art@cs.bu.edu (Al Thompson) (02/07/90)
In article <1461@east.East.Sun.COM> sgolson@pyrite.East.Sun.COM (Steve Golson) writes: |(originally posted to comp.lsi, with no replies) | |When did Gordon Moore first postulate Moore's Law? Quoting from |"Computer Structures" by Siewiorek, Bell, and Newell p. 64: | | In 1964, Gorden [sic] E. Moore, then director of research at | Fairchild Semiconductor, predicted that the component count per | IC chip would double every year. | |Unfortunately there is no reference. Anyone have a definitive answer? Ah, but there is a reference. I've been meaning to post this for a while, but just now got around to it. The following is quoted from: T.R. Reid, 1984, "The Chip", Simon and Schuster, New York. pp 123-124. "Noyce's friend and colleague, Gordon Moore, was asked in 1964-when the most advanced chips contained about 60 components-to predict how far the industry would advance in the next decade. 'I did it sort tongue in cheek,' Moore recalled later. 'I just noticed that the number of transistors on a chip had doubled for each of the last three years, so I said the rate would continue.' To his dismay, that off-the-cuff prediction was widely quoted and soon came to be known as 'Moore's law'. To his astonishment, the law held up well into the 1970's. 'At that time, I had no idea that anybody would expect us to keep doubling [capacity] for ten more years. If you extrapolated out to 1975, that would mean we'd have 65,000 transistors on a single integrated circuit. It just seemed ridiculous.' By 1975 the industry was producing a new series of memory chips that contained 65,536 tansistors. 'It amazes me,' Moore said recently. 'I still have a tough time believing we can make these things.'"
gillies@p.cs.uiuc.edu (02/07/90)
That's amusing. If Gordon Moore hadn't made such an outlandish prediction, maybe we *wouldn't* have seen the number of transistors double every year for a decade or moore (sic).
henry@utzoo.uucp (Henry Spencer) (02/07/90)
In article <51751@bu.edu.bu.edu> art@cs.bu.edu (Al Thompson) writes: >'I had no idea that anybody would expect us to keep doubling [capacity] for >ten more years. If you extrapolated out to 1975, that would mean we'd >have 65,000 transistors on a single integrated circuit. It just seemed >ridiculous.'... There is actually a still more impressive observation, somewhat related: the number of transistors on Earth has been doubling every 10 months for the last 30 years. The slowdown in chip-density growth has *not* slowed this growth down, last I heard. -- SVR4: every feature you ever | Henry Spencer at U of Toronto Zoology wanted, and plenty you didn't.| uunet!attcan!utzoo!henry henry@zoo.toronto.edu
gerry@zds-ux.UUCP (Gerry Gleason) (02/07/90)
In article <1990Feb7.001316.28775@utzoo.uucp> henry@utzoo.uucp (Henry Spencer) writes: >There is actually a still more impressive observation, somewhat related: >the number of transistors on Earth has been doubling every 10 months for >the last 30 years. The slowdown in chip-density growth has *not* slowed >this growth down, last I heard. Did you come up with this "rule" yourself, or get it from someone else? A quick calculation shows that 30yrs/10mo is about 36 doublings or about 64 billion times the number in 1960. That might work out to be reasonable, but what are the endpoint values, so I can do a quick check? And of course, who is estimating these values and what is the uncertainty? Gerry Gleason
nelson@m.cs.uiuc.edu (02/08/90)
Something here seems a bit flaky... If, in fact, that doubling every 10 months rule holds back to 1960, and we assume that there was only ONE transistor then, we have 20 trans. for every person on Earth. Now, if we say that there was only ONE Burroughs 5000 (introduced in 1960), then it follows that each person has the equivalent of 20 of those machines. This is still not reasonable, though. I think a figure of about TEN MILLION transistors in 1960 is reasonable. That equates to 200 million per person on Earth now, and that seems much too large...
henry@utzoo.uucp (Henry Spencer) (02/09/90)
In article <161@zds-ux.UUCP> gerry@zds-ux.UUCP (Gerry Gleason) writes: >>There is actually a still more impressive observation, somewhat related: >>the number of transistors on Earth has been doubling every 10 months for >>the last 30 years... > >Did you come up with this "rule" yourself, or get it from someone else? It's from somewhere else, but I'm afraid I can't cite a specific reference. It was an impressive number (although as others have pointed out, it seems a little bit too quick to be true), and the graph of transistor population (on a log scale) vs. time was impressively linear. Unfortunately, I read an awful lot of technical literature -- my mail is measured in kilograms per week -- and this was a year or two ago, so I have no idea where I saw it. Sorry. -- SVR4: every feature you ever | Henry Spencer at U of Toronto Zoology wanted, and plenty you didn't.| uunet!attcan!utzoo!henry henry@zoo.toronto.edu
sgolson@pyrite.East.Sun.COM (Steve Golson) (02/09/90)
