• 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.

Radeon PhysX GPU Acceleration?

i think it's AMD+ATI+INTEL vs nVidia

AMD's graphics department and CPU department are two totally different things.
 
ATI have their own stream computing thing, it's just that they're not crapping out press-releases about it. This just smells of faked goods, but it is possible - just put PhysX PPU *mulation in OpenCL running on some ATI GPU.

So, PhysX. How many games has it again? How many of them made it into top 10 sales lists and reviews? How many have you played besides the UT3 demo? Any one of them using it to enhance and improve the gaming experience rather than put it into a single demo level? (Except Cellfactor.)
 
They say it isn't running on 48 series cards because they don't have review samples? Ugh. Go buy one.

And on that note, What the heck is with all the bannings? 0_0 Dan_D of ALL people. I'm so lost.
 
This has to be fake. ATI cards dont have CUDA support which is the main driving force behind PhysX.
 
Screw IP and licensing, AMD should just make their own PhysX DLL or w/e that has the same function calls. Kinda like EAX support for non Creative cards years ago..
 
BioSehnsucht said:
Screw IP and licensing, AMD should just make their own PhysX DLL or w/e that has the same function calls. Kinda like EAX support for non Creative cards years ago..
Click to expand...

yeah, thats what they should do, do something to get their pants sued off by NV.....AMD cannot afford that sort of risk, they would lose that fight
 
Afaik, GPU physics operations are done via programmable shaders. Not some special hardware. Both camps have designed GPU's around DirectX and Opengl. That said, physics on GPU is not much more than a shader operation. Sure, Physx as we know it was written for CUDA but thats not to say the final result cant be the same. There could have been some underlying agreement between Ati and Ageia before the purchases. I'm not saying this screenshot is real by any means. But the idea itself is not beyond the realm of possibility. Until I see real results, its nothing more than an interesting idea.
 
kumquatsrus said:
http://forums.vr-zone.com/showthread.php?t=293936

here's the info on that screenshot. seems legit to me.
Click to expand...
CUDA does emit shaders for output. I wonder if NGO captured the shaders and intercept the PhysX calls. It's a pretty limited solution if that's what they did, basically what people complained about: a 3DMark only optimization. Cue crying in 3, 2, 1... wait they were the ones who were crying. :p
 
What cracks me up is the fact that everyone loves to hate synthetic benchmarks but can't seem to stop talking about it when the topic comes up LOL
 
Silus said:
Token number ?! Most of the games we see nowadays are part of the TWIMTBP program and you can be sure that that's the way NVIDIA is going to use to tout PhysX support.

Also, you don't see AMD signing onto a totally NVIDIA controlled PhysX. But does it make sense for them to use Havok, which is part of Intel ?
Click to expand...

Ever heard of the saying "the enemy of my enemy is my friend"?
 
if physiXs is supported via hacked/modified drivers - good luck stopping it once its out & NV may actually allow ATI to use it in order to kill off havok.

this is going to be interesting
 
messerchmidt said:
if physiXs is supported via hacked/modified drivers - good luck stopping it once its out & NV may actually allow ATI to use it in order to kill off havok.

this is going to be interesting
Click to expand...

so where would this leave intel and havok with larrabee?
 
Silus said:
Token number ?! Most of the games we see nowadays are part of the TWIMTBP program and you can be sure that that's the way NVIDIA is going to use to tout PhysX support.

Also, you don't see AMD signing onto a totally NVIDIA controlled PhysX. But does it make sense for them to use Havok, which is part of Intel ?
Click to expand...


Yeah, token number. Many devs would require Nv to damn near pay for the entire implementation of hardware accelerated physX in their games if they were not already on board. Nv has some cash but not enough to do that for every single twimtbp title. That is supposing Ati does not jump on the bandwagon. If they do, we will see more support and sooner.
 
messerchmidt said:
if physiXs is supported via hacked/modified drivers - good luck stopping it once its out & NV may actually allow ATI to use it in order to kill off havok.

this is going to be interesting
Click to expand...

That Press Release (http://www.amd.com/us-en/Corporate/VirtualPressRoom/0,,51_104_543~126548,00.html) posted just 14 Days ago, suggests that ATI will support Havok on their GPUs.

So it looks like Intel and ATI/AMD will go the Havok route to hold back nVidia's Ageia PhysX Tech..
 
