Path: utzoo!attcan!uunet!fernwood!apple!sun-barr!cs.utexas.edu!wuarchive!psuvax1!rutgers!cmcl2!kramden.acf.nyu.edu!brnstnd
From: brnstnd@kramden.acf.nyu.edu (Dan Bernstein)
Newsgroups: comp.lang.misc
Subject: Re: C's sins of commission
Message-ID: <2440:Nov607:32:5890@kramden.acf.nyu.edu>
Date: 6 Nov 90 07:32:58 GMT
References: <PCG.90Oct30201101@teachk.cs.aber.ac.uk> <3673@stl.stc.co.uk> <1990Nov2.052815.22188@murdoch.acc.Virginia.EDU>
Organization: IR
Lines: 25

In article <1990Nov2.052815.22188@murdoch.acc.Virginia.EDU> gl8f@astsun9.astro.Virginia.EDU (Greg Lindahl) writes:
  [ on one type of aliasing problem ]
> 	a[i] = b[i] / c[i];
> 	d[i] = b[i] * c[i];
  [ Fortran sees separate arrays and loads b[i] and c[i] only once ]
  [ C sees possible aliasing and loads them twice ]

Hmmm. I would almost automatically optimize the above into

  {
   register double bi = b[i];
   register double ci = c[i];

   a[i] = bi / ci;
   d[i] = bi * ci;
  }

Don't ask me to philosophize about this fragment.

> but
> don't make my code run 30% slower.

Hey, dude, write it 30% faster in the first place.

---Dan