Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!cs.utexas.edu!usc!zaphod.mps.ohio-state.edu!mips!hal!mark
From: mark@mips.COM (Mark G. Johnson)
Newsgroups: comp.arch
Subject: Re: New benchmark (was: Integer Multiply/Divide)
Message-ID: <34233@mips.mips.COM>
Date: 8 Jan 90 17:24:39 GMT
References: <2131@sunset.MATH.UCLA.EDU>
Sender: news@mips.COM
Reply-To: mark@mips.COM (Mark G. Johnson)
Lines: 44


Email failed me; therefore ...

Here are the results of running Peter Montgomery's "program bchmark" for
Integer Multiply and Divide, on a MIPS RC3240.  This is a deskside machine
in a PC-sized cabinet (23" x 7" x 18"); essentially it's an upgrade of the
earlier M/120 machine.

I didn't change the source code; just took it as-is and applied the naive
command line "f77 -O montgomery.f".  Running the executable file a.out gave:

MIPS RC3240 compiled with f77 -O
 Gener  Dble  Sngl Cmplx Intgr Modlr TrlDiv PrbPrm   GCD BinDec Cfrac  Geo mean
  0.08  0.26  0.23  0.34  0.39  0.46   2.84   0.10  0.63   0.60  0.46  0.36 sec

{This falls between the single-processor Alliant with vectorization and
optimization, and the 6-processor Alliant with V&O, in Geo mean performance.}

For those who desire to bicker about the benchmark's representative-ness
of "real code", or potential lack thereof, here is a teeny bit of profiling
data.  A single line of source, marked below with >>>>, is responsible for
over 1/4 of all cycles consumed by the program (dynamically, on RC3240):


            do 100 j = 5, NINT(SQRT(REAL(n))), 6                      [575]
      >>>>      if (MOD(n,j).eq.0 .or. MOD(n,j+2).eq.0) then          [576]
                    PRIME = .FALSE.                                   [577]
                    return                                            [578]
                end if                                                [579]
100         continue                                                  [580]


Here's the profiler output {for an RC3240, compiled with f77 -O):

    procedure (file)                           line bytes     cycles      %
    -----------------------------------------------------------------------
    prime_ (montgomery.f)                       576   108   15677188  25.67
    ... etc ...

Apologies in advance if I did the measurement incorrectly.,
-- 
 -- Mark Johnson	
 	MIPS Computer Systems, 930 E. Arques, Sunnyvale, CA 94086
	(408) 991-0208    mark@mips.com  {or ...!decwrl!mips!mark}