. Try switching a game to software rendering (CPU) sometime and see how your framerates go. :)
Dedicated hardware is not really creating a hardware limitation. They're making it possible to perform those tasks on anything less than $50,000 computers. You'd probably need a CPU like Cell to be able to provide real time rendering of 3D graphics - and that wouldn't leave much power for anything else to run.
It's also not the hardware that makes it possible for "anyone" to use 3D graphics - it's the libraries that provide easy access to 3D manipulations (OpenGL/DirectX). - epicbard, on 10/12/2007, -0/+8Now if they could only make a decent Linux driver on par with the Win driver they make...
- prockcore, on 10/12/2007, -6/+14Oh yeah? Well my dad can beat up your dad!
Let's see some actual physics cards, ATi. - motbob, on 10/12/2007, -5/+11This article doesn't have any real substance. All dailytech is reporting is that ATI claims it is better than the competition, which is not news. I'm sure that NVIDIA would say the same thing about their GPUs.
- thegsa, on 10/12/2007, -0/+6Patent That, Quickly!
- prockcore, on 10/12/2007, -4/+9Sorry, I meant actual physics examples using their GPU.
My brain is fried.. it's friday.
I think we'll all end up wih dedicated physics cards eventually. The GPU just isn't general enough to work well for physics. I remember when 3d cards just started coming out. I didn't think they'd take off, and I thought they'd stunt growth in 3d. I think I was right too. Before the first Voodoo card was available, the demoscene had realtime raytracing running off CPU.
12 years later, and realtime raytracing is still limited to the demoscene. 3d cards have created a hardware limitation that didn't exist before.
However, 3d cards have also made it easier for practically anyone to make a 3d game. Physics will go the same way. Sure the cards will create a hardware limitation, but it will also lower the barrier to entry. - Vokas, on 10/12/2007, -1/+6My money is on Bill.
- BlackChaos, on 10/12/2007, -1/+5And if they could make a quality Win driver while their at it, that would be cool too.
- RobotCitizen, on 10/12/2007, -1/+5All this hype over what is nothing but the return of math coprocessors. Is it 1988 again?
- DocDEB, on 10/12/2007, -0/+4DNF has been delayed?!? I thought it was just 3D Realms normal development cycle.
- inactive, on 10/12/2007, -1/+4yet ati manages to ***** up its every release of linux drivers...
i switched to nvidia sometime back after trying a radeon, and i cant tell you how happy i am, playing games in linux... - maxhrk, on 10/12/2007, -0/+3i think i rather have one pure about graphical and shade thing on one card, and other card for physic. I dont want all smash up into one card anyway.
- mike_p, on 10/12/2007, -1/+4Yep... right now, we begin the "hyping cycle"... I won't trust anything until I see some benchmarks using Intel/AMD/AMD64 (since it's supposed to be taking the load off the CPU)...
- inactive, on 10/12/2007, -0/+3guys ! shhh....
stop making so much noise..
3drealms will now have one more excuse to delay duke nukem forever...
shhhhh.... - kokobaroko, on 10/12/2007, -0/+3have you seen this?
http://graphics.cs.uni-sb.de/SaarCOR/
small 100MHz FPGA pumping more RT graphics than few P4 3GHz ... imagine what would happen if NVIDIA implemented such algos in their 600MHZ beast chips with 1200MHz ram - bobothn, on 10/12/2007, -1/+3remember the story a while back were ati and nvidea people were about in a fight?
- CaptainMal, on 10/12/2007, -0/+2AGEIA is GIVING AWAY THEIR API, man. You can license it for free. Even a modder could use it without having to shell out. Their model is that they'll make cash selling the cards, while handing all developers FREE PHYSICS TOOLS.
They are teh win.
It's ATi and Nvidia that are on separate paths; still trying to justify SLI and Crossfire, which will not even come close to doing what the Physics processor can do. Nobody who actually looks at the specs of the PPUs they are cranking out would call it "just a math co-processor". It's clearly designed (just like 3d cards are) to do one job and do it better than anything else can. It's the exact OPPOSITE of the Cell design, which is just to make basic generic computational chips and run them all in parallel.
For some applications, the Cell is the perfect solution. Supercomputers comes to mind. Parallelism and sick bandwidth with low latency between them.
For gaming, which is not as predictable as many HPC tasks, taking physics away from the CPU and GPU and building another specialized processor that can increase the opportunities for developers to realize more true-to-life realistic environments, is a HUGE step in the right direction. Keep in mind that it's largely due to the PCIe bus that this will be largely scalable and future-proofed, since even more data will be flowing over the system bus. AMD and the Hypertransport folk must be ecstatic.
