Re: for loop speed in different Windows versions

[Ben Diedrich]

Here's an update on the problem I am having. I ran a speed comparison
between three different Octave versions for different operations in a
'for' loop.

Ben Diedrich
------------

Run time results for three versions of Octave 2.1.5X. Tests provided by
Paul Thomas.

Pre-compiled is "octave-2.1.50a-inst.exe", a standalone Octave
installation with octave-forge, gnuplot (no X), and without ATLAS BLAS.
It is available from Octave Forge (octave.sourcefore.net). Compile
options and extra libraries are unknown.

Self-compiled versions (2.1.50 & 53) were compiled on a Cygwin 1.5.7-1
system with gcc 3.3.1-3 and ATLAS BLAS 3.6.0. Configuration options used
were:
"--enable-shared --disable-static"

The test system is a 2.4 GHz P4 running Windows 2000.

Test Strings
1. tic;z=1;x=[1:1e5];tot=0;for i=1:1e5;;end;toc
2. tic;z=1;x=[1:1e5];tot;  for i=1:1e5;tot;end;toc
3. tic;z=1;x=[1:1e5];tot=z;for i=1:1e5;tot=z;end;toc
4. tic;z=1;x=[1:1e5];tot=0;for i=1:1e5;tot=x(i);end;toc
5. tic;z=1;x=[1:1e5];tot=0;for i=1:1e5;tot=sin(z);end;toc

Results
Test  Pre-compiled  Self-compiled
      2.1.50        2.1.50   2.1.53
----  --------      -------  -------
1     0.066000      0.47900  0.16800
2     0.35700       2.1030   1.7170
3     0.61700       4.0720   3.0530
4     1.5470        9.5670   8.4720
5     1.3290        8.1820   6.3380



-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.octave.org
How to fund new projects:  http://www.octave.org/funding.html
Subscription information:  http://www.octave.org/archive.html
-------------------------------------------------------------