Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1a 12/4/83; site rlgvax.UUCP
Path: utzoo!watmath!clyde!akgua!mcnc!decvax!harpo!ulysses!mhuxl!houxm!hogpc!houti!ariel!vax135!floyd!cmcl2!seismo!rlgvax!guy
From: guy@rlgvax.UUCP
Newsgroups: net.unix-wizards
Subject: Re: 1620 lore
Message-ID: <1912@rlgvax.UUCP>
Date: Sat, 12-May-84 02:03:47 EDT
Article-I.D.: rlgvax.1912
Posted: Sat May 12 02:03:47 1984
Date-Received: Sun, 13-May-84 10:13:37 EDT
References: <514@sri-arpa.UUCP>
Organization: CCI Office Systems Group, Reston, VA
Lines: 38

> The description referred to as a description of STRETCH sounded more like
> a description of the IBM 7080 commercial processor...a very fast, expensive
> implementation more like the 1401/1410 than the 1620.  Stretch was the 7030.
> For details, see Bucholz's book "Design of a Computer System".  A lot of
> good ideas from Stretch appeared later in the /360 line (and a lot of good
> ideas didn't).

If you're referring to

> The IBM STRETCH (7030) ca 1957, was bit addressable. The instruction
> contained a word address (64 bit word) , bit offset, and a "byte length",
> or operand size. The machine was a masterpiece of baroque architecture:
> all sorts of nifty things that were practical to use only in assembler.
> The term "byte" was coined for the STRETCH, and originally ment a variable
> size operand.

the STRETCH did look like that.  It even had, if I remember correctly,
BCD add and subtract instructions.  It was, however, *primarily* a binary,
single-accumulator, architecture.  It had a 64-bit word, but fixed-point
arithmetic was done on bit fields of arbitrary length starting on an arbitrary
bit boundary (it was a fully *bit*-addressible machine) and was done an eight-
bit *byte* at a time; floating point arithmetic (which was supposed to be quite
fast for the time) dealt with 64-bit floating point numbers (possibly 32-bit
also, I don't remember) and was a good deal faster than fixed-point arithmetic.

It did not, however, resemble the 14xx series or the 1620 very much.

It *did* have one of the more interesting ideas for a multi-user system.
Since a lot of programming at the time was done on a single-user system
using the control panel (lights, switches, corsets, bustles - sorry, got
carried away a bit there :-)), they decided that the right way to do a
multi-programming system was to have the front panel be purely an I/O
device and have the driver for it permit a user to manipulate their process
using this software-driven control panel!  I have no idea whether the ultimate
OS for this beast actually did this or not.

	Guy Harris
	{seismo,ihnp4,allegra}!rlgvax!guy