Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10 5/3/83; site umcp-cs.UUCP
Path: utzoo!watmath!clyde!floyd!harpo!seismo!rlgvax!cvl!umcp-cs!zben
From: zben@umcp-cs.UUCP
Newsgroups: net.arch
Subject: Univac 1100 - Ones complement subtractive adder
Message-ID: <4939@umcp-cs.UUCP>
Date: Wed, 1-Feb-84 03:37:48 EST
Article-I.D.: umcp-cs.4939
Posted: Wed Feb  1 03:37:48 1984
Date-Received: Tue, 7-Feb-84 09:20:31 EST
Organization: Univ. of Maryland, Computer Science Dept.
Lines: 17

> No, a one's complement subtractive adder is *not* a snake - there
was one of 'em in the heart of the Univac 1108 that was my first and
still most favorite computer (the maintenance on a 10+ year old machine
was killing us so we scrapped it - sniff - I have one of the register
boards in my desk drawer).  Seems like normal hardware subtracts by
complementing the subtrahend and adding.  Well, this brain damaged
hardware added by complementing one augend and subtracting!  I think
it ran marginally faster, but they might have been designing around a
patent or something.  In any case, it seemed to generate minus zeroes
in a lot more cases than just (0)+(-0); you could generate a minus zero
using normal operands.  The Fortran compiler generated really contorted
code to handle the case.  I'm not sure if the current machines still have
this problem, but that might explain the note from the Univac Unix person.

Ben Cranston          ...seismo!umcp-cs!zben         zben@umd2.ARPA
                        (running 4.1 BSD)        (Univac 1100/80 but *not*
                                                  running Unix (yet?) )