Path: utzoo!mnetor!uunet!lll-winken!lll-lcc!ames!hao!husc6!mit-eddie!uw-beaver!cornell!rochester!PT.CS.CMU.EDU!sei!sei.cmu.edu!firth
From: firth@sei.cmu.edu (Robert Firth)
Newsgroups: comp.arch
Subject: Re: RISC is a nasty no-no!
Message-ID: <4399@aw.sei.cmu.edu>
Date: 1 Mar 88 13:38:59 GMT
References: <179@wsccs.UUCP: <696@nuchat.UUCP> <284@scdpyr.UUCP> <11199@duke.cs.duke.edu> <998@PT.CS.CMU.EDU>
Sender: netnews@sei.cmu.edu
Reply-To: firth@bd.sei.cmu.edu.UUCP (Robert Firth)
Organization: Carnegie-Mellon University, SEI, Pgh, Pa
Lines: 15

In article <998@PT.CS.CMU.EDU> edw@IUS1.CS.CMU.EDU (Eddie Wyatt) writes:

>EXCUSE ME, but when I declare an array to be an n by m matrix in
>C (float foo[n][m]) I get a contiguous block of memory.  The
>representation IS NOT row-vector or column-vector.  So when I access
>index number i,j someone has to perform the calculation 
>base + sizeof(type)*(i*m+j).

However, going by Knuth's statistics on array usage, more than 95%
of those multiplications occur in loops where the index is the
induction variable.  The optimisation called "strength reduction"
can then remove them.  If your target machine has a slow multiply,
such an optimisation is probably necessary.  How to do it is
explained in detail in many books, including eg Wulf's 'The Design
of an Optimising Compiler'.  The first Fortran compiler included
this optimisation, so it's been around for quite a while.