Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!rochester!pt.cs.cmu.edu!sei.cmu.edu!firth
From: firth@sei.cmu.edu (Robert Firth)
Newsgroups: comp.arch
Subject: Re: Optimization vs. the programmer
Message-ID: <920@aw.sei.cmu.edu.sei.cmu.edu>
Date: Mon, 13-Apr-87 08:27:16 EST
Article-I.D.: aw.920
Posted: Mon Apr 13 08:27:16 1987
Date-Received: Sun, 19-Apr-87 09:42:20 EST
References: <479@danews.ATT.COM> <3300003@uiucdcsm> <344@zuring.mcvax.cwi.nl>
Sender: netnews@sei.cmu.edu
Reply-To: firth@bd.sei.cmu.edu.UUCP (PUT YOUR NAME HERE)
Organization: Carnegie-Mellon University, SEI, Pgh, Pa
Lines: 10

In article <344@zuring.mcvax.cwi.nl> dik@zuring.UUCP (Dik T. Winter) writes:
>And that is right, otherwise you might see i * 14 transformed to
>(i << 4) - (i << 1), which is slower on some machines.
>(The Pascal compiler on one of the systems I use did this, this
>'optimization' has been turned off.)

Also uses an unnecessary register.  If a compiler makes this change it
will emit ((i<<3)-i)<<1.  Transformations such as these are VALID on
(almost) all machines, but they are not necessarily OPTIMISATIONS.  Mr
Winter and others are right - leave these tricks to the compiler.