Moore's Law: at least 10 to 20 years before the limit is reached

Paul Boughton

April saw the 40th anniversary of the publication of the article that became the basis for Moore's law, which looked at the period of 10 years from 1965 to 1975 and came to the conclusion that the density of components on a piece of silicon was doubling every year.

In 1975, he updated his prediction to once every two years. While originally intended as a rule of thumb in 1965, it has become the guiding principle for the industry to deliver ever-more-powerful semiconductor chips at proportionate decreases in cost.

Gordon E Moore is currently Chairman Emeritus of Intel Corporation. After working at Fairchild with Robert Noyce and Andrew Grove, Moore co-founded Intel in 1968, serving initially as Executive Vice President. He became President and Chief Executive Officer in 1975 and held that post until elected Chairman and Chief Executive Officer in 1979. He remained CEO until 1987, when he was replaced by Grove, and was named Chairman Emeritus in 1997.

Moore earned a BSc in Chemistry from the University of California at Berkeley and a PhD in Chemistry and Physics from the California Institute of Technology. He was born in San Francisco, California, on 3rd January,1929.

He is a director of Gilead Sciences Inc, a member of the National Academy of Engineering, and a Fellow of the IEEE. Moore also serves on the Board of Trustees of the California Institute of Technology. He received the National Medal of Technology from US President George Bush in 1990.

EEE: How did Moore's Law come about?

GM: 1965 was the very early days of integrated circuits (ICs) which typically had 30 components and I could see in the lab we had something with 60 components would be launched in the next years I looked back and saw that we had pretty much doubled the components every year so I took this an extrapolated for the next 10 years, to 60 000. That was a pretty wild extrapolation but it got the idea across, I hope, that making ICs would lead to cheaper components and electronics. I didn't expect it to be so precise. At the end of 1975 I went back to look at what happened and why and there were three components: more density, bigger chips and squeezing the waste space out of the chip. As we had squeezed all the spare space out by that stage, from 1975 it was doubling every couple of years rather than every year.

EEE: How long will Moore's Law continue to be valid?

GM: Something like this can't continue for ever. If you extrapolate too far you always end in disaster, and we are approaching the size of atoms, and that's a pretty clear limitation. But I can see the next two to three generations of technology will be likely to proceed so we have 10to20years before we reach a fundamental limit, and even then that's not the end of the progress as by then engineers will have a budget of literally millions of transistors on a chip for their designs.

EEE: What is the real Moore's Law?

GM: A lot of people have applied Moore's Law to anything that grows exponentially in the industry. The alaw' of computing power increasing by a factor of two every 18 months probably came from Dave House when he worked at Intel so I think he deserves the credit for that.

EEE: What is the potential for technologies such as nanotechnology?

GM: I'm a sceptic on that. The IC technology is the result of the cumulative investment in research and development of over $100bn and something to replace it has to spring almost full blown. Now, nanotechnology has many applications but I'm sceptical it will replace ICs in the mainstream industry. When you look at it, we are already operating well below 100nm so standard silicon technology has become nanotechnology in that respect but the idea of building up something atom by atom comes from a different direction.
Rather than replace ICs we are seeing something else happening as the technology of ICs is being used in other field such as gene chips and biochips, we have micromachined MEMS devices used in projection TVs and airbags and increasingly there's microfluidics with chemistry labs on a chip.

EEE: If Moore's Law is good for the consumer, do you replace your PC every 18months?

GM: I use a Centrino PC, and Intel's replacement cycle is two to three years, and I'm on that. In the original paper I did talk about the potential for cheaper computers in the home, but as time went on I forgot about it. When I as CEO I remember one engineer coming in and describing a small computer for the home but the only application we could think of for it was in the kitchen for storing recipes, and I didn't think that was a very powerful application.

EEE: With computing performance growing so fast, does the software hold us back?

GM: People can't develop and debug software properly until they have the hardware so software does tend to lag the hardware. I certainly want a simple interface although I don't know what it looks like. I think we are losing ground a bit in general purpose computing -- its not through a lack of effort but they want to add so many new features its hard to simplify it, but it works pretty well.

EEE: Will computer processing power ever approach that of the human brain?

GM: I think computers are going in the wrong direction to do that. They were developed to solve problem in a particular way and all the attempts at artificial intelligence have fallen short. I believe that to really get human intelligence you have to go back and look at how the brain works and take a completely different approach. But if you take a particular part of human intelligence and keep pushing computers you could probably do it -- one area that intrigues me is good language recognition for example, being able to distinguish between atoo' and atwo' by using the context. Once the computer recognises language you can start to have an intelligent conversation with your computer and that will dramatically change the way computer are used.

EEE: Can we get close to 100percent yields?

GM: We get amazingly close to 100percent yields today. In my office at Intel I have Intel's first 100percent yield wafer. That was a tremendous problem in the early days but its pretty tractable now

EEE: What about power consumption?

GM: Intel has been very concerned with power particularly in mobile systems but the desktops also need a lot of power. Our chief technology officer made an extrapolation that if we keep gong the way we are, the energy density on a chip would be greater than he surface of the sun, so we clearly had to do something. That's the reason there's all the emphasis on dual and multiple cores.