Cramming more components onto integrated circuits
Great link. It shows that Moore projecting his law from 1965 only to 1970 - and here we are able to confirm his projection has been valid for, not merely half a decade, but going on half a century.IMHO Moore's Law does not actually define the rate at which improved IC circuits are developed, but rather how fast they are demanded. That is, the fact that increased production of a good has, historically, always produced a concomitant reduction in cost for not just transistors but for all goods. The reason that the development of IC improvements has been as rapid as Moore predicted (for much longer than he actually initially predicted) is simply that two years from now the market will absorb twice as many transistors as now, at half the price per transistor as today. Doubling the production of a physical thing of constant size (a car, for example) would depress the price by much more than a factor of two. But two years from now I will be ready to listen if you offer to double my RAM for the same money as my original quantity of RAM.Even at this late date there are sure to be billions of people who don't have computers, so when you drive down the price with increased production you still have a huge market to tap. As it is, I think many of those people are at least getting cell phones . . .
And also note that the speed of a processor doesn’t scale with the number of components. The doubling of components doesn’t mean that you get a doubling of speed.