Moore's New Law
You have heard this one for years no doubt, but it has made a recent splash in the popular press. Walter Russel Mead is taking a quick look at Moore's Law and thinks that it may be running out of gas. He quotes The New York Times:
The problem is not that they cannot squeeze more transistors onto the chips — they surely can — but instead, like a city that cannot provide electricity for its entire streetlight system, that all those transistors could require too much power to run economically. They could overheat, too. The upshot could be that the gadget-crazy populace, accustomed to a retail drumbeat of breathtaking new products, may have to accept next-generation electronics that are only modestly better than their predecessors, rather than exponentially faster, cheaper and more wondrous.
There is a partial answer out there now. It is the GreenArrays GA144 chip. What is the big innovation in this chip with 144 cores? No clocks. So you have no clock domains to manage. No power domains to manage. What causes the big advantage? When you have nothing to do you do ABSOLUTELY nothing. If you have something to do you do it as fast as you possibly can using as little power as possible. The chip does on the order of 90 BIPS (that is B for billion) at a cost of about 650 mw. Of course that scales down to under 1 mw when everything is idle. With values in between depending on resources used. So we may have a different law to deal with. In fact a different Moore's Law. Chuck Moore's Law (wmv). Yep. It is a Forth chip. Since I last reported on this chip the company has made some progress. They have chips for sale at $20 ea. - minimum 10 pcs. Plus they have a development board the EVB001 [pdf] for $450 ea. If you just want to fool around they also have a chip simulator and development software you can play with for the price of a download. Plus manuals,videos and other help. I've been a Forther since way back (since it came out for the 8080/Z80 - Laxen and Perry IIRC - that would have been back in '78 or '79) so I find this development rather exciting. The GreenArrays people are of course not the only ones thinking along these lines (chips should do nothing unless they have something to do) but as far as I know this is the first commercial execution of such an idea.