Innovation
Rupert Goodwins' diary
Tuesday 03/04/2001It's instructive to remember that bigger is just as important to the chip business as smaller. The whole malarkey revolves around mass production, just like Henry Ford discovered: if you put one slice of silicon through an expensive process and make one chip, you have one very expensive chip.
Tuesday
03/04/2001 It's instructive to remember that bigger is just as important to the chip business as smaller. The whole malarkey revolves around mass production, just like Henry Ford discovered: if you put one slice of silicon through an expensive process and make one chip, you have one very expensive chip. But if you have hundreds of chips on that same wafer of silicon, they cost a hundredth as much for almost exactly the same production cost. So, trick one is to make the individual circuits smaller -- that way you get more per wafer. Simple. Then, and this is something that doesn't make the headlines, you make the wafer of silicon bigger -- same result. Of course, shrinking circuit size also makes your circuits faster and consume less power, while bigger wafers don't do anything but make more chips. Still, Intel's announcement that it's gone from 200mm wafers to 300mm is just as important for future processor and memory technologies. While everyone's cowering in the basement over the stock market collapse it's hard to think ahead, but nonetheless the costs of developing new silicon aren't going to go down any just because the market's gone soggy. Unless efficiency measures -- like larger wafers -- are high on the agenda, the next generation of processors will be too expensive to develop. And then what, eh?
03/04/2001 It's instructive to remember that bigger is just as important to the chip business as smaller. The whole malarkey revolves around mass production, just like Henry Ford discovered: if you put one slice of silicon through an expensive process and make one chip, you have one very expensive chip. But if you have hundreds of chips on that same wafer of silicon, they cost a hundredth as much for almost exactly the same production cost. So, trick one is to make the individual circuits smaller -- that way you get more per wafer. Simple. Then, and this is something that doesn't make the headlines, you make the wafer of silicon bigger -- same result. Of course, shrinking circuit size also makes your circuits faster and consume less power, while bigger wafers don't do anything but make more chips. Still, Intel's announcement that it's gone from 200mm wafers to 300mm is just as important for future processor and memory technologies. While everyone's cowering in the basement over the stock market collapse it's hard to think ahead, but nonetheless the costs of developing new silicon aren't going to go down any just because the market's gone soggy. Unless efficiency measures -- like larger wafers -- are high on the agenda, the next generation of processors will be too expensive to develop. And then what, eh?