GringsThe same guy from AMD (richard huddy, an ATI man) claims in another article that it was intel buying havok that killed off the idea of gpu physics

www.custompc.co.uk/news/601677/gpu-physics-dead-says-amd.html

Of course, Nvidia and ATI could talk to the other big player in physics, Ageia, about accelerating the PhysX API on GPUs, but Huddy claims that this isn’t really an option. ‘We talked to Ageia,’ admits Huddy, ‘and asked them whether they would like our GPU acceleration in the software case, but it would be somewhat embarrassing for them if we run it much faster, and certainly for the eye candy physics we do.’

Huddy pointed out that there are two main categories of physics in games – eye candy and interactive physics that actually change the way in which the game is played. The latter ‘tends to be better done on the CPU, or Ageia would argue the PPU,’ says Huddy, ‘but the rest can all be turfed off onto the GPU.’ Unsurprisingly, Huddy said that ‘Ageia aren’t really enthusiastic about that, because it does show just how much faster we are then them, and it undermines their message, so I see us having quite major problems in that kind of market dynamic.’

KetxxxAlso, combine that with running a Phenom with an FSB of 333MHz or faster, suddenly seems quite tasty eh? I bet a Phenom with an FSB to match a C2D at a comparable clock to a C2D would see both CPUs neck and neck. People all too easily forget Phenom is running on a lousy 200MHz.