I say bravo AGEIA, and they will most certainly get my $250 when UT2007 hits stores. I'll get sick physics, AND more performance out of my GPU and CPU. Can't wait. - warnipples, on 10/12/2007, -0/+2http://physx.ageia.com/footage.html look at this awesomeness
- swax, on 10/12/2007, -1/+3let the physics war begin, this is great. graphics updates haven't been doing much for me lately, still the same old, same old, static everything. this is exactly what we need nvidia and ati to be battling over to get some really cool games coming out. and I dont mean just ***** blowing up, but fully alive and changing enviroments, flowing streams that take their own path, different materials interacting, bending, breaking, and ***** you and I haven't even dreamed up yet.
- trogdoor, on 10/12/2007, -0/+2Can this do interactive physics to enhance gameplay or just make things _look_ more realistic?
- xodex, on 10/12/2007, -1/+3Sorry but ati/nvidia don't know the first thing about breasts, They'll need to talk to Itagaki from Team Ninja if they want any counseling on how it's done.
- AngryPenguin47, on 10/12/2007, -0/+2I would just be impressed if they put some effort into their linux drivers.
- alloneword, on 10/12/2007, -0/+2> Now if they could only make a decent Linux driver on par with the Win driver they make...
Thats why I chose an nVidia card. - SpringBoy, on 10/12/2007, -0/+2Well, it'd be nice if their OpenGL implementation was good in the first place. Then they can try writing a better driver for it. :)
- DEFSMAC, on 10/12/2007, -0/+2ati and nvidia reminds me of ford and chevy. they are both the same thing, do basically the same thing, yet people will forever support one brand and hate the other.
- JohntB, on 10/12/2007, -0/+2It could be that the hardware is wrong for the purpose. What if they'd held off creating GPUs until people were using realtime raytracers? Then, the hardware might be designed for raytracing, and not for polygon transformation and coloring. The existence of hardware designed for our existing model has locked us into that model.
- meatbites, on 10/12/2007, -0/+2What I'm concerned about is the lack of open standards -- and particularly with the reactions we're seeing here. All three of them -- NVIDIA, ATI and AGEIA -- are running their separate paths.
Why, oh why? If the past is, you know, anything to go by, this is surely the easy way to kill off a new technology.
This is about as fantastic as the new competing optical mediums.
In light of this, and also the issue of controlling who does and doesn't have such a physics processor, I see the consoles being the first to benefit, as it's far easier to control who has one. - iHawk, on 10/12/2007, -0/+1PC components are not like stereo components or televisions. Low-end in PC is usually last years hardware, and not typically an entry level offering so I'm not sure how you could argue that nVidia does a better job with the high-end hardware while suggesting that their mid and lower tier offerings are inferior to ATI. Perhaps you mean that nVidia cards do not stand the test of time?
I would argue that both ATI and nVidia offer high quality cards, but the hardware vendors have a lot to do with the end-user experience. nVidia seems to license a lot more vendors than ATI and some are certainly better than others. Stick with proven hardware vendors, whether using nVidia or ATI and you're bound to have a better experience than by purchasing a brand X card from a discount manufacturer. - meatbites, on 10/12/2007, -0/+1CaptainMal, I can see where you are coming from, however the same logic can also be applied to both ATI and NVIDIA -- and anybody else with enough care-factor to create a physics API/processor/both. That is to say: what if ATI and NVIDIA were to also offer a fully free API? You'd consider this a good thing? Sure, developers would have access to three or more APIs, which are available for free, but these developers would have to actually support all of these APIs. I would assume that would involve a lot of development time, much along the lines of the different video card APIs.
It would be an absolute nightmare, but that isn't my point. My point is that, from my understanding, this isn't an open standard -- AGEIA controls 100% of its API and the direction it heads in. This way, only AGEIA can sell the hardware -- you said this yourself. I don't consider that a positive situation for the PC market. For consoles, proprietary hardware (regardless of 'free-as-in-beer' APIs) is of little concern to anyone, but on PCs it's a whole new bucket of Yakety Sax. - warnipples, on 10/12/2007, -1/+2I totally agree we should wait for the mighty carmack to comment on this all other opinion is false!!!
- anenokoji, on 10/12/2007, -3/+4I disagree, a physics card will be needed... for vertice transformation, you have to run a sin(), cos() and power() function for every x,y,z. and you have 3 vertices per poly. based on a 7800gtx max vertices/sec its about 7gflops a second(based on 60fps), just for vertice transformations. doesnt include ai or any other garbage like that.
gaming will always be infinately parallel? because the polys will always increase in number. - TheGrizah, on 10/12/2007, -1/+2This isn't the first time ATi stated that physics could be done on a GPU or a Graphics Card. I would rather see a PPU integrated onto the same card as a GPU or something like that though.
