• 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.

PhysX Support Coming in New ForceWare Drivers

Terry Olaes

Terry Olaes

I Used to be the [H] News Guy
2FA
Joined
Nov 27, 2006
Messages
4,646
Looks like we’ll finally see the fruits of Nvidia’s February 2008 acquisition of Ageia. HotHardware is reporting that ForceWare 177.39 will add GPU-accelerated PhysX capabilities to the 9800GTX, 9800GTX+, and the GTX 260/280 cards. Full support for 8 and 9 series cards will gradually roll out. How this will translate to real world performance has yet to be seen but it’s certainly interesting.
 
Hopefull nvidia can push for hardware acceleration better than Ageia was able to. There were plenty of PhysX engine games, but the PPU made little difference for most aside from varying quality effect enhancements, a few special levels and the Cell Factor "game."
 
Interesting, but are there any new titles coming that are still supposed to support hardware accelerated PhysX? Do Ati and Intel intend to support it? If not, hardware accelerated PhysX is pretty much dead in it's current form imho. We will get a few Nv sponsored games that use it and then it will quietly vanish or become a truly open standard imho.

Yes, PhysX is supposedly open. But, Nv controls everything regarding it, and can ensure they always perform best by changing physX to best suit their hardware or to least suit competitor hardware. Which is prolly why Ati and Intel have not signed on. If it's calls get added to Dx and/or Ogl, or becomes a truly open standard that Nv no longer controls, it will have a much better chance of acceptance imho.
 
Sweet! Now..
1) is PhysX in more than 1-2 games worth caring about if you run out of AAA, AA, A, B-titles to play? No, games with a special PhysX gimmick level do not count.
2) are sites going to fall for it and consider them ten times better than ATI cards just because Nvidia cards using this get get higher scores on non-graphical tests like that?

(1: No, 2: Yes)
 
177.39 works fine on the 8800GT with hardware NVidia PhysX support already.
 
Nenu said:
177.39 works fine on the 8800GT with hardware NVidia PhysX support already.
Click to expand...

Might want to expand that to include G92 based cards, ie. 8800GT and GTS 512MB. Got it working on both a 8800GT and a 8800GTS 512MB G92.
 
ir0nw0lf said:
Might want to expand that to include G92 based cards, ie. 8800GT and GTS 512MB. Got it working on both a 8800GT and a 8800GTS 512MB G92.
Click to expand...

I didnt to want to expand on it as I only have experience of the driver with my 8800GT.
I'm glad you can clarify it works on other cards :)
 
So does 177.39 Forceware also support CUDA? And do I need to download a seperate PhysX driver?
 
Nenu said:
177.39 works fine on the 8800GT with hardware NVidia PhysX support already.
Click to expand...

ir0nw0lf said:
Might want to expand that to include G92 based cards, ie. 8800GT and GTS 512MB. Got it working on both a 8800GT and a 8800GTS 512MB G92.
Click to expand...

Can you guys confirm that PhysX support is detected and enabled in a PhysX game like UT3 or GRAW/GRAW2? Maybe even Cell Factor? If you can confirm it, I'll post it.
 
I downloaded the improved PhysX levels for UT3 and they run like a dream on my 8800GT, 177.39 driver and NVidia PhysX driver.
I dont know what they would have run like before though.
 
Well, this is good news for me. I've been contemplating a physics card for a while, but now I may jump from ATI to NVIDIA.

Sure, there may not be very many games. But look how many PhysX cards there were out in the consumer boxes. But, how many NVIDIA cards are there? A whole lot more. This will entice developers to program for physics in games. There is a much larger market now. There will be a lot more games in the future. Cool part about that is, you don't have to buy the hardware and have it sit in your PC doing nothing. It's a video card upgrade, so you see benefits from that. And it does a great job at that. So, I see it as a win/win situation.
 
from the 9800gtx+ vs 4850 review at Legit:

3dmark_vantage_cpu.jpg
 
This isn't going to mean a hell of a whole lot, most physics level in games can be handled by todays dual and quad cores easily, I don't see this being a big advantage for them as long as these games support dualcore and quadcore processors.
 
