• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Nvidia offers PhysX support to AMD / ATI

That's pretty sweet but really it makes a lot of sense. No one is going to develop for PhysX if Nvidia cards are the only ones that can use it. Nvidia has to get ATI onboard with PhysX or it will never go anywhere.

We saw this with the add-in cards, no one wanted to develop for a card no one had, and no one wanted a card that no one was developing for. The same would go for excluding all the ATI customers out there. Only once nearly everyone has PhysX capabilities, will we start seeing it in more titles, and as more than simply effects physics, but gameplay physics as well.
 
Great news. It would be silly if Amd/Ati didn't join in. All the same I love the fact I'll be able to get more juice out of my 8800.
 
Yeah, we'll see... They're going to ditch the physics processor just so they can sell more video cards, they may "support" the physx cards but they sure as hell won't make any more. Though there is a possibility this could be used as a stepping stone in getting people to adopt dedicated cards.

Everybody has a video card, with cuda, they can run the physics in "enabled" games, then, when there are games with the support, the dedicated cards can will take over in doing the tasks and then you'll see the performance improvement. But my vision is that nvidia will drop the physx card and just stick with it being on the video card despite the fact that video cards are suppose to be for just that, video.
 
abudhu said:
Great news. It would be silly if Amd/Ati didn't join in. All the same I love the fact I'll be able to get more juice out of my 8800.
Click to expand...
I don't think that you will see Physx hardware enhancements until after the 9000 series and I think that it is highly unlikely that you will see any software Physx enhancements for the 8800 series for a while if at all.
 
We will probably see a Geforce F@H client before PhysX shows up on any GPU platform.
 
Biskquik said:
We will probably see a Geforce F@H client before PhysX shows up on any GPU platform.
Click to expand...
Why do you think that? No one is even working on a F@H client specifically for nvidia GPUs (because of the way it's tightly controlled by Pande/Stanford), while the CEO said nvidia is already porting PhysX to CUDA: "We're working toward the (PhysX)-engine-to-CUDA port as we speak. And we intend to throw a lot of resources at it."
 
[21CW]killerofall;1032252638 said:
I don't think that you will see Physx hardware enhancements until after the 9000 series and I think that it is highly unlikely that you will see any software Physx enhancements for the 8800 series for a while if at all.
Click to expand...

Yeah, but one can dream. :)
 
LOL nvidia will support AMD/ATI even though no support outside 8xxx and 9xxx series?
Give me a break, you guys really thin nvidia will support other chip makers before they support their own full line?They will support it the same way PhysX work now on ATI/Nvidia, that is software!
Yep ATi owners just don't hold your breath on hardware support, would make tons more sense had they announce support accross the board for the nvidia's own line and then some more...As far as we're concerned they only tangible information we have is that the big "N" is working into integrating PhysX API to CUDA language, everything else is just wishfull, if anything happens announcement will follow.

And the biggest laugh comes from PPU comments:"support will be provided as demand dictates" WTF!!! please mr. PR lemme drink your words...This means what exactly?

-As long as PPU sales do not sink under a certain amount?
-As long as manufacturers contracts hold?
-As long as it's good for PR?
-As long as the API is integrated to CUDA?
-As long as it doesn't cost you a dime?
-As long as it looks good?
-As long as the SDK don't outgrow the current drivers set?
-As long as anything, as long as it keep raining outside the office...
 
BoudahXL said:
LOL nvidia will support AMD/ATI even though no support outside 8xxx and 9xxx series?
Give me a break, you guys really thin nvidia will support other chip makers before they support their own full line?They will support it the same way PhysX work now on ATI/Nvidia, that is software!
Yep ATi owners just don't hold your breath on hardware support, would make tons more sense had they announce support accross the board for the nvidia's own line and then some more...As far as we're concerned they only tangible information we have is that the big "N" is working into integrating PhysX API to CUDA language, everything else is just wishfull, if anything happens announcement will follow.

And the biggest laugh comes from PPU comments:"support will be provided as demand dictates" WTF!!! please mr. PR lemme drink your words...This means what exactly?

-As long as PPU sales do not sink under a certain amount?
-As long as manufacturers contracts hold?
-As long as it's good for PR?
-As long as the API is integrated to CUDA?
-As long as it doesn't cost you a dime?
-As long as it looks good?
-As long as the SDK don't outgrow the current drivers set?
-As long as anything, as long as it keep raining outside the office...
Click to expand...

Supposing your right, then dedicated physics just died. The majority of devs will then only code for it if Nv throws money at them. Nv has a limited amount of money to throw at it. If it only runs on Nv cards outside of software emulation, it will die, or just be mostly ignored. Just like it is now.

However, Nv could allow it to run in hardware on Ati cards, but make sure it does not run as well as it does on Nv's own stuff, with er, um, updates to the sdk and the built in advantage of knowing what is going to happen next with the api, and drivers. Doing that they can maintain the appearance of fair play while always being ahead of Ati in the physics dept.

We are both being a bit cynical. But I have seldom been given reason to not be.
 
