Originally Posted by SIrPauly
The way I did take that information is PhysX is scalable, meaning if an end-user has more PhysX hardware, the engine, if coded, may offer many more PhysX effect capabilities for the higher-end sku's than the lower-end Sku's.
So, if a gamer had Tri-Sli, one may have more PhysX effects than a single GPU. If one had a 240 core product used as a discrete PhysX GPU one may have more PhysX effects than a 128 core discrete PhysX GPU. Not taking this information as right now and in everything PhysX but eventually in some titles or engines moving forward.
It makes a lot of sense to be scalable considering the PhysX hardware may differ now and in the future. What about future nVidia hardware? Many more cores as well. Why should all PhysX owners have the same effects when some hardware may be much more powerful?
That's a fair point. I was thinking in terms of physX scaling out from single GPU to multiples, which would be very difficult IMO (because of the lack of shared fast access storage).
But yeah, considering a physX implementation that scales out from a 96SP card to a 240SP card is a good thought.