CompMage said:
Them numbers have to be a biggest crock of shit I've ever seen.
Click to expand...
A bit angry there? :p

The chart is for the CPU score, which consists of 2 tests. One test gets an 11x speedup from the GPU accelerated PhysX. The relative scores are fine for what it is measuring in those two tests.

Whether or not a 3DMark score means anything for gaming is open for debate. I feel it doesn't mean much since it doesn't represent any games (no games use the 3DMark engine), and the two GPU vendors optimize drivers to the hilt to get a high score. The shader usage and model complexity is also atypical. To top it off, it's just a glorified scripted timedemo that means much less than actual gameplay with all parts of a game's engines running.
 
pxc said:
A bit angry there? :p

The chart is for the CPU score, which consists of 2 tests. One test gets an 11x speedup from the GPU accelerated PhysX. The relative scores are fine for what it is measuring in those two tests.
Click to expand...

The other question would be - what about the actual PhysX hardware? Running the 9800GTX+ with PhysX drivers is well and good, but it's a massive, power-hungry, double slot card. What if you paired a Radeon 4850 with one of the Ageia PhysX cards themselves? Lower power, better performance? Or....???
 
dderidex said:
The other question would be - what about the actual PhysX hardware? ... What if you paired a Radeon 4850 with one of the Ageia PhysX cards themselves?
Click to expand...
Yeah, what about actual PhysX hardware? Since it's now an nvidia product and essentially discontinued (aside from the Asus PCI-E version), it's up to nvidia and Futuremark to enable support in the drivers and benchmark for the old PPU to run.

I'm going to guess that nvidia won't be in a hurry to do that for Vantage, although they will still support the old PPU in PhysX games that support hardware acceleration. :p nvidia will be pushing GPU physics because it now has a large lead over everyone else. It's still iffy whether game developers will actually support it considering the lackluster hardware support the PPU even with a "free" SDK. If the gamble pays off (i.e. devs announce new titles and most existing PhysX titles work with GPU physics), nvidia will have a pretty good trump card this generation.
 
pxc said:
Yeah, what about actual PhysX hardware? Since it's now an nvidia product and essentially discontinued (aside from the Asus PCI-E version), it's up to nvidia and Futuremark to enable support in the drivers and benchmark for the old PPU to run.

I'm going to guess that nvidia won't be in a hurry to do that for Vantage, although they will still support the old PPU in PhysX games that support hardware acceleration. :p nvidia will be pushing GPU physics because it now has a large lead over everyone else. It's still iffy whether game developers will actually support it considering the lackluster hardware support the PPU even with a "free" SDK. If the gamble pays off (i.e. devs announce new titles and most existing PhysX titles work with GPU physics), nvidia will have a pretty good trump card this generation.
Click to expand...

It's already there. 3dMark Vantage, as released, 'officially' supported only the Ageia PhysX card for physic acceleration, and it make a fairly large difference in the CPU and physics scores. It also nearly doubled the number of gates in some of the tests to increase load at the same time.

This whole 'driver update increases 3dMark score' is just nVidia stacking the current GPU implementation directly on top of the older Ageia API, so that the app itself (in this case, 3dMark) can't tell one way or the other.
 
So, I guess, to take another tack at this...

...that chart shows a HUGE jump in CPU scores with a 9800 GTX+ running the PhysX driver on it vs a regular 9800 GTX+. What would a 9800 GTX+ with an Ageia PhysX card look like compared to THAT, though? Even higher? Lower? What?

They don't test it, though, so there is no way to tell if the current implementation is 'better' or 'worse' than the previous implementation. All they are showing is that using 'PhysX hardware' in a test designed for PhysX hardware performs better than when not using any.

Well...yeah. That's pretty obvious. They don't test the real question - is this better than a standalone PhysX card or not?
 
dderidex said:
What would a 9800 GTX+ with an Ageia PhysX card look like compared to THAT, though? Even higher? Lower? What?
Click to expand...

