- Speed test. Obfuscation

From: Campbell, Robert D (SELEX) (UK Edinburgh) <"Campbell,>
Date: Mon, 6 Nov 2006 13:25:31 -0000

----boundary-LibPST-iamunique-1605591794_-_-
Content-type: text/plain




I have used a simple CPU / Maths speed test for a while.

It does not attempt to capture cache effects. While these can certainly
be large,
they also seem to be quite variable & unpredictable, so I doubt if
anything
can capture them reliably. Anyway this routine has seemed to give
indicative
performance metrics quite reliably for our kind of 'big' sums.

The algorithm in BlackBox:

PROCEDURE DoReal (trials : INTEGER);
  CONST
    pi = 3.14159265358979323846;
  VAR
    i, n : INTEGER;
    con, sum : REAL;
  BEGIN
    f.Str ('Number of trials = '); f.IntComma (trials, 0, 3); f.Ln;
    Timer.Reset;

    con := pi / 180.;

    FOR n := 1 TO trials DO
      sum := 0.;
      FOR i := 1 TO 100 DO
        sum := sum + Math.Sin (con * i)
      END
    END;

    Timer.Mark;
    f.Str ('Con = '); f.RealComma (con, 20, 15, 5, ','); f.Ln;
    f.Str ('Sum = '); f.RealComma (Sum, 20, 15, 5, ','); f.Str ('
in');
    Timer.WrtMark (f, TRUE)
  END DoReal;


(In this code 'f' is something that knows how to write to the Log, and
Timer
is a Module that can do simple timing tasks - the details are not
significant.)


One of my old results is for an early (1992 I think) DEC VAXstation &
FORTRAN:
  5.75 secs for 1000 trials.

One of my slow results was for an 8 MHz Motorola 68000 with software
floating point:
  687 secs for 1000 trials. (I actually used that machine a lot!)

One of my interesting results was for Maple:
  Very slow, but it got the (exactly) correct answer: a page of A4
involving root 2's,
  root 3's, root 5's, and lots of pi's.

Anyway about today's hardware. We have been considering .NET and
recently made these measurements:

        Visual Basic 6, compiled, pre .NET : 10.6 sec for 1 million
trials.
        Visual Basic.NET (code embedded) : 4.9 sec for 1 million
trials.
        Visual Basic.NET (calling a C# dll)): 5.5 sec for 1 million
trials.


        C#.NET (code embedded)) : 6.4 sec for 1 million trials.
        C#.NET (calling a C# dll)): 5.5 sec for 1 million trials.
        C++.NET (code embedded)) : 5.3 sec for 1 million trials.


        Matlab: 90 secs for 1 million trials (time scaled from a
different machine)


        BlackBox: 6.4 sec for 1 million trials.
        Gardens Point Component Pascal : About 5.5 secs - I have lost
the exact time.




Notes:
  1 - All results (except Maple) 64-bit floating point.

  2 - To be fair to Matlab (although I don't know why I should be!) you
can recover almost all the speed if:
  a) You can
&
  b) you can be bothered
to rewite the algorithm in 'Vectorised' form.

  3 - It would be interesting to have some comparable Java timings.
Informally
I think it is generally comparable, BUT it's trig routines do silly, and
very slow,
things if the angle is far outside the range +/- pi.


Cheers

Robert


********************************************************************
This email and any attachments are confidential to the intended
recipient and may also be privileged. If you are not the intended
recipient please delete it from your system and notify the sender.
You should not copy it or use it for any purpose nor disclose or
distribute its contents to any other person.
********************************************************************

--- BlackBox
--- send subject HELP or UNSUBSCRIBE to blackbox{([at]})nowhere.xy


----boundary-LibPST-iamunique-1605591794_-_-
Content-type: application/rtf
Content-transfer-encoding: base64
Content-Disposition: attachment; filename="rtf-body.rtf"
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----boundary-LibPST-iamunique-1605591794_-_---
Received on Mon Nov 06 2006 - 14:25:31 UTC