Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.3 4.3bsd-beta 6/6/85; site ut-sally.UUCP
Path: utzoo!utcs!mnetor!lsuc!pesnta!pyramid!ut-sally!crandell
From: crandell@ut-sally.UUCP (Jim Crandell)
Newsgroups: net.arch
Subject: Re: Right shift vs. divide
Message-ID: <4027@ut-sally.UUCP>
Date: Tue, 21-Jan-86 19:46:34 EST
Article-I.D.: ut-sally.4027
Posted: Tue Jan 21 19:46:34 1986
Date-Received: Wed, 22-Jan-86 01:47:11 EST
References: <3000002@convexs> <4782@alice.UUCP>
Reply-To: crandell@sally.UUCP (Jim Crandell)
Organization: U. Texas CS Dept., Austin, Texas
Lines: 15

>> arithmetic right shifting is equivalent(numerically) to division if the
>> algorithm is implemented correctly.  just because most machines do it
>> wrong doesn't mean it doesn't work.
>>
>> if you want, try the infamous right shift all 1's (-1).
>> 
>Still more evidence that the problem is harder than it looks:
>the above algorithm fails if the dividend is the most negative number....

Seymour Cray would really get a bang out of this discussion.  Anybody want
to send him a copy of it?
-- 

    Jim Crandell, C. S. Dept., The University of Texas at Austin
               {ihnp4,seismo,ctvax}!ut-sally!crandell