Researchers develop carbon nanotube computer

Researchers at Stanford University have built what is claimed to be the first fully-working computer based on a processor built from carbon nanotube transistors, in a move hoped to help string out Moore's Law for a few more decades.

Named for Intel co-founder Gordon Moore, Moore's Law is the observation that the complexity of computer circuits as measured by the number of transistors crammed onto a die roughly doubles every eighteen months. Although originally merely a historical observation, it has proved remarkably accurate in predicting - or, some would argue, defining - the growth of computing.

Now, however, Moore's Law is reaching its limits. Increasing the number of transistors in a given processor requires decreasing their size, and decreasing their size introduces a number of physics-related issues from heat generation to current leakage. As a result, some scientists believe that within the next few processor generations we will hit a hard size limit below which it will be impossible to manufacture a traditional processor.

That's where non-traditional materials come in. A team at the Stanford Robust Systems Group claims to have developed the world's first working computer built using a microprocessor which uses carbon nanotube transistors in place of the more traditional type used in modern computing.

It's a move that others in the industry have been investigating - IBM, in particular, is a proponent of carbon nanotubes as one possible solution to the coming struggles to keep up with Moore's Law - but which has yet to be realised beyond small-scale experimental systems. 'People have been talking about a new era of carbon nanotube electronics moving beyond silicon,'[i] claimed Subhasish Mitra, lead professor on the project and an electrical engineer and computer scientist. '[i]But there have been few demonstrations of complete digital systems using this exciting technology. Here is the proof.'

The computer, admittedly, is basic by modern standards: its processor is built on a process node of one micron - 1,000 nanometres, orders of magnitude larger than today's smallest 14nm parts - and features just 142 of the carbon nanotube transistors. That was enough, however, for the team to implement a small subset of the MIPS instruction set arcitecture and to produce what they claim is a Turing-complete, general-purpose programmable computer system.

The team has already shown the computer is capable of performing useful tasks, albeit slowly: using a customised multi-tasking operating system, the computer has performed counting and number-sorting operations - but the team claimed that the research is more than just the creation of a slow, expensive microcomputer. 'It's not just about the CNT computer,' explained Mitra. 'It's about a change in directions that shows you can build something real using nanotechnologies that move beyond silicon and its cousins.'

The team's work is published in the Nature journal this month.