Happy 40th Moore's Law
Posted by Phil Aaronson at 4:38 PM
The Computer History Museum put on, what they called the only west coast celebration of the 40th anniversary of Moore's Law. They celebrated with a "conversation" between Gordon Moore and Carver Mead. It was Mead who coined the phrase "Moore's Law".
There seems to be some confusion about what exactly Moore's law is saying. I was wrong for example. And as it turned out, the law itself has been updated once already. Initially it was simply that the number of transistors that could be fabricated on a chip would double every year. Starting in 1959 with a single transistor. The increasing number of transistors were driven by three primary factors: simply fabricating smaller and smaller devices, reducing defect density and by squeezing out wasted space. But by 1979, while working on some devilishly complex CCD memory chips (I had no idea that CCD devices started out as memory) they had removed virtually all the wasted space. It was then that Moore updated his law to doubling the number of transistors to every two years.
Doubling the number of transistors every two years in effect means that the performance of your computer doubles every 18 months. By some measure of performance. They didn't say how these relate exactly, but it was the 18 month performance number that early Intel marketing pushed. And hence added to the confusion. Doubling first one year, now two years and through marketing, a confusing 18 months.
Carver Mead waxed the more philosophical of the two. He saw Moore's law as one of the few tangible views into the future that we've had. And without that view into the future the semiconductor industry could not have sustained the growth that it has enjoyed.
It was a real pleasure see these two in person. They talked about much more than just Moore's Law. From how Gordon Moore got into chemistry, he's a physical chemist by training: by creating explosives. To why he left Shockley to form Fairchild as part of the original founding eight: Shockley's management style left much to be desired. To an interesting glimpse at a pre-PC Intel casting around for applications that required complex chips in high volume. It was a great evening.
Links:
There seems to be some confusion about what exactly Moore's law is saying. I was wrong for example. And as it turned out, the law itself has been updated once already. Initially it was simply that the number of transistors that could be fabricated on a chip would double every year. Starting in 1959 with a single transistor. The increasing number of transistors were driven by three primary factors: simply fabricating smaller and smaller devices, reducing defect density and by squeezing out wasted space. But by 1979, while working on some devilishly complex CCD memory chips (I had no idea that CCD devices started out as memory) they had removed virtually all the wasted space. It was then that Moore updated his law to doubling the number of transistors to every two years.
Doubling the number of transistors every two years in effect means that the performance of your computer doubles every 18 months. By some measure of performance. They didn't say how these relate exactly, but it was the 18 month performance number that early Intel marketing pushed. And hence added to the confusion. Doubling first one year, now two years and through marketing, a confusing 18 months.
Carver Mead waxed the more philosophical of the two. He saw Moore's law as one of the few tangible views into the future that we've had. And without that view into the future the semiconductor industry could not have sustained the growth that it has enjoyed.
It was a real pleasure see these two in person. They talked about much more than just Moore's Law. From how Gordon Moore got into chemistry, he's a physical chemist by training: by creating explosives. To why he left Shockley to form Fairchild as part of the original founding eight: Shockley's management style left much to be desired. To an interesting glimpse at a pre-PC Intel casting around for applications that required complex chips in high volume. It was a great evening.
Links:
- The 40th Anniversary of Moore's Law with Gordon Moore, Co-founder and Chairman Emeritus, Intel, in conversation with Carver Mead, Chairman and Founder, Foveon.
0 Comments:
Post a Comment
<< Home