Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site mmintl.UUCP
Path: utzoo!watmath!clyde!bonnie!akgua!whuxlm!harpo!decvax!linus!philabs!pwa-b!mmintl!franka
From: franka@mmintl.UUCP (Frank Adams)
Newsgroups: net.arch
Subject: Re: RISC (really on multiplication due to structures)
Message-ID: <476@mmintl.UUCP>
Date: Tue, 9-Jul-85 13:41:23 EDT
Article-I.D.: mmintl.476
Posted: Tue Jul  9 13:41:23 1985
Date-Received: Sun, 14-Jul-85 09:17:15 EDT
References: <639@vax2.fluke.UUCP> <2743@nsc.UUCP>
Reply-To: franka@mmintl.UUCP (Frank Adams)
Distribution: net
Organization: Multimate International, E. Hartford, CT
Lines: 17
Summary: 


In article <149@mips.UUCP> mash@mips.UUCP (John Mashey) writes:
(concerning integer multiplies) > the real
>question, is how frequent are these cases, really? [Not an answer, but
>a question whose answer needs to be known when you're making tradeoffs
>in CPU design].

Actually, this still isn't quite the right question.  The right question
is how common *should* these cases be?  By which I mean, assuming good
programming techniques.  I have seen too many structure definitions and
arrays which were padded or shrunk to make the size a power of two.
Programmers shouldn't (and shouldn't have to) worry about such things!

Analysis of current programming practices reflects the reactions to
current (and past) architectures, and tends to propogate mistakes.

All right, I will admit that such analysis is better than no analysis
at all.  But be aware of its limitations.