Full Text of that Press Release;


AMD and Havok to Optimize Physics for Gaming

AMD and Havok announce plans to optimize physics processing on the AMD platform

SUNNYVALE and SAN FRANSISCO, Calif. -- June 12, 2008 --AMD (NYSE: AMD) and Havok today announced plans to jointly investigate the optimization of physics effects utilizing AMD’s full line of products.
With over 100 developers and 300 leading titles already using Havok’s physics engine - Havok Physics – the company has clearly defined its position as the leading developer of game physics. By working together, both companies are demonstrating their commitment to open standards and continued support for the needs of the game community.
“As the complexity and visual fidelity of video games increases, AMD wants to take advantage of opportunities to improve the game experience,” said Rick Bergman, senior vice president and general manager, Graphics Products Group, AMD. “By working with the clear market leader in physics software, AMD can optimize our platforms to consistently deliver the best possible visual experience to the gamer.”
Havok Physics scales extremely well across the entire family of AMD processors, including quad-core products such as the AMD Phenom™ X4. As part of the collaboration, Havok and AMD plan to further optimize the full range of Havok technologies on AMD x86 superscalar processors. The two companies will also investigate the use of AMD’s massively parallel ATI Radeon GPUs to manage appropriate aspects of physical world simulation in the future.
“The success of Havok as a cross platform software company is predicated on our willingness to listen to the needs of our customers,” said David O’Meara, managing director of Havok. “The feedback that we consistently receive from leading game developers is that core game play simulation should be performed on CPU cores. The clear priority of game developers is performance and scalability on of the CPU. Beyond core simulation, however, the capabilities of massively parallel products offer technical possibilities for computing certain types of simulation. We look forward to working with AMD to explore these possibilities.”
Today’s game experience demands a balanced platform approach, one that combines the right CPU and GPU horsepower. A number of game aspects, including advanced physics processing, are optimized for CPUs, while the latest DirectX 10.1 games demand the latest GPUs. AMD is striving to deliver the best of both worlds, with highly capable CPU and GPU technology that partners like Havok can use as their canvas to deliver the best experience possible.
"Physics is an important part of today's game experience, and in the quest for greater realism the usage and requirements of physics computation is expected to increase,” said Dean McCarron, principal analyst, Mercury Research. New technologies that improve physics simulation will enhance the gaming experience."
About AMD
Advanced Micro Devices (NYSE: AMD) is a leading global provider of innovative processing solutions in the computing, graphics and consumer electronics markets. AMD is dedicated to driving open innovation, choice and industry growth by delivering superior customer-centric solutions that empower consumers and businesses worldwide. For more information, visit http://www.amd.com.
About Havok
Havok, an Intel company, was founded in Dublin, Ireland in 1998, and is the premier provider of interactive software and services for digital media creators in the games and movie industries. With world leading expertise in physics, animation and tools, Havok’s business is to turn our customers’ creative aspirations into technical realities. Havok’s modular suite of tools gives power to the creator, making sure that our clients can reach new standards of realism and interactivity, while mitigating the overall cost and risks associated with creating today’s leading video games and movies.
Havok works in partnership with the world’s best known game developers - including Sony, Nintendo, Microsoft, EA, Ubisoft and Pandemic Studios. Havok’s cross-platform, professionally supported technology is available for the PlayStation®2, PLAYSTATION®3, PSP™, Xbox™, Xbox360™, Wii™, GameCube™, and the PC. Havok’s combination of superior technology and dedication to delivering for our customers every time has led to our technology being used in more than 150 of the world’s best known game titles, including BioShock, Halo 3, MotorStorm, Stranglehold, Crackdown, Age of Empires III and Cars.
Havok products have been used to drive special effects in movies such as Poseidon, The Matrix, Troy, Kingdom of Heaven and Charlie and the Chocolate Factory.
Havok has offices in Dublin, San Francisco, San Antonio, Calcutta, Munich, and Tokyo.
Havok.com, the Havok logo and the buzz-saw logo are trademarks of Havok.com Inc (or its licensors), in the United States and other countries. All other brand names, product names or trademarks belong to their respective holders.
©2008, Advanced Micro Devices, Inc. AMD, the AMD Arrow logo, ATI, the ATI logo, PowerXpress, Radeon, The Ultimate Visual Experience and combinations thereof, are trademarks of Advanced Micro Devices, Inc. Havok.com, the Havok logo and the buzz-saw logo are trademarks of Havok.com Inc (or its licensors), in the United States and other countries. Other names are for informational purposes only and may be trademarks of their respective owners.
Click to expand...
 
There is little difference between this and the whole Crossfire/SLI line of bullshit. There is NO technical reason why SLI or Crossfire wouldn't work on any 2 PCIe (x16 physical connector) motherboard, regardless of WHO made the motherboard.... other than, the deliberate PROHIBITION written into the drivers to dis-allow it.

