Quote Originally Posted by Final8ty View Post
It would likely increase sales for them.
But they will still have to live with that people will think the CPU path is deliberately performance crippled because of what they have seen with Havok.
Quote Originally Posted by RaZz! View Post
but think about it again. there must be a reason why nvidia isn't doing exactly that. maybe it's because physx wouldn't be that much of a performance hit in most games at all if it's integrated properly?
don't get me wrong, i do think computing stuff like physics on a gpu will result in better performance, no matter what, but on the other hand i think a quadcore where a core is dedicated for physics calculation could work out very well as well. maybe that's exactly what nvidia fears and why they locked down physx for their gpus only.
Perhaps, but that's the thing. Havok doesn't do these things and it's the premier CPU physics library. All the examples of "impressive" Havok usage are simple rigid body effects that PhysX does on the CPU as well. So what are they afraid of? If Havok were to come out with a fast CPU solution for particle systems and cloth and fluid simulation they would have to beef up PhysX on the CPU anyway to compete. I can understand why they don't want to waste time optimizing for the CPU while there's no competition, but I don't see why they don't just let it run slowly instead of disabling it outright.