Moore's Law, 40 years old this month
Posted 15 April 2005 - 05:06 PM
Considering that the rule has largely proved true, it means that Intel could fit 2,300 transistors on a silicon chip in 1971 and later this year will unveil a chip with 1.7 billion transistors. It's why your kid today can have an Emoto-tronic Furby that contains more computing power than NASA's Lunar Lander circa 1970.
But Moore's Law began as a back-of-a-napkin extrapolation — a quick guess about the future based on recent events, such as your Dad's saying on a family car vacation that since you'd gotten to Cleveland in two hours, you'd make Milwaukee by nightfall.
Gordon Moore has said he was so embarrassed that people had dubbed his prognostication Moore's Law that he couldn't bring himself to utter that phrase until the mid-1970s.
The world learned of Moore's Law in a seemingly unremarkable article that Moore wrote for the April 19, 1965, issue of Electronics — a magazine read primarily by men who wore crew cuts, glasses with black plastic frames and short-sleeved white shirts, and who worked as engineers at places like General Dynamics.
Among the other articles in that issue was one headlined: "Using cold cathode tubes to count and store." To celebrate what the magazine dubbed as the 35th anniversary of electronics, it asked five people from the industry, including Moore, to write about the future of electronics.
A Motorola executive wrote a piece called "New electronic products will make life easier." A guy from NASA wrote about space. Henri Busignies of ITT tossed out the idea of fiber optics in his contribution, "More and faster communications: Phone vision may be carried on laser beams." (Busignies apparently missed the prediction that ITT wouldn't stick around long enough to be a part of it.)
Amid all that, Moore's article seems downright nerdy. It's oh-so-alluring title: "Cramming more components onto integrated circuits."
He starts out by predicting that the future of electronics is tied to the future of integrated circuits — which seems like a "duh!" point now but in 1965 actually wasn't so obvious. Integrated circuits, which would grow up to be microprocessors such as the Pentium, were still pretty new and unproven.
In the magazine's credits, Moore — then at Fairchild Semiconductor, which he helped start before moving on to co-found Intel — was labeled "one of the new breed of electronic engineers."
Moore continued: "Integrated circuits will lead to such wonders as home computers — or at least terminals connected to a central computer — automated controls for automobiles, and personal portable communications equipment."
Not bad for 1965. He basically caught PCs and cell phones.
Anyway, not until paragraph 16 does Moore veer into the Moore's Law-ish stuff. He writes about the pace of improvements in integrated circuits to that point and notes that power, for the same cost, had roughly doubled every year.
"Certainly over the short term, this rate can be expected to continue, if not to increase," Moore wrote. "Over the longer term, the rate of increase is a bit more uncertain, although there is no reason to believe it will not remain nearly constant for at least 10 years."
And that's it. That's what became Moore's Law. The only difference is that in 1965, he said power would double every year. In 1975, Moore updated it to say power would double every two years.
In a 1997 interview, he told me, "I never said 18 months." Even though that's how everyone quotes Moore's Law these days.
So Moore's Law wasn't some scientific formula worked out on a white board or a deep insight that arose from years of study. Moore just did a little extrapolation for an article about the future that he probably thought would be forgotten within two weeks.
What's funny, too, is that it's not a law at all — not like the law of gravity, which describes something that is resolutely true. Chips don't have to double in power at a certain pace. But since it's so far been possible to improve chips at that pace, the industry has embraced Moore's Law as the de facto standard for how fast the industry should move.
I've heard lots of descriptions of the role of Moore's Law. It's the industry's metronome, some say. Or it's like the guy in the galley on those ancient ships in Charlton Heston movies who beats a drum to set the rowing pace for the slaves who man the oars. As a metaphor for the tech industry, it's not bad: If you don't keep up, you get whipped.
Or maybe Moore is like John O'Sullivan, the editor who in the 1840s first published the term "Manifest Destiny" to describe the USA's ultimate goal of stretching from coast to coast. It was just a few words in a story, but it took on a life of its own, becoming a rallying cry and a presumed outcome.
Strange how modestly these things can begin.
0 user(s) are reading this topic
0 members, 0 guests, 0 anonymous users