Comments on: The end of Moore’s Law http://timworstall.com/2009/07/30/the-end-of-moores-law/ It is all obvious or trivial except... Fri, 25 May 2012 14:11:27 +0000 hourly 1 http://wordpress.org/?v=3.3.1 By: Rob Fisher http://timworstall.com/2009/07/30/the-end-of-moores-law/comment-page-1/#comment-32942 Rob Fisher Thu, 30 Jul 2009 16:22:22 +0000 http://timworstall.com/?p=8994#comment-32942 I think Moore's law is quite specifically about the density of transistors. Those who anticipate its imminent end often fail to predict new techniques. Ray Kurzweil points out that Moore's law is part of a bigger picture of increasing computing power, traces exponential growth back to mechanical computers, and considers new technologies like vacuum tubes and silicone transistors the paradigm shifts that broke the limits of their preceding technologies. So I suppose Moore's law might come to an end but don't see any reason for computers in general to stop getting more powerful. As for costs and benefits, I can *imagine* infinite benefits -- like being able to live forever in a universe of your own creation -- from theoretically possible computers. So I don't see an endpoint where everyone is content with what they have. What might cause trouble are what might be called local minima in the cost/benefit curve between here and there. Sort of like space travel not catching on because although you *could* make lots of money mining the asteroids, intermediate technology you need to develop first isn't cost effective. I'd like to think that development would just slow down -- space flight continues to advance, even if slowly -- but I suppose it's possible we could get stuck in a cost benefit minima, which would be disappointing. I think Moore’s law is quite specifically about the density of transistors. Those who anticipate its imminent end often fail to predict new techniques.

Ray Kurzweil points out that Moore’s law is part of a bigger picture of increasing computing power, traces exponential growth back to mechanical computers, and considers new technologies like vacuum tubes and silicone transistors the paradigm shifts that broke the limits of their preceding technologies.

So I suppose Moore’s law might come to an end but don’t see any reason for computers in general to stop getting more powerful.

As for costs and benefits, I can *imagine* infinite benefits — like being able to live forever in a universe of your own creation — from theoretically possible computers. So I don’t see an endpoint where everyone is content with what they have.

What might cause trouble are what might be called local minima in the cost/benefit curve between here and there. Sort of like space travel not catching on because although you *could* make lots of money mining the asteroids, intermediate technology you need to develop first isn’t cost effective.

I’d like to think that development would just slow down — space flight continues to advance, even if slowly — but I suppose it’s possible we could get stuck in a cost benefit minima, which would be disappointing.

]]> By: AntiCitizenOne http://timworstall.com/2009/07/30/the-end-of-moores-law/comment-page-1/#comment-32927 AntiCitizenOne Thu, 30 Jul 2009 12:34:13 +0000 http://timworstall.com/?p=8994#comment-32927 Moore's law has been over-achieved in GPU technology. http://www.nvidia.com/object/product_tesla_s1070_us.html Moore’s law has been over-achieved in GPU technology.

http://www.nvidia.com/object/product_tesla_s1070_us.html

]]> By: Kay Tie http://timworstall.com/2009/07/30/the-end-of-moores-law/comment-page-1/#comment-32911 Kay Tie Thu, 30 Jul 2009 08:42:10 +0000 http://timworstall.com/?p=8994#comment-32911 What affects costs most is yield: the number of live chips to dead ones. This is dependent on the size of the chip and the size of the wafer: a flaw in the silicon on a big wafer of small chips knocks out only one of many. A flaw on a small wafer knocks proportionally more. The size of the chip is, of course, dictated by the size and number of transistors. What affects costs most is yield: the number of live chips to dead ones. This is dependent on the size of the chip and the size of the wafer: a flaw in the silicon on a big wafer of small chips knocks out only one of many. A flaw on a small wafer knocks proportionally more. The size of the chip is, of course, dictated by the size and number of transistors.

]]>