Originally Posted by Heinz68
Well the dedicated PsyX processor is one of the reason why the NVidia PnysX is going to loose to Havok Physics.
The other reason is, it is NV proprietary technology. Does NVIDIA think that AMD and Intel is going help sell more NV hardware?
Anyway this article says it better than I can.
The Inevitable Bleak Outcome of NVIDIA's CUDA + PhysX Strategy
Tuesday, May 12, 2009
There is another on going pool at TechPowerUp
about buying decision and it doesn't even address the question about the need for buying a "dedicated PhysX processor". I think if it was pointed out the numbers would be even lower.
There are many more Titles that use Havok Products
and more Upcoming Releases
And lastly it looks like AMD and Intel will unite against NVIDIA PhysX
that's the thing i was trying to say. YES physx will fail, it's a single vendor solution, that won't win market support.
but, who cares if it fails, the cuda backbone of being able to compute on graphics cards is undoubtedly more powerful and useful than for doing physx calculations alone.
If i'm not mistaken DX11 will have a uniformed api for dealing with physics calculations, so whether it's physx or havok it doesn't matter from DX11 onward, it'll go through the api and card vendors will have to conform to that to be DX11 compliant. But the fact that physx is already there shows that nvidia is ready to make this jump because they've already put a considerable amount of time into engineering their hardware in ways in which it can do these calculations..... don't twist my words though, i'm not saying they're dx11 compliant... but they do know their stuff, think of SLI, they had a good 2 year lead on ati in the multi-gpu sector because of SLI. Crossfire was dodgy, didn't scale well (then again sli wasn't great at scaling either), but it required a horrible dvi-dvi dongle for the 2 cards to communicate, whereas nvidia had already put the time into designing the really elegant (in comparison) sli bridge.
TLDR version: Who cares if it fails or not, physx is a step towrads the end result... which neither uses the current instantiation of physx or havok