🚀 go-pugleaf

RetroBBS NetNews Server

Inspired by RockSolid Light RIP Retro Guy

Article View: alt.folklore.computers
Article #121711

Re: GE 6xx (was Re: Honeywell 200)

#121711
From: mul... Richard Shetron
Date: Wed, 10 Dec 1997 00:00
43 lines
2832 bytes
In article <66cb1d$6...@rosebud.sdsc.edu>,
Tom Perrine <t...@galt.sdsc.edu> wrote:
[nsip]

>I was also extremely fortunate to serve a long internship under John
>Wertz, was one of the primary designers of the architecture of the
>instructions to be provided by "EIS box" (pronounced ICE box), which
>was the primary difference between the (original) 600 and 6000 series.
>
>As I recall, the EIS box was an option on later 600 models, the 6000
>had the functionality of the EIS box incorporated into the overall
>design of it CPU.

I think the later GE 6xx models may have actually been renamed to the
HIS 6xxx series.  The 655 (?) was the first 6xx series to use integrated
circuits, the 645 and earlier were discrete component or very small scale
integration.  I would not be surprised to find that the HIS 6180 was the
GE 655 with a new model number and name.  EIS was an option on the 6xxx
as far as I know, but was required for Multics, optional for GCOS.

RADC had 2 6180's, one without cache ran GCOS and the one with cache
ran Multics.  The Multics 6180 CPU failed and they tried to switch
to the GCOS 6180 and found the EIS hardware needed so much repair it
was quicker to repair the Multics 6180 and fix the GCOS 6180 later.
The maxload went from 36 without cache to 42 with cache.

>The EIS was the Extended Instruction Set, which had all kinds of
>complex instructions for supporting things that COBOL and FORTRAN
>liked to do.  I seem to remember one insttuction (moved under edit
>with translation), which would move a value from one place to another,
>convert formats (say from floating point or BCD to ASCII characters,
>with a COBOL-like editting field to put in $-signs, commas, and
>decomal points).  One seriously studly instruction :-), but it made
>the COBOL benchmarks *scream*, much to the chagrin of CDC and
>Burroughs sales people :-)
>Tom E. Perrine (t...@SDSC.EDU) | San Diego Supercomputer Center 

The EIS was real wierd.  Several instructions had both 2 and 3 operand
modes:  a=a <op> b and c = a <op> b.  the weird part was a, b, and c
could all be different data types and the appropriate conversions would
be done as needed.  The EIS string instructions oculd operation on up
to a segment (just under 1MB) of data at a time.
Technical Details
Message-ID: <348e065a.0@news.wizvax.net>
Path: mbox-mo1.pugleaf.net!unknown!not-for-mail
References: <65g3r9$8gg@bgtnsc03.worldnet.att.net> <34860CF2.86D1A0B2@monmouth.com> <bertschc-0512972318300001@ip-23-174.phx.primenet.com> <66cb1d$6sr@rosebud.sdsc.edu>

Show Full Headers
Complete Article Headers
From: mul...@wizvax.wizvax.net (Richard Shetron)
Subject: Re: GE 6xx (was Re: Honeywell 200)
Date: 1997/12/10
Message-ID: <348e065a.0@news.wizvax.net>
X-Deja-AN: 296794514
References: <65g3r9$8gg@bgtnsc03.worldnet.att.net> <34860CF2.86D1A0B2@monmouth.com> <bertschc-0512972318300001@ip-23-174.phx.primenet.com> <66cb1d$6sr@rosebud.sdsc.edu>
Organization: Wizvax Communications, Troy, NY. USA
Newsgroups: alt.folklore.computers
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit