Printable View
It seems you need a dual card solution for this to work well. Performance hit with a single card solution (in games) is quite bad apparently. Any one in Melbourne with a motherboard with dual PCIE 16x slots and a ATi card? I want to try something out...
Just as I thought, the physics is too taxing on a single GPU. I still like the idea of a dedicated card, I may just get a PhysX card depending on how the on GPU physics turns out.
There have been mixed reports.... I guess we'll have to wait until TT or VRZone look into it.
We'll just have to wait and see.
All I know is, it seems possible to have a NVidia card running PhysX and an ATi card running graphics. Could this mean NVidia may release a card with no ports but with a GPU core?
EDIT. After posting this I found this interview with Derek Perez, NVIDIA's Head of Public relations. the quotes below in red are things he says that support my earlier conclusions.
From Nvidia;
"Delivering physics in games is no easy task. It's an extremely compute-intensive environment based on a unique set of physics algorithms that require tremendous amounts of simultaneous mathematical and logical calculations.
This is where NVIDIAŽ PhysX™ Technology and GeForceŽ processors come in. NVIDIA PhysX is a powerful physics engine which enables real-time physics in leading edge PC and console games. PhysX software is widely adopted by over 150 games, is used by more than 10,000 registered users and is supported on Sony Playstation 3, Microsoft Xbox 360, Nintendo Wii and PC.
In addition, PhysX is designed specifically for hardware acceleration by powerful processors with hundreds of cores. Combined with the tremendous parallel processing capability of the GPU, PhysX will provide an exponential increase in physics processing power and will take gaming to a new level delivering rich, immersive physical gaming environments with features such as:
* Explosions that cause dust and collateral debris
* Characters with complex, jointed geometries for more life-like motion and interaction
* Spectacular new weapons with incredible effects
* Cloth that drapes and tears naturally
* Dense smoke & fog that billow around objects in motion
The only way to get real physics with the scale, sophistication, fidelity and level of interactivity that dramatically alters your entertainment experience will be with one of the millions of NVIDIA PhysX-ready GeForce processors.*
*Note: NVIDIA will deploy PhysX on CUDA-enabled GPUs later this year. The exact models and availability will be announced in the near future."
~~~~~~~~~~~~~~
Couple of points:
1. The move will be towards no separate PhysX cards, although currently still available - PhysX algorithms are being incorporated into the NVidia drivers. The GPU has plenty of stream processors that are perfectly suited to PhysX work - much more so than the CPU. With new driver releases your 8 or 9 series Nvidia GPU will become more and more PhysX enabled in and of itself.
No separate PhysX cards:
"Physics is a natural for processing on the GPU"
FiringSquad: "Are there any plans to release new stand alone PhysX graphics chips or will AGEIA's technology be integrated into new NVIDIA GeForce graphics chips or nForce motherboards?"
Derek Perez:: "We have no announcements at this time." <------- WOOOOO Look for the lie "The only way to get real physics with the scale, sophistication, fidelity and level of interactivity that dramatically alters your entertainment experience will be with one of the millions of NVIDIA PhysX-ready GeForce processors.*" - from the above Nvidia release.
"we are going to enable GPU Physics as soon as possible."
"AGEIA designed its multithreaded PhysX software specifically for hardware acceleration in massively parallel environments. NVIDIA’s GPUs are well-suited to take advantage of PhysX software and AGEIA’s expertise".
2. Backwards roll out for 8 and 9 series GPU's is due this year. Exactly which models will be supported yet to be confirmed but it looks like perhaps all 8 and 9 series cards (i.e.all CUDA capable cards which implies they are porting the PhysX algorithms straight into the CUDA language and injecting it into their forceware).
3. "The only way to get real physics with the scale, sophistication, fidelity and level of interactivity that dramatically alters your entertainment experience will be with one of the millions of NVIDIA PhysX-ready GeForce processors." I do not think Nvidia will be making it easy for ATI to use this technology. However it may well be possible to use an Nvidia card dedicated to the PhysX work and an ATI card for the actual rendering and graphics output.
4. As to releasing a card with no ports but a GPU I think it is unlikely. If you think about it since the day SLI was released there has been a VERY strong argument for such cards which would offer the scaling of SLI at a lower cost as the second card would have a simpler production, layout, less materials etc. But it hasn't been done. I think Nvidia are moving away from separate PhysX cards with this move and I would be surprised if the AGEIA cards do not come off the shelves relatively quickly. What would an Nvidia GPU with no ports and PhysX incorporated be? It would be a PhysX card and I suspect these are on the way out.
6. I believe with CUDA development, buying AGEIA and some of Nvidia's other recent moves they are seeing a vast change in the nature of PC architecture on the horizon and are discerning a hard-fight period for the industry - and are therefore working towards positioning themselves to best take advantage of both the new developments and the inevitable industry shake ups.
