Originally Posted by Prox
Rockwell will be 16nm... you really think heat will be that much of a problem by then? No.
heat will still be a problem... its safe to say that about 60-70% of heat is from the switching (other losses in gate/oxide, and such) so if your packing in more transistors and switching them faster you will always get more heat. the only advantage is with a smaller transistor (which the posted nm is not always the ful size of the transistor, usualy just the gate size) there is less capacitance so they just need to drive it less, but the switching loss will still be large.
The transition to smaller transistors does not get rid of the large heat factor.
the AMD X2 2.9GHz was a TDP of 65 Watts Thats two cores, assume 32.5 Watts per core on a 65nm tech. so lets compare it to an equal CPU of new tech.
Athlon II X2 270 - 45nm 3.4GHz but same TPD 65 Watts.
so with a 30% smaller transistor and a 14.5% increase of speed was obtained keeping the TDP equal.
Right now AMD X6 1075T is a 3.5GHz with 45nm tech and 95W TDP.
using equal convention of the above,
if this went to 19nm tech, that is 57% smaller Transistor, it can be assumed if the above is true about a 28% increase in speed. to keep the same TDP. 4.45GHz.
Thats if and when we gets to 19nm, in most cases I think a different architecture will be built. using Moore's Law, every two years the Number of transistors will double in the same area, we can assume that with an incease of over 50% in size to go from 45nm to 19nm (we have to remember production not a single lab making 1 chip every 4 months). will take about 3-4 years if not a little longer for it to be in the hands of the consumers. so say we will have 19nm dies in 6 years... I hope we have moved past DDR4 well before then...