Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!cornell!uw-beaver!Teknowledge.COM!polya!Gang-of-Four!andy
From: andy@Gang-of-Four.Stanford.EDU (Andy Freeman)
Newsgroups: comp.arch
Subject: Trends (was Re: Fed up with MIPS)
Message-ID: <12570@polya.Stanford.EDU>
Date: 20 Oct 89 21:40:22 GMT
References: <76700077@p.cs.uiuc.edu> <318@ssp1.idca.tds.philips.nl>
Sender: news@polya.Stanford.EDU
Reply-To: andy@Gang-of-Four.Stanford.EDU (Andy Freeman)
Organization: Computer Science Department, Stanford University
Lines: 30

In article <318@ssp1.idca.tds.philips.nl> roelof@idca.tds.PHILIPS.nl (R. Vuurboom) writes:
[Comparing the PC industry with the auto industry:]
>Theres another analogy with the automobile industry and thats the rate
>at which new models were introduced...in the seventies the rate of
>model introduction was reduced (to me the sign of a mature technology).
>This (I assume) will also occur in the computer industry at a "certain
>point in time".

The 70s trend may have been a reduced introduction rate (and I'm not
convinced it was; complete redesign rate may have gone down, but the
partial revision rate went up), but the 80s trend has been towards an
increased model introduction rate.  The Japanese (especially Honda)
started this with the every three-year complete revisions, and the US
automakers have cut a substantial amount of time out of their redesign
cycle.

It looks to me like the introduction rate increased in the computer
industry in the mid 80s.  Some of it was start-up (the MIPS folk got
the R3000 out the door fairly quickly after the R2000), but even Intel
and Motorola got on the bandwagon.  That's why the killer micro attack
is working.  Cray's cycle time is too long.  (I'm talking about
Seymour.)

It's the old "lots of small improvements" vs "infrequent large
changes" problem.  Both make lots of progress, but at different times.

-andy
UUCP:    {arpa gateways, sun, decwrl, uunet, rutgers}!polya.stanford.edu!andy
ARPA:    andy@polya.stanford.edu
BELLNET: (415) 723-3088