Moore's Law is dead. Long live Moore's Law

16 June 2009

Len Jelinek, director and chief analyst at iSuppli, has stuck his neck out and called the end of Moore's Law as an economic driver in 2014. He hasn't said development has to stop technologically, more that the costs will not outweigh the advantages of going further than the 20nm or 18nm node.

He is at odds with the official Intel position. Leading Intel technologist Kelin Kuhn was in London to talk about some of the potential roadblocks on the way to 15nm and beyond. And she was bullish about the future, even though it relies on some very big changes to the way chips are made with the next five years.

And you don't have to wait long for someone to tell you that, in the 1980s, a bunch of researchers reckoned 1µm was the limit. The story is relayed in much the same way history teachers tell of Victorian fears that riding in cars at more than 30mph would tear people's heads off. So, has Jelinek got it wrong?

OK. So the first thing to get out of the way is Moore's Law: It's not really a law is it? Not like Newton's laws of motion or the laws of thermodynamics in which bad things happen if they don't work. But it has proved to be a very good observation of an industry that has been good for about 40 years. It's actually an observation on what price elasticity and technological development can do for an industry.

I think it should actually be called the Moore-Noyce Law because it was Bob Noyce, Moore's colleague at Fairchild and then Intel, who came up with the pricing model that meant Moore's Law became the key to predicting a market sector driven by deflation.

You might be thinking: who cares? So what if it stops? The world certainly won't stop but long-held assumptions about the development of computing and the development of a technology-driven society do rest on the idea that there is this constant upgrade cycle in action.

Second, Moore's Law isn't really very detailed. When Moore plotted some graphs and extrapolated in the mid-1960s, there wasn't any such thing as a 'process node'. There was no International Technology Roadmap for Semiconductors (ITRS) to tell you what the half-pitch measurement for the first metal layer was expected to be from year to year. Moore explained later that the actual reduction in size of the transistors and circuits was responsible for just one-third of the increase in chip function per dollar every two years. (It started off as a doubling every year but this soon levelled out to two by the time Moore gave his more detailed analysis of the graph he drew in the mid-1970s). The other two-thirds came from an increase in chip size and improvements in design techniques. Although the electronics industry was one of the first to employ computers for design, most layout was done by hand on sheets of plastic even ten years into the Moore's Law period.

What has happened since is that the one-third down to shrinkage in two dimensions now accounts for the bulk of the biennial improvement in density. So, it's easy to equate Moore's Law with process technology. But there is no reason for things to stay that way. The original graph only plots two things: 'number of components per function' versus time. There is no declaration of how any of that is actually to be achieved.

Personally, I reckon, and I'm not alone by a long shot, unless there's some kind of miracle in extreme ultraviolet lithography or someone comes up with some crazy way of continuing to use 193nm lithography, there's going to be a bit of a hiatus between 22nm and 18nm or 15nm.

The question is: will anybody outside the industry actually notice. Because there are few tricks the chipmakers can pull to honour the promise of Moore's Law but which utterly break the connection between process node and the trend that has driven chipmaking since the mid-1960s. The lead feature in the upcoming E&T deals with part of that. There is also one from a few weeks back that goes into one of the ways out. But I'll add the links in, as well as some of the iSuppli charts as they do explain a lot, and follow-up later. I wanted to get this post out of the way quickly.

2 Comments

...or Moore's Comfort Blanket?

Never could understand the almost anally-retentive and seemingly slavish desire to adhere to Moore's Prediction. It seems to me to have been fudged and bent several times to keep it on track.

Moore's 'Law' could almost be seen as a comfort blanket - something to hold onto for reassurance as the tech spirals wildly upwards!

Oh - and who would I side with in a fist fight betwen an analyst and Intel? Dur!!

It's not a comfort blanket - if it were, it would be a very expensive comfort blanket. I don't think anyone has made the case for whether it is cause or effect, but the general opinion is that the doubling every two years is a byproduct of the economics that were, effectively, introduced when Bob Noyce gave the early IC customers in the military an offer they couldn't refuse.

Intel vs analyst? Well...But it's not just iSuppli. There is a growing feeling among chipmakers that 22nm or 20nm is not necessarily a hard limit, but it could turn out to be a bump in the road. The question is whether the general public will notice as the industry swerves off in a different direction.