Moore’s Law

by Dan Murray

Published February 21, 2001



Faster, smaller and more powerful computing devices generate heat and consume too much energy. But research and innovative discoveries keep pace, and the rate of advancement continues to accelerate.

A prophetic observation in 1965 by Intel co-founder Gordon Moore, been dubbed Moore’s Law. Moore accurately stated that the number of transistors on a chip, and thus the performance, will double every 18 to 24 months.

Whenever insurmountable barriers have stalled progress, a resolution, breakthrough or quantum leap has paved the way. Technological advances in the next six years are projected to exceed those of the past 60 years. Where else in the economy have product values increased as prices are continually lowered?

More transistors amounts to more power consumption and thus heat, the enemy to reliable electronic constancy. Chip manufacturers’ problems and rewards escalate. Circuit complexity are inherently unending. But solutions abound.

Asynchronous Digital Design (ADD) in Pasadena, CA is pursuing a new high-performance microprocessor design that doesn’t require a clock. The radical business endeavor, based on a decade of research from the California Institute of Technology, could produce semiconductors that have the same or better performance of today’s advanced microprocessor chips, but consume less energy.

Url Cummings and Andrew Lines, ADD’s co-founders and classmates at Caltech, are pressing for speed enhancements at reduced heat and power consumption. It’s called the asynchronous chip.

Micro-processors today use clocks to regulate their internal timing. Clock’s are measured in hundreds of Megahertz (millions of clock cycles per second). Our modern synchronous computer chips are like constantly humming motors doing only a little work once in a while.

Not only do typical computing processors “run in place,” but all subparts are constricted to be in (lock) step with the clock. The overall process is, by design, restricted to it’s slowest stage. For this and other reasons, not all 400 MHz computers operate the same, and comparing processors by clock speed is erroneous.

Transistors in asynchronous chips can theoretically switch on and off, as needed, and function independently. The subparts optimally work together without the constraint of being in step with a master clock. Cummings & Lines expect their first chip will perform comparably to a traditional processor operating at 1,000 MHz.

There you have it: no marching in step at blistering clock speeds, conserving energy and reducing heat without a timekeeper. The asynchronous model sounds like an ideal working environment for people too, doesn’t it?

Peter Coffee of PC Week wrote this analogy of synchronous logic: “Imagine working in an office where everyone picks up the phone every 60 seconds to see if anyone is calling.” Dick Pountain of Byte Magazine described asynchronous logic as, “A taxi service where cabs depart, not at fixed times, but only when they’re carrying passengers.”

Potential uses for the low-power chip will certainly include mobile communications, portable hand-held computers, and Net-connected devices.

For years, computer chip manufacturers have successively reduced the size of their microscopic components and their relative distance to one another. Electrons moving along a labyrinth of tiny copper and gold lines reach their destinations sooner if the inner-distances are shorter.

At the present pace of development, according to Moore’s Law, miniaturization will reach the molecular level in about 12 more years. Meanwhile, the optical lithography process has just about reached its limit. The very high frequencies of light are simply too course to etch thinner circuit paths onto a silicon wafer’s surface.

IBM and Intel invent new processes to focus light “tighter.” One method, called Extreme Ultraviolet Lithography, concentrates invisible light through specially shaped mirrors. A chip’s circuit paths are then focused onto a wafer through refined lenses.

Lucent Technologies’ Bell Labs is pursuing multiple streams of electron beams to clarify a finer etching, like a minuscule pencil writing one line at a time. X-rays, which are still shorter wavelengths, are producing progressively finer circuit patterns.

In other areas of research, scientists are concerned for the rough road over which these electronics must pass. At higher computing speeds, electrons have the effect of riding on the surface of the metal, and ridges are like mountains. New methods of refining a smoother silicon surface improve the performance of electron movement.

As the physics of size, heat and higher computing performance are resolved, other new and interesting possibilities emerge. Hydrogen and chlorine may replace silicon. The control of spinning atoms may replace transistors. It’s an interesting time to be alive.