Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!uunet!mcsun!ukc!warwick!nott-cs!bhamcs!spsl
From: spsl@cs.bham.ac.uk (Steven Lam)
Newsgroups: comp.sys.super
Subject: Re: Massively Parallel LINPACK on the Intel Touchstone Delta machine
Message-ID: <1991Jun20.120808.27762@cs.bham.ac.uk>
Date: 20 Jun 91 12:08:08 GMT
References: <1991Jun10.144354.695@chpc.utexas.edu> <1991Jun10.235501.7039@ariel.unm.edu> <11806@mirsa.inria.fr>
Sender: news@cs.bham.ac.uk
Organization: Birmingham University Computer Science
Lines: 24


Dear Sir,

I understand that you've been using both Cray and CM-2. Can you give me some
figures, such as the price and the performance of each machine? I personnally
believe that fine-grain computers are much cheaper to build and hence less
dependent on integration technology. Does the compiler for CM-2 allow you to
grap several number of processing cells for your applications and other users
may do the same? Have you ever experience the communication bottleneck when
working with Cray? There are a few papers on CM-2 machine recently, they found
that the connection machine is not very good in computing singular value
decompostion or similar problem, as the cell utilization may fall to 20-40%,
what is your comment?

Stephen Lam

+++++++++++++++++++++++
School of Electronic Engineering
University of Birmingham
Birmingham B15 2TT
e-mail: spsl@uk.ac.bham.cs
+++++++++++++++++++++

Thank you.