Path: utzoo!mnetor!uunet!husc6!mit-eddie!uw-beaver!cornell!rochester!PT.CS.CMU.EDU!K.GP.CS.CMU.EDU!lindsay
From: lindsay@K.GP.CS.CMU.EDU (Donald Lindsay)
Newsgroups: comp.arch
Subject: Re: 2-D arrays (was: RISC is a nasty no-no)
Message-ID: <1020@PT.CS.CMU.EDU>
Date: 3 Mar 88 04:45:10 GMT
References: <179@wsccs.UUCP: <696@nuchat.UUCP> <284@scdpyr.UUCP> <25699@linus.UUCP> <2378@pdn.UUCP>
Sender: netnews@PT.CS.CMU.EDU
Organization: Carnegie-Mellon University, CS/RI
Lines: 20


2-D arrays should not in general be implemented as vector-of-pointer-to-
vector. The Burroughs 6700 had to do it this way, and it was a defect of the
machine.

This implementation method can be useful in specific cases. In general, it
has bad properties.

For one, Fortran can't use it. Fortran specifies that you can step past the
end of one row and find yourself at the front of the next. (Or rather,
columns .. ah, yes, Fortran! )

For another, it isn't uniform w.r.t. rows-versus-columns. One is preferred,
the other penalized.  The problem is that an algorithm may accidentally be
coded to vary the "wrong" subscript more often than the "right" one.  The
performance hit (and the ensuing "optimization") are needless.


-- 
	Don		lindsay@k.gp.cs.cmu.edu    CMU Computer Science