Xref: utzoo comp.lang.c:10480 comp.lang.fortran:696
Path: utzoo!attcan!uunet!husc6!purdue!umd5!mimsy!chris
From: chris@mimsy.UUCP (Chris Torek)
Newsgroups: comp.lang.c,comp.lang.fortran
Subject: Re: no noalias not negligible: more red herrings, conclusion unchanged
Message-ID: <11711@mimsy.UUCP>
Date: 28 May 88 18:37:40 GMT
References: <54713@sun.uucp>
Organization: U of Maryland, Dept. of Computer Science, Coll. Pk., MD 20742
Lines: 34

In article <54713@sun.uucp> dgh%dgh@Sun.COM (David Hough) writes:
>... the following versions of ROLLED source; the first results in unrolled
>object code; the second and most other equivalent ones do not:

[Some text deleted.  First version:]
>        i = 0;
>        do {
>                dy[i] = dy[i] + da * dx[i];
>                i++;
>        } while (i < n);

[Second:]
>        i = 0;
>        do {
>                dy[i] = dy[i] + da * dx[i];
>        } while (i++ < n);

This illustrates a very important point.  These two loops are *not*
equivalent!  If n is 2, the first loop executes twice (as it should);
the second executes three times.  So remember:

  Make it right before you make it fast, and do try not to break it
  when you make it fast. :-)

(I still do not understand the insistence on avoiding the form

	do {
		dy[i] += da * dx[i];
	} while (++i < n);

which is much easier to read, and remains correct.)
-- 
In-Real-Life: Chris Torek, Univ of MD Comp Sci Dept (+1 301 454 7163)
Domain:	chris@mimsy.umd.edu	Path:	uunet!mimsy!chris