by Anton Shilov
08/29/2013 | 11:50 PM
From time to time various experts come up with predictions regarding the end of Moore’s law, economical and technological viability to double the number of transistors per chip every two years. Some companies believe that economical efficiencies of Moore’s law are going to drastically decrease on 20nm node. But a DARPA expert believes that the law will lose its feasibility in 2020 – 2022, or at 5nm or 7nm nodes.
“For planning horizons, I pick 2020 as the earliest date we could call it dead. You could talk me into 2022, but whether it will come at 7 or 5nm, it's a big deal," said Robert Colwell, director of the microsystems group at the Defense Advanced Research Projects Agency (DARPA), who previously served as the chief IA-32 architect on the Pentium Pro, Pentium II, Pentium III, and Pentium 4 microprocessors at Intel, during the Hot Chips conference, reports EETimes web-site.
Exponentials, such as Moore’s law, always come to an end by the very nature of their unsustainably heady growth. Nonetheless, following Moore’s law microprocessor performance has increased by several thousand times over forty years thanks to upsurge of frequencies from 1MHz to over 4GHz as well as increase of the amount of cores from one to eight and more.
The former IA-32 chief architect points to slow progress of central processing units’ performance in the recent years and notes that even that progress costs a lot of money. Therefore, the rapid skyrocket of the microprocessor industry is nearing its end due to economic reasons.
"When Moore's Law stops it will be economics that stops it, not physics, so keep your eye on the money. […] I do not expect to see another 3500x [performance] increase in electronics – maybe 50x in the next 30 years. I do not think the world's going to give us a lot of extra money for 10% [annual] benefit increases," Mr. Colwell told an audience of processor designers.
DARPA has a list of around 30 possible alternatives to the CMOS technology that has been the corner stone of Moore's law. DARPA's microsystems group has "a fair amount of money chasing" two programs. One is exploring approximate computing in a program called Upside; another is exploring the effects of spin-torque oscillators to settle on partial solutions at relatively low power.
It is noteworthy that Mr. Colwell does not believe in technologies like 3D stacking, new micro-architectures and apps, new switching technologies, better human interfaces and so on.