home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.arch
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!moe.ksu.ksu.edu!ux1.cso.uiuc.edu!news.cso.uiuc.edu!uxa.cso.uiuc.edu!j-grout
- From: j-grout@uxa.cso.uiuc.edu (grout john robert)
- Subject: Re: A Little History
- References: <1478@seqp4.sequoia.com> <1992Aug21.195858.24945@sctc.com>
- Message-ID: <BtCtLM.41F@news.cso.uiuc.edu>
- Sender: usenet@news.cso.uiuc.edu (Net Noise owner)
- Reply-To: j-grout@uiuc.edu
- Organization: University of Illinois at Urbana
- Date: Fri, 21 Aug 1992 22:10:33 GMT
- Lines: 31
-
- smith@sctc.com (Rick Smith) writes:
-
- >While it's nice to have history provide the personal touch
- >(associating a person with a concept) history is especially
- >significant in computer architecture because many of the concepts
- >don't really become obsolete. It's amazing how much of it is
- >independent of the implementation technology. In other words, things
- >that worked in tube machines are still being used in VLSI. Sometimes
- >the only published example of a technique is a historic example, since
- >today's vendors don't always describe things in as much detail as we
- >might need. In other cases, "everybody" knows about various classic
- >examples (Illiac IV for multiprocessing, Cray for vector registers,
- >B5000 for "thunks") and the peculiarities of the implementation point
- >the rest of us in particular directions as a result.
-
- >What goes around comes around. So it's good to know what's coming.
-
- In "What's All This about Gallium Arsenide", a presentation Seymour
- Cray made to Supercomputing (in Orlando, FL), he talked about how he
- learned to build reliable computers out of unreliable semiconductors
- in the 1950's (use diode-transistor logic)... then startled his
- audience by telling him that he reused that knowledge in the Cray-3
- (because gallium arsenide is unreliable than more conventional
- semiconductors)... everyone laughed when they realized just _why_
- he'd been telling us about these experiences of his from the 1950's.
-
- --
- John R. Grout
- University of Illinois, Urbana-Champaign
- Center for Supercomputing Research and Development
-
-