Nvidia Mulls Over Porting PhysX to OpenCL

[Slaymate]

It's not necessarily that simple. ATI's and Nvidia's architectures are radically different and what works well on one doesn't necessarily work well on the other.

It is that simple. ATI could work with the F@H Team to develop a ATI Client but they don't, or they won't.

Any gpu, whether it's from ATI, Nvidia, or Intel is capable of supporting physics processing provided the physics engine is implemented using a vendor agnostic api like OpenCl or DirectX compute. The problem is no companies have done that, Nvidia didn't want anyone else to be able to use Physx so they implemented it using a proprietary api, CUDA, that can only run on their cards and then they crippled the cpu version so they could use Physx as a marketing tool. There was a rumor that Havoc was going to be released for OpenCL but when Intel canceled larrabee they no longer had a incentive to move forward with it.

So it's Nvidia and Intel's fault that ATI won't develop it's own Physics Processing Software. Why should Nvidia or Intel be responsible for enabling features for ATI.

Nvidia's dominance of the gpgpu market is little more than marketing. Also for the record I have heard that ATI has the more mature OpenCL compiler.

Nvidia's dominance of the gpgpu market is a direct result of trying, something ATI doesn't want to do. A more mature OpenCL complier. When you have less features to code for it's easy to say more mature, ATI saves more money by not offering advanced software research and development. That's why I haven't bought an ATI product since the X800XT.

 

[patonb]

It's not ATI or FaHs fault. Its not getting all the right "other" programs together to allow fah to build a great program.

Also, until the new smp was released, all programs were writen by volunteers, so yha they weren't programed the greatest.. But the new programs are by actual paid programmers.

FaH would love to see ATI perform as well, why would they want to slow half their doners? Plus i think theyre getting tired of the winers.

 

[zmatt]

It is that simple. ATI could work with the F@H Team to develop a ATI Client but they don't, or they won't.

Now I know you don't know what you are talking about. programming on GPU's isn't easy at all, but I'm sure you and your CS degree know that already. ATI was on version 2.0 of its OpenCL SDK when Nvidia was on beta. You know, i remember some time ago when F@H made a different branch for ATI cards and they beat Nvidia ones, a lot of people who invested a lot in nvidia cards got ticked off so they pulled it. Not only that but F@H isn't the be all end all of GPGPU, far from it. You can show in other cases where ATI out performs Nvidia in GPGPU. You cannot take one metric (whose objectivity has been brought into question) and paint it as the case for everything. That is called the fallacy of hasty generalization.

So it's Nvidia and Intel's fault that ATI won't develop it's own Physics Processing Software. Why should Nvidia or Intel be responsible for enabling features for ATI.

ATI demonstrated over a year ago Havok running accelerated on their cards.

Nvidia's dominance of the gpgpu market is a direct result of trying, something ATI doesn't want to do. A more mature OpenCL complier. When you have less features to code for it's easy to say more mature, ATI saves more money by not offering advanced software research and development. That's why I haven't bought an ATI product since the X800XT.

HAHAHAHAHAHAHAHAHA
If only you knew what Nvidia's "trying" got people in a compiler. Nvidia does not have a lead in GPGPU. Physx is poorly adopted and on the verge of dying, ATI has a stronger presence in the TOP500 super computers (a metric that matters) and a far easier to use environment.

 

[S0ULphIRE]

I'd respect what you're saying and maybe you'd actually convince me, but you're coming off like an angry fanboy. Tone it down, fix up your spelling, eat a cookie.

 

[zmatt]

What spelling mistakes? I spell check my posts.

Sorry if I am coming off as the F word, but I used to think I knew what I was talking about. Now I know for certain what I am talking about and I have very little tolerance for anything else.

 

[S0ULphIRE]

dieing? ATI has a 'stringer' presence? Nvidia 'does nto'? lol, you really got into a frenzy towards the end

 

[Puddle Jumper]

Yeah it did get a little bit rough there at the end.

Regardless of our preferences on the issue I think we have to take Zmatt's statements as fact since the rest of us don't have any experience with gpgpu development, at least not that I'm aware of.

 

[zmatt]

dieing? ATI has a 'stringer' presence? Nvidia 'does nto'? lol, you really got into a frenzy towards the end

Odd, spell check didn't pick up any of those. I guess it goes to show not to rely too much on technology.

 

[Slaymate]

What I see is Nvidia hating ATI fanboyism.

The fact is ATI has never had good software support for any of their products. You can make all the wild claims you want and continue to support their lack of effort by buying their products. In return you will continue to be second fiddle, good luck with that.

 

[Puddle Jumper]

What I see is Nvidia hating ATI fanboyism.

The fact is ATI has never had good software support for any of their products. You can make all the wild claims you want and continue to support their lack of effort by buying their products. In return you will continue to be second fiddle, good luck with that.

So the fact that he actually has experience with gpgpu development is completely irrelevant to a discussion about gpgpu?