SLI physics? I bet Nvidia's driver team will love to implement that. So we'll have SLI Hybrid Power, SLI graphics, SLI physics and possibly 3/4 way graphics/physics also. Implementing the onboard gpu to do physics while the discreet gpu's do graphics seems easy enough to do...along with multi monitor support and multi gpu folding support.
 
dderidex said:
It's already there. 3dMark Vantage, as released, 'officially' supported only the Ageia PhysX card for physic acceleration, and it make a fairly large difference in the CPU and physics scores. It also nearly doubled the number of gates in some of the tests to increase load at the same time.
Click to expand...
Ah, I didn't realize that. Since no one seemed to be talking about PPU scores in Vantage, I thought it wasn't enabled.

I looked through the ORB and found this PPU score, the top one for a PPU I think: http://service.futuremark.com/resultComparison.action?compareResultId=183472&compareResultType=19

The PPU runs CPU test 2 about 3.5 times faster than a Core 2 Quad 3.0GHz. The 9800GTX+ runs that test 11x faster than the Core 2 Quad 3GHz. CPU test 1 depends on CPU speed. Even with the CPU in that link overclocked to 4.7GHz, the total CPU score is less than 2/3 the score in the graph posted in this thread: 20571 (PPU + 4.7GHz o/c QX9650) vs 33504 (QX9770 3.2GHz + GPU PhysX on stock 9800GTX+).

But who really cares about Vantage. The physics test isn't part of a game engine. Does anyone have recent PPU UT3 PhysX level benchmarks to compare? Of course the GPU will have to give up some performance to run physics, but the UT3 benchmarks seem to be physics bound anyways.
 
GORANKAR said:
Interesting, but are there any new titles coming that are still supposed to support hardware accelerated PhysX? Do Ati and Intel intend to support it? If not, hardware accelerated PhysX is pretty much dead in it's current form imho. We will get a few Nv sponsored games that use it and then it will quietly vanish or become a truly open standard imho.

Yes, PhysX is supposedly open. But, Nv controls everything regarding it, and can ensure they always perform best by changing physX to best suit their hardware or to least suit competitor hardware. Which is prolly why Ati and Intel have not signed on. If it's calls get added to Dx and/or Ogl, or becomes a truly open standard that Nv no longer controls, it will have a much better chance of acceptance imho.
Click to expand...

Intel is not going to sign on, they own Havok physics.

Intel graphics cards suck or are non existent so their adoption will be nil.

AMD is left holding an empty bag.

That leaves Nvidia PhysX as the winner with a HUGE installed base of GeForce cards.

I see what your saying, but I don't think the fact that nv owns PhysX will be a problem. Intel owns the x86 architecture and we still have AMD CPU's.
 
i want to know why G80 GPU's are not okay to use with this driver now....or is it that they are and just the .ini file needs to be edited for them to work...i have seen other reports on the web saying as much
 
nobody_here said:
i want to know why G80 GPU's are not okay to use with this driver now....or is it that they are and just the .ini file needs to be edited for them to work...i have seen other reports on the web saying as much
Click to expand...
Because nvidia didn't enable it for the G80 series yet. :p This preliminary support seems rushed to slow down ATI's pretty successful HD 4850 launch. Look at how much attention the 9800GTX+ is getting and it's not even shipping yet. But working GPU physics is pretty much a bombshell announcement considering nvidia purchased Ageia only 4 months ago.

Anyways, full G8x/G9x series support is supposed to be available next month and a G8x supporting beta will probably "leak" before then.
 
They're may be staggering the release so they dont get support issues from all the cards at once.
 
Dear [H]. how about making it so clicking on the title of the front page news item causes you to see the item rather than another page that has exactly the same blurb as the one on the front page.

Anyways this will be great, it might cause game designers to work a bit on imagination rather than just rehashing something with better graphics and less intuitive controls.
 
DeFex said:
Dear [H]. how about making it so clicking on the title of the front page news item causes you to see the item rather than another page that has exactly the same blurb as the one on the front page.

