Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!clyde.concordia.ca!uunet!mailrus!cs.utexas.edu!swrinde!zaphod.mps.ohio-state.edu!sol.ctr.columbia.edu!emory!hubcap!watmath!cpsc.ucalgary.ca!cleary
From: watmath!cpsc.ucalgary.ca!cleary@uunet.UU.NET (John Cleary)
Newsgroups: comp.parallel
Subject: Re: Using the Butterfly Parallel Processor
Summary: paging on the Butterfly (performance problems)
Message-ID: <7430@hubcap.clemson.edu>
Date: 14 Dec 89 15:46:04 GMT
Sender: fpst@hubcap.clemson.edu
Lines: 26
Approved: parallel@hubcap.clemson.edu

In article <7385@hubcap.clemson.edu>, slim@cis.ohio-state.edu (Scott Whitman) writes:
> 	I am in the process of optimizing my program for running on
> the BBN Butterfly GP1000.  I was wondering if anyone has any
> suggestions for looking at how to tune the program for good speedup.
> I had been working on a small Butterfly (9 nodes) and got reasonable
> speedup.  When I finally moved to a larger machine, I noticed that
> at 24 nodes, the performance was worse than with 8.  Needless to say,
> this is not a desirable affect.

There is a known problem on the Buuterfly implementation of Mach where it has
a tendency to keep te page tables for a number of nodes on one node.
There are also a number of other places whree there are centralized
data structures.  I dont understand all the details of this but others
have reported effects just like yours.  As to a fix I cant help much.
Maybe wait for the next release oft he operating system :-)

	John G. Cleary
	Department of Computer Science
	University of Calgary
	2500 University Drive 
	N.W. Calgary
	Alberta T2N 1N4
	Canada

	cleary@cpsc.UCalgary.ca
	Phone: (403)282-5711 or (403)220-6087