[RC5] Re: rc5 Digest, Vol 6, Issue 1
Timothy.Marsh at usm.edu
Wed Dec 10 12:39:01 EST 2003
> Date: Sun, 30 Nov 2003 03:41:11 -0500
> From: aSe <aSe at SysFail.com>
> Subject: Re: [RC5] 0.060 completion of rc5-72
> To: "D.net Discussion" <rc5 at lists.distributed.net>
> At 02:06 PM 11/30/2003 +0800, you wrote:
> >On Saturday, Nov 29, 2003, at 19:24 Asia/Hong_Kong, Ñåðãåé wrote:
> >>Okey, at current speed it would be nearly 1200 years to complete, but...
> >>Moore's law
> >>More than 25 years ago, when Intel was developing the first
> >>company cofounder Gordon Moore predicted that the number of transistors
> >>microprocessor (and computing power) would double approximately every 18
> >>months. To date, Moore's law has proven remarkably accurate.
> >I'm not so sure about the 'number of transistors' bit though (25 years is
> >about 17 18-month periods, and if transistors on a chip is, say, 10
> >million, that would mean under 100 transistors when Moore made such a
> >I don't think you can do much with 100 transistors.
> A good link FYI..
> -aSe [aSe at SysFail.com]
It's true that you can't do much with 100 transistors. In fact, reading the
link above and the original paper, you find that there were onyl about 50
components per square inch of chip. That's why personal computers as we
know them didn't really exist yet. Moore's observation was made in 1965 and
he predicted that it would hold true for about 10 years, and it did. From
1975 on the law has been modified to 18 months (from 12) for doubling. That
adjusted law is what has held up till now. Some say it is actually a
self-fulfilling law since researchers use it as a goal. Of course, the law
is based on the current type of technology, silicon based semi-conductor
chips. When we reach the limits of this technology, or we change
technologies, then the rate of increase would probably be completely
different (hopefully even faster).
More information about the rc5