Xref: utzoo comp.arch:8598 comp.lang.misc:2707 comp.lang.c:16712
Path: utzoo!mnetor!tmsoft!dptcdc!jarvis.csri.toronto.edu!mailrus!cornell!uw-beaver!mit-eddie!fenchurch!jbs
From: jbs@fenchurch.mit.edu (Jeff Siegal)
Newsgroups: comp.arch,comp.lang.misc,comp.lang.c
Subject: Re: Peephole optimisation
Message-ID: <11234@eddie.MIT.EDU>
Date: 4 Mar 89 17:23:48 GMT
References: <740@tetons.UUCP> <76700068@p.cs.uiuc.edu> <671@oracle.oracle.com> <730@microsoft.UUCP> <1153@l.cc.purdue.edu> <8650@aw.sei.cmu.edu> <11201@eddie.MIT.EDU> <420@chem.ucsd.EDU>
Sender: uucp@eddie.MIT.EDU
Reply-To: jbs@fenchurch.UUCP (Jeff Siegal)
Organization: MIT, EE/CS Computer Facilities, Cambridge, MA
Lines: 16

In article <420@chem.ucsd.EDU> tps@chem.ucsd.edu (Tom Stockfisch) writes:
>I don't think three instructions is enough if your RISC machine has
>a lot of parallelization and scheduling is considered.
>[...]Perhaps some
>specialized scheduler pass would be more practical.

I have to agree with this last assessment.  In general, scheduling is
very different from peephole optimization.  A peephole optimizer
usually transforms a code sequence into an equivalent (but shorter)
code sequence, while a scheduler will be concerned with ordering of
instructions.  

In addition, the strategies for selecting the optimal (or at least
better) are quite different from those used in a peephole optimizer.

Jeff Siegal