I spend time at a couple of different universities quite frequently. The CS departments are focusing and still trying to get a grasp on Quantum Computing. Not sure if that is where things are going though, it's just my anecdotal observations

http://plato.stanford.edu/entries/qt-quantcomp/

Quantum computing.

PP is not the future of gaming. There is too much interdependance of between aspects of games that PP can't be utilized on a large scale effectively for games. Forcing PP where it is not appropriate will actually constrain efficiency, not increase it.

Say you put each object's AI on a separate thread, you will still need to wait until the rendering thread finishes before you can proceed to the next frame. PP in games is only as fast as its slowest thread.

@Crash Yeah, look at Killzone and the 60% usage deal. If that is coming from 6 SPEs then look @ the redundancy from 180 cores. Not saying that they couldn;t get more out of KZ2 (it is still early in the PS3's life) but it goes to show that extra is not always going to equal 100% once done.

Then again, graphics is example of opposite. It's very easy to use PP in graphics and because of that some GPU's on PC have hundreds of cores (Nvidia GTX 280 has 280 cores for example). Even $100 GPU's on PC has dozens of cores.

1) software engineers know where they can add more things to optimize

2) games are insane to optimize well for any processor because you have to account for the worst situation, not the normal situation.  that means that all games on any cpu are effectively inefficient.

3) multicore software design is in its infancy.  they will get better at it.

4) multicore processors will get better.  the hardware designers will learn to increase the efficiency of their designs.

M$ and IBM are developing tools to increase the efficiency and effectiveness of creating apps designed for multiple cores/threads.

perpendicular processing of course!

Well an increase in threads per core is one idea, i know Intel is returning Hyper Threading to the Core duo line.

quantum processing, but not for years to decades... there is a problem with parallel processing, as the number of cores increases, the efficiency of each cores communication decreases. at a point, with adding more cores will not give a gain in speed even with a higher mhz boost, the increase will cause a loss in net power... where this is in terms of practicality has not been found yet in practically produced units. this could vary wildly between type of processor in use...(x86-ppc).

all of this is part of the reason ibm/sony worked on the cell with asymmetrical processing to see if they could improve on performance until the time that quantum processors are ready... this is a big gamble, and they need to convince people like MS, and apple to code for asymmetrical cell cpus