The demise of Moore's Law foretold - yet again
© Chemical Heritage Foundation / CC licence / Wikimedia
- Manipulation of components at atomic level will require a new approach
- End of the Age of Silicon in sight?
- IEEE and other industry bodies seem to think so
- Quantum computing back on the agenda – and graphene
Way back in 1965, Gordon Moore, the co-founder of Intel, observed that the number of components that could be etched onto the surface of a silicon chip was doubling at regular intervals of two years. After more observation and experimentation, he concluded that the doubling-on-doubling-on-doubling would continue, if not infinitum, then at least into the foreseeable future - which in those days was at least a decade and longer.
Times have changed beyond belief since then. In 1965 the income of the average American was US$6,450 per annum, gas cost 31 cents a gallon, annual inflation was running at 1.9 per cent and the St. Louis Arch, the "Gateway to the West" was completed. Elsewhere, the Beatles released four LPs in a single year, Australia sent troops to fight with US forces in Vietnam, the Canadians adopted their new "Maple Leaf" flag and, in London, the Post Office Tower opened. Also in 1965, the densest of memory chips could store a mere 1,000 bits of information. Today's chips have 20 billion transistors.
Despite the gloomy prognostication of many a Cassandra ever since Dr. Moore first framed his famous "law" it has been proven correct time-after-time for more than 50 years now. Mark Twain once quipped, "Reports of my death have been greatly exaggerated" and the same applies to Moore's Law. Its demise has frequently been forecast and indeed, back in 2005, Gordon Moore himself admitted that the doubling-up cannot continue indefinitely. He said, "It can't continue forever. The nature of exponentials is that you push them out and eventually disaster happens".
He added that transistors eventually would reach the limits of miniaturisation when they approach the levels of atomic particles. "In terms of size you can see that we're approaching the size of atoms which is a fundamental barrier," he said, "but it'll be two or three generations before we get that far. We have another 10 to 20 years before we reach a fundamental limit."
That was 11 years ago and now, as periodically happens, the limits of Moore's Law are once again under the spotlight - or, more properly perhaps, under the scanning electron microscope.
New semiconductor 'roadmap'
An industry body, the Semiconductor Industry Associations of the United States, Europe, Japan, South Korea and Taiwan is to make one last report based on the International Technology Roadmap for Semiconductors, (ITRS), a dated, nay, venerable chip technology forecasting system. Change is on the way and it will happen because almost every one of the world's major chip makers belong to the ITRS, including IBM, Samsung and Intel itself is a member.
Semiconductor manufacturer are getting close to the point where they will be manipulating components at the atomic level and no-one really knows what will happen then, although some think the Age of Silicon may almost be over.
Thus the IEEE (the much respected Institute of Electrical and Electronics Engineers) is to create a new forecasting system. This one will be a called International Roadmap for Devices and Systems, or IRDS, and it will take into account a much wider range of computer technologies.
Thomas Conte, a renowned computer scientist at the famous Georgia Institute of Technology, and the co-chairman of the group tasked with defining the new set of benchmarks that will replace the old chip reports, say, “The end of Moore’s Law is what led to this. Just relying on the semiconductor industry is no longer enough. We have to shift and punch through some walls and break through some barriers.” Among those barriers are quantum computing and the use of graphene to replace silicon.
Interestingly, although Intel is an important member of the ITRS, it is not participating in that body's final report and maintains that it can see ways to continue to use silicon at the atomic level.
And the company has quite a few theoretical physicists on its side. Some say that the limit of Moore's Law could actually be between 250 and 600 years in the future. Now that's what I call a long-range forecast.