Path: utzoo!utgpu!watmath!clyde!att!osu-cis!tut.cis.ohio-state.edu!rutgers!bellcore!texbell!sugar!peter
From: peter@sugar.uu.net (Peter da Silva)
Newsgroups: comp.sys.amiga.tech
Subject: Re: Aztec compiler ineffeciencies
Message-ID: <3024@sugar.uu.net>
Date: 26 Nov 88 22:31:50 GMT
References: <3013@sugar.uu.net> <7949@dasys1.UUCP>
Organization: Sugar Land Unix - Houston, TX
Lines: 34

In article <7949@dasys1.UUCP>, rodd@dasys1.UUCP (Rod Dorman) writes:
> In article <3013@sugar.uu.net> karl@sugar.uu.net (Karl Lehenbauer) writes:
> >I've been examining the output of Aztec 3.6a and have discovered that it does
> >no optimization on divides, such as turning divides by constants into shifts
> >	asr.l	#1,d4

> As many compiler writers have found in the past and I'm sure many will
> rediscover in the future, this only works for positive values.  Try
> shifting -1 and you'll get -1 as a result, *not* zero as one would expect.

First consider (unsigned)/2. Then...

As many compiler users in the past have discovered, having -1/2 == -1 is
often much more useful... particularly with graphics applications. This
depends on how you interpret integer arithmetic... should division be
truncated towards zero (trunc), to the numerically lower value (floor),
or the numerically higher value (cieling). Mathemeticians like trunc, but
this leads to a singularity at 0. Say that, for some reason, you're drawing
a line at arctan(1/2) using a straightforward algorithm, and it goes through
the origin.

	With trunc			     With floor
                  **                                  **
                **                                  **
             ***                                  **
           **                                   **
         **                                   **

See comp.lang.c for assorted flames.
-- 
		    Peter da Silva  `-_-'  peter@sugar.uu.net
		     Have you hugged  U  your wolf today?

	          Disclaimer: My typos are my own damn busines#!rne