4.- juanflaiter: A good portion of a video card’s (or in your case a physics card) price is the onboard memory. Could it be possible to implement shared memory architecture with either the main memory or GPU memory?
MH: Although technically feasible, the benefit is not meaningful either in cost or performance. For the gamer who wants the best system, limiting GPU memory or stealing main memory just means the system isn’t the highest level system. Also, shared memory means that bandwidth is shared and will most likely lower performance. This tradeoff might be good for a mainstream product down the road, but not a performance product.
This is an interesting interview, although nothing much has changed since our interview with them back in July. It’s really great to see AGEIA’s PhysX bing used in a lot more places than just games though, such as 3D Studio Max. Check out the full interview at Overclockers.cl.