- inactive, on 10/12/2007, -0/+1Never fear; PCI express is here!
- inactive, on 10/12/2007, -2/+3From every example I've ever seen, they're talking about physics in a visual-physical sense, not a virtual-physical sense. So no, it doesn't create new levels of interaction, only pretty graphics. If it created new levels of interaction that would make a game hugely flawed. Players with the physics cards would be interacting with game elements that players without the card can't.
Of course, one argument I saw about these new physics processors was that they actually penalize those who have them. If you have the card, your line of sight is blocked by huge clouds of smoke and debris, while someone without the card sees a small puff of smoke and still has clear line of sight to you. - Mysk, on 10/12/2007, -1/+2ATi is a company of nothing but press releases. They talk big, but they don't deliver. This article is nothing but a way for them to keep their name in the press.
In the mean time I'll be enjoying Havok physics being computed on my trusty 'ol 6800 GT.
The next time that I spend $300+ for gaming hardware it's going to be on a shiny new 3D card that will do all of the great gaming bells & whistles and should be (to my tastes anyway) perfectly capable of doing the physics as well. - iFindout, on 10/12/2007, -0/+1That's them fighting words! I guess their recent shouting showdown was the drop that tipped the bucket. Lets watch, shall we?...
- inactive, on 10/12/2007, -3/+4The thing in this article that excites me is the idea of being able to use my old graphics card as a discreet physics processor without having to have it linked to the main GPU, meaning I can use my old graphics card as a physics processor, rather than having to buy one of these new physics cards
If this is the case then it's an absolutely fantastic idea, and a bigger deal than I think most people are realising - CaptainMal, on 10/12/2007, -1/+2Not really, they're just trying to come up with something useful for those SLI people to use their extraneous video card for, since it only translates into a 20% improvement in performance for 200% of the cost.
AGEIA is going to spank them both soundly, and Havok can't hold a candle to the AGEIA physics capabilities.
The Unreal 3 Engine uses AGEIA for physics, so start saving. - CaptainMal, on 10/12/2007, -0/+1Look at it this way meatbites;
Right now, they're trying to cram physics into an API built for graphics or CPU.
Much like they all jumped at the chance to use a new specialized API to offload graphics from the CPU and allow for more realistic gaming experience, they will also jump at the chance to offload physics from the CPU and GPUs, to again improve the experience and make a better game.
Witness the fact that Unreal Engine 3 signed on, among many, many other companies, and you'll see that there is no doom-saying necessary: gaming companies are dying to provide a better experience for their customers, and the biggest names in the business recognize that.
AGEIA won't be the only ones selling their hardware anymore than Nvidia or ATI are the only ones, and nothing about their business model prevents another company from providing a compatible and competitive product. Just like AMD competes with Intel and ATI and Nvidia compete with everybody else. For instance, BFG will be producing AGEIA cards, and I think ASUS has hopped on as well. - DocDEB, on 10/12/2007, -1/+2Consoles are also where physics as integral to game play can be implemented. On the PC side it can't be more than graphic fluff since the developer could never count on a physics proc being present — unless of course all graphics cards come with an onboard physics proc — but then that would likely be only the high end boards and you are back to the same problem.
- cwcheang, on 10/12/2007, -0/+1Awesome comments. insightful, interesting, funny, informative.. everything.
You love digg more everyday - CaptainMal, on 10/12/2007, -0/+1AGEIA is giving away their API man. Look at Havok's business model; they charge the developer for it up front.
AGEIA is going to win this hands down. I say 3 years tops before every gamer has a PPU in his rig. AGEIA has really approached this brilliantly. - kaiser79, on 10/12/2007, -0/+1Ageia wins (I hope). If gamers have proven anything, it's their willingness to spend lots of money on hardware. Meanwhile, game developers continue to push the hardware to the absolute limit. There certainly isn't extra headroom on the GPU to take up the physics processing responsibilities. I suppose this just means selling more hardware to nVidia and ATi.
The bigger problem is when the programmers have to code for 3 different physics APIs. Games will cost *even more* to program and content will continue to subsist. And who wants to innovate when it means hours of frustration getting it to work on all these different APIs and hardware combinations?
I hope Microsoft will do something like they did for pixel shaders in DirectX (HLSL pixel shader programming language). While they're at it, how about an AI Model? Yeah yeah, MS is the devil and all that, but credit where credit is due. - TheHim, on 10/12/2007, -0/+1Would sure interest me in terms of content creation. Investing into an anyway required powerful graphics card and make use of it for non-realtime physics simulation instead of using the CPU alone or buying expensive additional hardware would be fantastic.
-
Show 51 - 66 of 66 discussions
© Digg Inc. 2009
— Content created and posted by Digg users is