Xref: utzoo alt.fractals:123 comp.arch:15008 comp.sys.sequent:530
Path: utzoo!utgpu!news-server.csri.toronto.edu!clyde.concordia.ca!uunet!cs.utexas.edu!usc!jarthur!mwilkins
From: mwilkins@jarthur.Claremont.EDU (Mark Wilkins)
Newsgroups: alt.fractals,comp.arch,comp.sys.sequent
Subject: Re: Sure, with a CM (was Re: Other Fractal Topics)
Message-ID: <5601@jarthur.Claremont.EDU>
Date: 29 Mar 90 13:44:05 GMT
References: <MWB#`L*@rpi.edu> <2015@ariel.unm.edu> <3194@draken.nada.kth.se> <1990Mar29.101949.18049@oracle.com>
Followup-To: comp.sys.sequent
Distribution: na
Organization: Harvey Mudd College, Claremont, CA 91711
Lines: 19

In article <1990Mar29.101949.18049@oracle.com> csimmons@oracle.com writes:

[ Describes relative Mandelbrot set benchmarks of parallel computers ]

>	A Sequent Symmetry (at least the one I benchmarked) gets at
>	most .5 Mandelflops per CPU.

[ Goes on to define the "Mandelflop" as an MFlop when one measures the 
  operations in the inner loop of a mandelbrot set, a benchmark which
  appears generous compared to MFlops by Whetstone.                   ]

  Almost certainly not if you have the machine to yourself.  My Macintosh II
with a 16MHz processor and 2 wait-state memory gets 0.5 MFlops by Whetstone,
and a Sequent runs with 0 wait-state memory.  Even with operating system
overhead, it's not that slow.  Is it?
  Followups to comp.sys.sequent.

-- Mark Wilkins
   mwilkins@jarthur.claremont.edu