Vast change in the nature of PC architecture on the horizon:
"Second, the computer industry is moving towards a heterogeneous computing model, combining a flexible CPU and a massively parallel processor like the GPU to perform computationally intensive applications like real-time computer graphics. Physics is a natural for processing on the GPU because, like graphics, it is made up of thousands of parallel computations, and with our CUDA technology, which is rapidly becoming one of the most pervasive parallel computing programming environments in history, we can open this exciting parallel processing world to applications desperate for a giant step in computing performance—such as physics processing, computer vision, video/image processing, and a world of exciting applications we’ve not yet imagined."
When a university can build a CUDA based supercomputer based around 4 x 9800 GX2 graphic cards for a total cost of $4,000, and which in some tests outperforms the $13.5 Million 1000 CPU cluster they built a couple of years ago, you KNOW there is major change in the air.
Crazy
EDIT2 HAHAHA Funny "Interview"Code:}
:rant
print RandomStrings$
--$ IF brain == null THEN END ELSE GOTO rant {{$
"Tech ARP : But will this mean an increase in price for NVIDIA graphics cards that come with the PhysX processor? Or will NVIDIA continue to produce cheaper cards without the PhysX processor?
NVIDIA : Not at all. We not only intend to integrate the PhysX processor for free, we will now price our cards at least 20% cheaper than ATI's cards. If we have a vision for the next 5 years, it's to drive ATI into the ground, once and for all. When that happens, maybe we will buy those boys out and then we will be at the top of the telephone books too."
That wasn't a real interview. ;)Quote:
Originally Posted by Crazy Buddhist
Nvidia have no intention of buying ATI, as it would mean the following:
a.) AMD (which they would need to buy also) would probably lose their x86 license due to Nvidia not being based in the US which is why AMD managed to get that license, which means their CPUs would be in no position to be used in Windows based desktops.
b.) Nvidia has explicitly said before that they'd prefer to buy AMD as a processor firm, but wouldn't want the rest of it's 'baggage' - e.g. ATI.
c.) They'd be fighting a war on three fronts rather than one if they did manage to get the x86 license to stay in their hands: Chipsets, Graphics and Processors, which is part of the reason why AMD is in so much trouble.
d.) AMD is currently making a loss; the only part making a profit is from ATI, so they couldn't afford to sell ATI on it's own as they'd soon become bankrupt (even with the money made from selling ATI, they'd probably not have enough money to guarantee the ness. research.)
e.) Intel is making their own dedicated graphics solution anyway, so they'd be back to square one as soon as that was released (1 company vs 1 company) and Intel has a much larger amount of money to research etc.
f.) AMD would probably have much less of a claim of 'monopoly' against Intel due to becoming part of a larger company; this would allow Intel to use more aggressive business tactics without getting into trouble.
Blimey you ARE clever. I didn't spot that. If i had I would have said something like 'EDIT2 HAHAHA Funny "Interview"' putting the word "interview" in quotationn marks for emphasis. Call me an Englishman and rub my nose in a ducks bottom.Quote:
Originally Posted by Anarchist
I did not, anywhere, say, propose or imply they did.Quote:
Originally Posted by Anarchist
Quote:
Originally Posted by Anarchist
Blah blah blah .... a) is rubbbish. b) see below c) boy you really missed the point of what I wrote d) rubbish e) yes we know this f) rubbish ...
Nvidia won't buy AMD because they would also be buying the only other significant GPU business as part of the deal and would have a 90%+ monopoloy in the Graphics market through AMD's ownership of ATI. They would then be forced to sell the ATI business by both US and European monopoly authorities and there is no obvious buyer.
In this other thread I predicted something about current trends in the market. What I predicted in that thread is that Nvidia will make more integrated "systsem on a die" computers, that that is where the industry as a whole is headed and that the coming war in the industry will be Nvidia vs AMD and Intel.
Crazy.
Completely wrong. :rolleyes:Quote:
Originally Posted by Crazy Buddhist
The majority of the graphics market, even without releasing their dedicated graphics solution yet is controlled by Intel and both courts in the US and Europe would recognize this. As long as people were able to run desktop computers, with a graphics solution, capable of running modern operating systems, without having to use dedicated graphics they would not risk monopoly claims - for example, you CAN play pretty much every game on Intel integrated graphics, it's just not that great of an option. :rolleyes:
Look up the terms of the Intel/AMD x86 license agreement, you will find a is right. The agreement is not valid overseas. For d. - "The chipmaker's net loss for the quarter was $358 million" http://www.ecommercetimes.com/story/62648.html Loss was greater in last quarter of 2007.
f I believe you've just contradicted yourself. ;)
One GIANT head f*** for me lol.