All politics, no engineering. :eek::rolleyes::p
 
ImperfectLink said:
Afaik, GPU physics operations are done via programmable shaders. Not some special hardware. Both camps have designed GPU's around DirectX and Opengl. That said, physics on GPU is not much more than a shader operation. Sure, Physx as we know it was written for CUDA but thats not to say the final result cant be the same. There could have been some underlying agreement between Ati and Ageia before the purchases. I'm not saying this screenshot is real by any means. But the idea itself is not beyond the realm of possibility. Until I see real results, its nothing more than an interesting idea.
Click to expand...

if that were the case, the PPU cards BFG was trying to pimp would have been able to do video acceleration, but they can't, so yes, it is a specific hardware supported feature
 
nobody_here said:
if that were the case, the PPU cards BFG was trying to pimp would have been able to do video acceleration, but they can't, so yes, it is a specific hardware supported feature
Click to expand...

Apples to oranges, man. The ability to do on GPU physics and other general purpose tasks is a result of programmable shaders, NOT due to nvidia and ATI thinking it would be some cool, random thing to throw on their cards. The "specific hardware supported featue" is the ability to run custom code rather than predetermined routines, which then allows for everything else such as physics, video encoding, etc.. all of which uses the same feature of modern video cards which came as a result of DirectX and OpenGL.
 
kllrnohj said:
Apples to oranges, man. The ability to do on GPU physics and other general purpose tasks is a result of programmable shaders, NOT due to nvidia and ATI thinking it would be some cool, random thing to throw on their cards. The "specific hardware supported featue" is the ability to run custom code rather than predetermined routines, which then allows for everything else such as physics, video encoding, etc.. all of which uses the same feature of modern video cards which came as a result of DirectX and OpenGL.
Click to expand...

Um, what he said... ;-)
 
0mega said:
And on that note, What the heck is with all the bannings? 0_0 Dan_D of ALL people. I'm so lost.
Click to expand...

This same thing has me puzzled. Noticed it while reading this thread and saw like 3 or 4 people banned. Hope it's temporary.
 
PhysX support on ATI cards would be possible if ATI got the license for it. They would just need to do a wrapper that translates the PhysX calls into Brook+ or CAL which can run on their GPU... just like nvidia did theirs that translates the PhysX calls into CUDA.
 
If I remember correctly, didn't Nvidia say something about CUDA going to be compatible with ATI cards as well?

CUDA and Brook+ are both C style languages. It seems like it really wouldn't be that difficult to make either one cross platform.... now performance wise, I bet that you would need GPU specific optimizations (different libraries, etc.) to be able to get good speed on different GPUs.
 
nobody_here said:
if that were the case, the PPU cards BFG was trying to pimp would have been able to do video acceleration, but they can't, so yes, it is a specific hardware supported feature
Click to expand...

Actually, there was a Cuda-like 'GPPPU' programming library for the PhysX chip. It was very similar to a GPU in terms of programmable stream processors. The main difference was its fast processor-to-processor communication at the time. But with the G80 series and Cuda, nVidia introduced that aswell by using a large shared cache for all processors.

The G80 shaders aren't as specific as one might think. They can do a lot more than just graphics, and Cuda allows you to use them while bypassing the fixed graphics-related hardware, so you are free of the old "create vertexbuffer, render to texture" approach of GPGPU. It's now programmable much like a CPU, with very flexible memory access.
So no, there is no specific hardware. The unified shaders on G80 and above are flexible and programmable enough to perform a variety of tasks, including things like physics and video encoding.
More or less the same goes for AMD, although they designed their shaders in a different way.
 
This is all interesting, but exactly WHY do we want physics processing on our graphics cards? In terms of gaming, I thought the GPU is still the bottle neck. Wouldn't it be better to use spare CPU cycles to do this processing? (quads are pretty cheap)

Has anyone actually seen gaming benchmarks comparing frame rates with and without PhysX enabled on an NV card? I'd be very interested to know what the frame rate hit is.

Perhaps there interesting uses for PhysX outside of gaming where the GPU has some spare cycles and I just am not aware of them.
 
[H]ocusPocus;1032677960 said:
This is all interesting, but exactly WHY do we want physics processing on our graphics cards? In terms of gaming, I thought the GPU is still the bottle neck.
Click to expand...
Because, aside from a few exceptions, physics is limited in gaming by CPU speed. Read up on the old Ageia PhysX (p)reviews, but ignore the hype. Interaction and realistic physics effects could improve immensely. Is it worth trading off a few frames per second for better physics? I think so.
 
