Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!uunet!zephyr.ens.tek.com!uw-beaver!ssc-vax!carroll
From: carroll@ssc-vax (Jeff Carroll)
Newsgroups: comp.arch
Subject: Re: massive linpack
Message-ID: <4112@ssc-bee.ssc-vax.UUCP>
Date: 13 Jun 91 21:11:59 GMT
References: <9106070135.AA02947@ucbvax.Berkeley.EDU> <1991Jun8.055711.13457@marlin.jcu.edu.au>
Sender: news@ssc-vax.UUCP
Reply-To: carroll@ssc-vax.UUCP (Jeff Carroll)
Organization: Boeing Aerospace & Electronics
Lines: 30

In article <1991Jun8.055711.13457@marlin.jcu.edu.au> csrdh@marlin.jcu.edu.au (Rowan Hughes) writes:
>
>I would have thought  (N**2)*C*E was more appropriate.
>
>>         Richard Lethin asks:
>>Aren't really large problems of interest sparse matrices anyway?  Who
>>holds the record for massive sparse matrix solves?
>
>Yes, large problems (N > 10,000) are typically sparse, and with strong
>diagonal banding. These can only be solved with iterative methods, and
>only on vector, or large parallel machines. Iterative methods usually

There is at least one fallacy here. I have worked on at least one class
of problem involving boundary integral methods which is both large by
this definition and *dense*. We typically solve these by LU decomposition.

I submit that if we can solve large dense problems by direct methods, we
can certainly solve sparse problems of the same size by the same methods.

One (here nameless) vendor of vector processors put himself at a
considerable disadvantage in a procurement here once by assuming that a
large problem had to be sparse, rather than reading the RFP.



-- 
Jeff Carroll		carroll@ssc-vax.boeing.com

"...and of their daughters it is written, 'Cursed be he who lies with 
any manner of animal.'" - Talmud