In article <1990Feb7.001316.28775@utzoo.uucp> henry@utzoo.uucp (Henry Spencer) writes: >There is actually a still more impressive observation, somewhat related: >the number of transistors on Earth has been doubling every 10 months for >the last 30 years. The slowdown in chip-density growth has *not* slowed >this growth down, last I heard. Notes from a lecture by Gordon Moore, 19 September 1989, MIT VLSI Seminar: "VLSI Industry Trends, Technological and Economic" The number of transistors produced per year increases 10x every 3 years, or about 2x every year. Some numbers: 1968 - 10^9 transistors produced 1978 - 10^12 1988 - 10^15 This used to be difficult to count; now all you have to do is count how many DRAM bits were shipped and multiply by 2. Another way of looking at today's number: that is 2x10^6 transistors for every person in the developed world. Have YOU bought your 2M transistors this year? Your quota for 1995 will be 100M... Another way of looking at this: in 1990 we will construct more transistors than have ever existed before this year. Steve Golson sgolson@East.sun.com golson@cup.portal.com Trilobyte Systems -- 33 Sunset Road -- Carlisle MA 01741 -- 508/369-9669 (consultant for, but not employed by, Sun Microsystems) "As the people here grow colder, I turn to my computer..." -- Kate Bush
pkr@maddog.sgi.com (Phil Ronzone) (02/09/90)
In article <3300099@m.cs.uiuc.edu> nelson@m.cs.uiuc.edu writes: > >Something here seems a bit flaky... If, in fact, that doubling every 10 > months rule holds back to 1960, and we assume that there was only ONE > transistor then, we have 20 trans. for every person on Earth. Now, if > we say that there was only ONE Burroughs 5000 (introduced in 1960), then > it follows that each person has the equivalent of 20 of those machines. > This is still not reasonable, though. I think a figure of about > TEN MILLION transistors in 1960 is reasonable. That equates to 200 > million per person on Earth now, and that seems much too large... I though so at first, but then: I just read about ONE company with ONE fab line makeing ~800,000 1MB DRAMs a month. That's 800 BILLION transistors per month. Of course, I am assumimg that "transistor" is not restricted to discrete component transitors ... ------Me and my dyslexic keyboard---------------------------------------------- Phil Ronzone Manager Secure UNIX pkr@sgi.COM {decwrl,sun}!sgi!pkr Silicon Graphics, Inc. "I never vote, it only encourages 'em ..." -----In honor of Minas, no spell checker was run on this posting---------------
crisp@mips.COM (Richard Crisp) (02/09/90)
In article <3793@odin.SGI.COM> pkr@maddog.sgi.com (Phil Ronzone) writes: > >I though so at first, but then: > >I just read about ONE company with ONE fab line makeing ~800,000 >1MB DRAMs a month. That's 800 BILLION transistors per month. > Here's the numbers I saw recently: Toshiba was producing 8 million 1meg DRAMS/month, Hitachi 3.5 million/month etc. -- Making the world safe for technocracy.
art@cs.bu.edu (Al Thompson) (02/09/90)
In article <3793@odin.SGI.COM> pkr@maddog.sgi.com (Phil Ronzone) writes: |In article <3300099@m.cs.uiuc.edu> nelson@m.cs.uiuc.edu writes: |> |>Something here seems a bit flaky... If, in fact, that doubling every 10 |> months rule holds back to 1960, and we assume that there was only ONE |> transistor then, we have 20 trans. for every person on Earth. Now, if |> we say that there was only ONE Burroughs 5000 (introduced in 1960), then |> it follows that each person has the equivalent of 20 of those machines. |> This is still not reasonable, though. I think a figure of about |> TEN MILLION transistors in 1960 is reasonable. That equates to 200 |> million per person on Earth now, and that seems much too large... | |I though so at first, but then: | |I just read about ONE company with ONE fab line makeing ~800,000 |1MB DRAMs a month. That's 800 BILLION transistors per month. | |Of course, I am assumimg that "transistor" is not restricted to |discrete component transitors ... I've been following this with some interest. It seems that some people are flabbergasted at the idea of hundreds of billions of transistors. Well, given today's feature sizes numbers like that are not astounding, let alone surprising. Let's err on the high side and say gates are 2u long. That means that five gates can be fitted across one diameter of a human red blood cell (approx dia=10u). Now, how many red cells do you suppose you carry around with you? An adult male has about 5*10^6 red cells per cubic MILLImeter of blood. There are a million cubic millimeters in a liter. So, we have about 5*10^12 red cells per liter. In these terms, hundreds of billions of transistors is nothing to write home about.
johnl@esegue.segue.boston.ma.us (John R. Levine) (02/09/90)
In article <3300099@m.cs.uiuc.edu> nelson@m.cs.uiuc.edu writes: > >Something here seems a bit flaky... If, in fact, that doubling every 10 > months rule holds back to 1960, and we assume that there was only ONE > transistor then, we have 20 trans. for every person on Earth. ... There was one transistor in 1947. By 1960 there were many transistorized computers as well as all sorts of non-digital equipment like portable radios. As noted elsewhere, RAM makers are churning out billions of transistors per month these days, and as they move to 4MB and 16MB chips, the rate will only increase. I bought about 20 million transistors last year (8MB of parity RAM and other random stuff, I bet there's another million in my CD player), how about you? -- John R. Levine, Segue Software, POB 349, Cambridge MA 02238, +1 617 864 9650 johnl@esegue.segue.boston.ma.us, {ima|lotus|spdcc}!esegue!johnl "Now, we are all jelly doughnuts."
gillies@p.cs.uiuc.edu (02/09/90)
Re: 10^15 transistors produced in 1990.
A more interesting number would be the number of working transistors
thrown away. We probably throw away a trillion working transistors
every year, just because a few thousand or hundred thousand
transistors or connection lines are broken.
What a waste. I think we should use lasers to cut out & recycle
these working transistors. 8-)
Don Gillies, Dept. of Computer Science, University of Illinois
1304 W. Springfield, Urbana, Ill 61801
ARPA: gillies@cs.uiuc.edu UUCP: {uunet,harvard}!uiucdcs!gillies