According to this over at pcper http://www.pcper.com/article.php?aid=584 the only poblem with phisX on amd cards is THE FOOLS DO NOT KNOW HOW TO COMMUNICATE WITH EACH OTHER

proof to quote josh "So I am here to put, in writing, a message from NVIDIA to AMD. Richard Huddy, please call Roy Taylor at NVIDIA ASAP. He is awaiting your call. "

maybe they can put there heads together / work threw this issue in therapy!
 
pxc said:
Is it worth trading off a few frames per second for better physics? I think so.
Click to expand...

If its just a few fps that would be good, but are there any reviews out there yet showing actual numbers? This should just work with existing titles that already have PhysX support, correct?
 
maxius said:
According to this over at pcper http://www.pcper.com/article.php?aid=584 the only poblem with phisX on amd cards is THE FOOLS DO NOT KNOW HOW TO COMMUNICATE WITH EACH OTHER

proof to quote josh "So I am here to put, in writing, a message from NVIDIA to AMD. Richard Huddy, please call Roy Taylor at NVIDIA ASAP. He is awaiting your call. "

maybe they can put there heads together / work threw this issue in therapy!
Click to expand...

actually this part
Do users sit for hours on end just watching the benchmark run over and over again?
Click to expand...

is true for some people, i used to be one, would spend all night tweaking, tuning, overclocking, doing volt mods, whatever and running benchmarks all night long.......so yes, there are people like that Mark Rein :p
 
[H]ocusPocus;1032678594 said:
If its just a few fps that would be good, but are there any reviews out there yet showing actual numbers? This should just work with existing titles that already have PhysX support, correct?
Click to expand...

well, a good way to see is to try the physx exclusive ut3 maps with pre-physx drivers vs physx enabled drivers and compare fps performance. i would try, but i don't have the game to test. it's only one game, but it should give you an idea of what can be achieved.

http://www.hardforum.com/showthread.php?t=1317359

here are the drivers, with some already testing.

keep in mind that only some titles currently have access to hardware acceleration while others will be supported in software via cpu. though, more hardware accelerated physx gams are expected in the near future.
 
if i could get the damn drivers to work with my 8800's i would try as i have UT3....but from what i have read the modded .inf trick only works on G92 and up GPU's, not G80's, at least as of right now
 
nobody_here said:
if i could get the damn drivers to work with my 8800's i would try as i have UT3....but from what i have read the modded .inf trick only works on G92 and up GPU's, not G80's, at least as of right now
Click to expand...

i guess you'll have to wait till july when the official g80/g92 physx supported drivers roll out
 
I'm giving this a bump. The drivers may be early but techgage did the kind of test we all wanted to see - UT3s PhysX level - the other games are not yet supported in Nvidias drivers which is a bad sign; they need to be added manually, there's no global "enable physx gpu ppu".

(9800 GTX)
http://techgage.com/article/nvidias_physx_performance_and_status_report/3

The TL;DR is that the GPU takes the expected impact from running PhysX, on a higher resolution (2560x1600) the FPS drops as if you were running software accelerated PhysX, while it is a lot better off at 1680x1050 - even though the FPS basically drops from 60 to 50 at that resolution.
 
MrLonghair said:
The TL;DR is that the GPU takes the expected impact from running PhysX, on a higher resolution (2560x1600) the FPS drops as if you were running software accelerated PhysX, while it is a lot better off at 1680x1050 - even though the FPS basically drops from 60 to 50 at that resolution.
Click to expand...

Most important thing is that with CPU physics you're completely CPU-limited.
The framerate at 1680x1050 is the same as at 2560x1600 with the CPU, always around 30 fps.
The GPU delivers 50 fps at 1680x1050, so you gain 20 fps over the CPU physics, which is an improvement in framerate of about 67%.... So even assuming perfect linear scaling, the CPU would require an additional 3 cores to deliver that kind of performance. In practice you probably won't even get this kind of performance out of a system with two quadcores.

So, assuming that most people game at resolutions lower than 2560x1600, I think GPU physics will be a nice improvement... especially for those that don't have a fast quadcore processor (my E6600 dualcore doesn't even get playable framerates in those PhysX levels at all).
Also, I think that as GPUs evolve and the physics acceleration matures, the difference between CPU and GPU only gets larger (note also that this is 'only' a 9800GTX, a 9800GX2 or GTX260/280 will deliver far more performance).
 
Thanks MrLonghair, that's the kind of testing I was curious about. I found the cpu usage numbers a bit odd as enabling physics on the cpu had very little impact on average cpu load.
 
You must log in or register to reply here.
Back
Top