Carbon nanotube computer hints at silicon's super-low-power successor

Swapping carbon for silicon is still a long way off, though, and processor makers have plenty of tricks left

Researchers at Stanford University have built the first working computer out of carbon nanotube transistors, a technology widely bruited as a replacement for silicon in microprocessors.

But the technology is a long way from being practical, and the short-term work being done with reduced-power processing and system-on-chip technology using conventional silicon will remain mainstays.

Named "Cedric," the carbon-nanotube-built computer is extremely primitive by today's standards -- a single-bit processor that runs at 1kHz and sports a mere 178 transistors. But making it work represents a triumph for the scientists of the Stanford team, who overcame a number of previous technical stumbling blocks, including how to make carbon tubes grow in parallel lines and how to disable malformed tubes that work like conductors instead of switchable transistors.

Another key success -- and one vital to making the process commercially viable -- is how the computer was created with many of the same processes used to create existing silicon. Many of the tools and techniques currently in use to create processors could be retooled for carbon.

Silicon has been widely regarded as being near the end of its useful life as a material for processors, since engineers are rapidly approaching the lower boundary for how small a transistor can be made with silicon. Carbon offers a way forward, since it's theoretically possible to create carbon-nanotube transistors that are an order of magnitude more energy efficient and a good deal smaller than their silicon ancestors.

But practical carbon-processor fabrication is still at least decades away by any measure. Consequently, processor makers right now are focusing on research to squeeze as much life out of silicon as possible.

As reducing transistor sizes becomes increasingly difficult, more efficient instruction sets and integrated system-on-chip designs that inherently use less power have come to the fore, since those approaches solve that many more problems at once in terms of system design.

The most familiar arena for this kind of of work is mobile processors. Apple's recent 64-bit A7 ARM processor, for instance, sports more than double the performance of its predecessor but, according to Apple, uses even less power. Servers are the next big deployment target for ARM processors as well; the theory is that many low-power ARM processors can outperform the wattage equivalent of current x86 processors. (This makes a bit more of Microsoft's committing to ARM on Windows easier to understand.)

In short, by the time carbon transistor technology becomes commercially viable, the overall way silicon is deployed in the context of a whole system -- not just as a transistor technology -- may well have gone through that many more revolutions all by itself.

As far as post-silicon research goes, IBM has long been doing remarkable work in this area. It has experimented with strongly correlated materials -- that is, metal oxides --and has built low-power transistors that use graphene, although those seem best confined right now to analog applications like signal processing.

This story, "Carbon nanotube computer hints at silicon's super-low-power successor," was originally published at Get the first word on what the important tech news really means with the InfoWorld Tech Watch blog. For the latest developments in business technology news, follow on Twitter.

Copyright © 2013 IDG Communications, Inc.