Supposing your right, then dedicated physics just died. The majority of devs will then only code for it if Nv throws money at them. Nv has a limited amount of money to throw at it. If it only runs on Nv cards outside of software emulation, it will die, or just be mostly ignored. Just like it is now.

However, Nv could allow it to run in hardware on Ati cards, but make sure it does not run as well as it does on Nv's own stuff, with er, um, updates to the sdk and the built in advantage of knowing what is going to happen next with the api, and drivers. Doing that they can maintain the appearance of fair play while always being ahead of Ati in the physics dept.
Click to expand...
Yep thinking in the same direction making OpenSDK not so open suddenly ;)
 
Hmm, AMD and Intel will both have 8-core processors soon....and of course their own GPUs. How in the world will those companies leverage their position? By propping up NVIDIA GPU PhysX? I don't think so. Physics in the future has already been decided by AMD and Intel and it will continue to be done on both of its multicore processors. That is my take on it tonight though. NVIDIA is polishing the brass on the Titanic.
 
(this is if I understand GPU architecture right, could be higher if each pipelines is shader clock and not core clock)

8 core cpu at 3ghz: 24ghz proccesing power, for $1000

8800GT: 112 * .6 = 67.2ghz proccesing power, for under $200

Or at same price range:
3X 9800GTX = 259.2 GHZ

The potential requirements of a physics engine are basically unlimited. In order to be totally realistic, you would need to be able to do trajectories for millions of particles in real time. A dedicated GPU on CUDA can do 250,000.
 
You seem to be comparing performance....I am talking about business. Two totally different things. Remeber that 8 core will soon be just like dual core now days....
 
Fair enough. The point I was trying to make is that physics are way behind graphics, and CPU's don't have the performance to catch up. For example, In Half Life 2, Alex's face is nearly photo-realistic, yet a crowbar passes straight through it.

I think companies are more likely to end up using the existing dedicated graphics hardware than inventing their own. The only exception to this is Intel, with Larrabee (sp?). AMD has neither the money nor time to invest in building a physics platform onto the CPU or GPU. For them, this is the easy way out. And Intels previous attempts to deviat from industry standards (wireless A, for example) have been less than succesful.
 
I agree the card never had a chance without widespread support. Suprised it lasted this long, I do hope that NV does do something with this tech though.
 
I don't think nvidia would have spent that kinda money if Ageia's technology wasn't good. The fact that they have invested so much means something at least. Either they will adapt the Physx engine to work on the GPU or maybe even bring out PPU enabled video-cards (which is really what Ageia should have pushed for from the start). If that happens nvidia will likely still try to support the early-adopters of the cards, while everyone else just runs it on a second video card in SLI (or something to that effect). Or maybe not, we'll see.
 
cybereality said:
I don't think nvidia would have spent that kinda money if Ageia's technology wasn't good. The fact that they have invested so much means something at least. Either they will adapt the Physx engine to work on the GPU or maybe even bring out PPU enabled video-cards (which is really what Ageia should have pushed for from the start). If that happens nvidia will likely still try to support the early-adopters of the cards, while everyone else just runs it on a second video card in SLI (or something to that effect). Or maybe not, we'll see.
Click to expand...
3DFX anyone?
 
cybereality said:
I don't think nvidia would have spent that kinda money if Ageia's technology wasn't good. The fact that they have invested so much means something at least.
Click to expand...
I'm not sure it was that much. nvidia is very stingy with investments, but this also seemed like a desperation move to counter Intel's purchase of Havok. A hint or possibly actual cost of how much nvidia paid should be in the next quarterly report.

The Ageia PPU was pretty much dead, but the PhysX engine did have paying customers (consoles and PC). It wasn't like nvidia purchased Ageia only for the PPU.

---
nvidia didn't buy 3dfx. It purchased some IP after 3dfx went bankrupt.
 
Excellent news, AMD would be silly not to support it. Maybe then we'd see some decent games supporting it.
 
I disagree that we're going to see the dedicated PhysX cards disappear. I think nVidia is making a wise move.

Get PhysX "technology" to become nearly ubiquitous by having a subset of it available on video adapters. This will get developers to start using it. Then make money from enthusiasts by offering a dedicated PhysX adapter that has more capabilities and can further enhance what these users are already seeing on their video cards.
 
I think nvidia would just offer a headless GPU as a physics accelerator if it wanted to sell a "dedicated" card. I don't think the CUDA port is going to be a subset either. The Ageia PPU was just a general processor (PCE) with 4 FP vector engines (FPE) accelerating PhysX engine with a program running on it.
 
pxc said:
I think nvidia would just offer a headless GPU as a physics accelerator if it wanted to sell a "dedicated" card. I don't think the CUDA port is going to be a subset either. The Ageia PPU was just a general processor (PCE) with 4 FP vector engines (FPE) accelerating PhysX engine with a program running on it.
Click to expand...


That is simplifying it a tad too much...
 
You must log in or register to reply here.
Back
Top