Thanks for that. I'm really interested in where the fundamental limits are. Already on the machine I'm writing this on (65nm core 2 duo), the transistors are on the order of 100 atoms long, so clearly the limits of 2D transistor density are already closing in rapidly. From there, there's frequency, efficient design (i.e. actual computing power per transistor - multicore design, stream processors etc) and 3D, so there's a lot more to improve.