I don't have an in depth knowledge of the subject but one obvious question from my point of view is why can't physics calculations be offloaded to a thread running on 1/2 cores or 2/4 cores while the other cores handle any pre-GPU Geometry processing or other rendering tasks? Maybe a simplistic question, so perhaps someone has a simple answer. I'm just finding it difficult to understand why all the GFX engine makers can't devise methods to make better use of true multi/core threading.

Futher to that, with all multi core becoming common place, I don't really see the benefits of dedicated hardware physics processors. If you take a Havoc based game or something like Flatout 2 for example where there are 1000's of objects being thrown around, and todays CPU's seem quite capable of handling that without much of a impact on frame rate. Physics Processors just seem like overkill a bit.