MathGroup Archive 1995

[Date Index] [Thread Index] [Author Index]

Search the Archive

Timing results

  • To: mathgroup at christensen.cybernetics.net
  • Subject: [mg1938] Timing results
  • From: maddog at cybernet.com (MD Squiers)
  • Date: Wed, 23 Aug 1995 22:32:56 -0400
  • Organization: Cybernet Systems Corporation

I am in the process of evaluating Mathematica, both in terms of
functionality and performance.  This message deals with performance
issues:

I am trying to decide which platform is best suited to run Mathematica.  I
received timing tests from Wolfram Research and am interested in other
people's experiences.  A surprising (then again, maybe not) feature of these
tests is that the PowerMac 9500/132 either outperformed or was at the top
of each performance test.  Some examples:

Timing[N[Sin[1/2],2500]][[1]]

   PowerMac 9500/132       1.46 sec
   DEC 3000/600            1.83
   Pentium 100 Mhz         5.33

Timing[N[Pi,10001]][[1]]

   DEC 3000/600            2.1
   PowerMac 9500/132       3.06
   Pentium                 7.96

First[Timing[Eigenvalues[Table[Random[],{200},{200}]]]]

   IBM RS/6000             3.45
   PowerMac                4.61
   DEC 3000/600            8.07

f[x_]:=4x-4x^2;
Timing[ Nest[f,0.6,5000]][[1]]

   HP9000/735              1.74
   PowerMac                1.94
   DEC 3000/600            3.2

This is by no means as exhaustive as the list I have.  In each category, I
have listed the best performance, along with the PowerMac and DEC
3000/600.  Since I can potentially put Mathematica on any platform - I'd
like verifications of these timings, if possible.

Thank you,
MD Squiers,
Research Engineer,
Cybernet Systems Corporation


  • Prev by Date: Re: Producing eps files
  • Next by Date: Q: ArcTan[ Tan[ x ] ] = x
  • Previous by thread: Fourier Coefficients
  • Next by thread: Q: ArcTan[ Tan[ x ] ] = x