Parallel processing is a somewhat old idea by now, still much need to be enhanced yet before the cores get near perfect scalable performance. The video card industry got there first and are leading knowhow in that area, AMD is yet to show all the benefits from getting ATI technology but much of it is already there and it ranges from the chipset itself as the miniaturization that allowed to be more at par with Intel at the raw core speed competition.
The future points to it, a no far away someday we'll be seeing dozen cores CPUs, should not be surprise the count would be by hundreds... tight interaction with the videocard, with full duplex communication allows for self adapting physics emulation with the CPU getting feedback of 'debris' to adjust response, and so on.
That is where I think the next real step leads.