Anyways this will be great, it might cause game designers to work a bit on imagination rather than just rehashing something with better graphics and less intuitive controls.
Click to expand...


agreed, it has been brought up many times here, the people running the site have heard it many times and dont care to change it.....i dunno, probably has something to do with one more page that gets to load ads, and from that viewpoint i do understand it, but i know what you mean, i have been dealing with that on a daily basis and to be hoenst it annoys me every single time....STILL...it's just annoying and unnecessary IMO
 
I've tried it with the modified inf on my 8600 GTS (G84), but with no luck.
 
nobody_here said:
agreed, it has been brought up many times here, the people running the site have heard it many times and dont care to change it.....i dunno, probably has something to do with one more page that gets to load ads, and from that viewpoint i do understand it, but i know what you mean, i have been dealing with that on a daily basis and to be hoenst it annoys me every single time....STILL...it's just annoying and unnecessary IMO
Click to expand...

It's to make sure people goto HardOCP and not just the forums.
 
k1DBLITZ said:
Intel is not going to sign on, they own Havok physics.

Intel graphics cards suck or are non existent so their adoption will be nil.

AMD is left holding an empty bag.

That leaves Nvidia PhysX as the winner with a HUGE installed base of GeForce cards.

I see what your saying, but I don't think the fact that nv owns PhysX will be a problem. Intel owns the x86 architecture and we still have AMD CPU's.
Click to expand...

A small hole in your evaluation there. AMD is on board with havok. So no, AMD isnt left holding an empty bag, at least not yet until Intel fucks them over (we all know its coming just a matter of when).
 
DemonDiablo said:
A small hole in your evaluation there. AMD is on board with havok. So no, AMD isnt left holding an empty bag,
Click to expand...
The announcement was for *CPU* physics. That bag is pretty empty until Havok has GPU physics available (probably around the sale time as Larrabee comes out), and pretty much all GPUs will benefit.

PhysX is a separate library and AMD's refusal to support CUDA could put it at a disadvantage if GPU physics takes off much sooner than Havok has GPU available.
 
nVidia has offered AMD support for CUDA, and AMD turned it down? Well if thats not a huge blunder. One of the only reasons that puts me in huge sway over to nVidia at the moment is because its going to have the ability to encode hi-def videos likity split come this septemeber. Such power and ability cannot be ignored by encoders. For less than 200 dollars I can buy a gpu that can process 4 hd movies in the time it takes Intel's best quad cores to do 1. Only issue is its in h264 and not x264 but I'm fully confident in nVidias partners abilities to come out with a program that can support that codec as well.

Also though, even if its only cpu physics, sooner or later it'll trickle up to the gpu sector. Either way you look at it though, AMD is in the same boat as Intel so I wouldnt say the bag is completely empty unless Intel decides this whole gpu thing was just a childish adolecent fade and backs out. Then AMD gots a problem.
 
Vashypooh said:
It's to make sure people goto HardOCP and not just the forums.
Click to expand...

How does a link back to itself make people go to HardOCP? That makes no sense to me. Having a pointless link on that page has never given me any desire to go to hardOCP, nor has accidently clicking the link.
 
Nemesis999 said:
How does a link back to itself make people go to HardOCP? That makes no sense to me. Having a pointless link on that page has never given me any desire to go to hardOCP, nor has accidently clicking the link.
Click to expand...

i guess purely statistically it does, so if someone googles something and the forum activity has gotten more hits (as it likely would because it is updated on a minute by minute basis by reply's to posts)

also a casual passer by could easily end up at the forums and never realize there was a front page with other articles....

it would be nice if the site would direct signed in members directly to the article but redirect anonymous surfers to the OCP article link.....i think this would accomplish the same purpose without being an annoyance for all of the members who are well aware there is a HardOCP front page and know how to get there....LOL

oh well

i want to see this Physx thing working on my G80 cards
 
nobody_here said:
i guess purely statistically it does, so if someone googles something and the forum activity has gotten more hits (as it likely would because it is updated on a minute by minute basis by reply's to posts)

also a casual passer by could easily end up at the forums and never realize there was a front page with other articles....
Click to expand...

It's not the fact that the link on this page goes to hardOCP, it's the fact that the HardOCP page itself links back to itself, for seemingly no reason.
 
Nemesis999 said:
It's not the fact that the link on this page goes to hardOCP, it's the fact that the HardOCP page itself links back to itself, for seemingly no reason.
Click to expand...



oh okay...yeah, i know what you are talking about now.....i was confused...yes, i agree, it is odd
 
You must log in or register to reply here.
Back
Top