Newsgroups: comp.lang.fortran,comp.parallel.pvm
From: genew@netcom.com (Gene Wagenbreth)
Subject: Re: fortran77 programming
Message-Id: <genewF3Kp9E.LKE@netcom.com>
Organization: Applied Parallel Rsch (Topanga)
References: <3660D892.E1EDBF12@est.it> <3665044D.A0F57E59@kings.uq.edu.au>
    <3665E491.7993355@flash.net> <y6k909mqww.fsf@tweedledumb.cygnus.com>
    <m21zmgnuau.fsf@blinky.bfr.co.il> 	<y6emqejtgq.fsf@tweedledumb.cygnus.com>
    <m2d85xi988.fsf@blinky.bfr.co.il> <y6yaokiw0r.fsf@tweedledumb.cygnus.com>
Date: Mon, 7 Dec 1998 02:22:26 GMT
Sender: genew@netcom18.netcom.com
Xref: ukc comp.lang.fortran:61889 comp.parallel.pvm:7844


You mentioned that there are twice as many registers on 2 processors.
There is also twice as much cache. If yu choose one processor that
is twice as fast, it may need twice as much cache. I have seen
super linear speedup due to a problem fitting entirely in cache when
run on multiprocessors.
-- 
-------------------------------------------------------------------
Gene Wagenbreth  Programming for Dollars
My opinions are mine, I think.
-------------------------------------------------------------------

