[Plplot-general] Relative speed of C, yPLot and Python/Numeric for processing arrays to be plotted

[Alan W. I. <ir...@be...>]

On 2006-12-14 08:45-0800 Alan W. Irwin wrote:

> However,
> have you tried some speed comparisons for the standard examples? The array
> preparation in those examples is done with Numeric which should be much (at
> least an order of magnitude) faster than ordinary Python array processing.

This relative speed topic engaged my curiosity so here is the actual
comparison between the speed of the C interface, the Yorick interface, and
the Python interface for example 8 (which uses quite a few arrays).
Note, the same core PLplot library was used in each case so the speed
differences are not due to the actual plotting, but instead in the preparation
of the arrays and other data that are required for the plots.

C:

time ./x08c -dev psc -o test.ps

real    0m0.487s
user    0m0.458s
sys     0m0.024s

Yorick interface (yplot):

time ../yplotl -batch x08.i -dev psc -o x08.ps

real    0m0.529s
user    0m0.498s
sys     0m0.026s

Python (with Numeric preparation of arrays that is done in xw08.py):

time ./x08 -dev psc -o test.ps

real    0m0.766s
user    0m0.584s
sys     0m0.041s

So yPlot and C have a modest speed advantage over Python/Numeric for this
test case, but nothing spectacular.  Thus, if you already like the Python
language for other reasons (like many scientists do), there is no need to
move to something else just to process numerical arrays.  Note, 
Python/Numeric was designed by those who were familiar with Yorick (which
was designed in the same institution, Lawrence Livermore Labs), and it
serves similar scientific high-level language needs.  Thus, in many ways it
is a successor to Yorick, and, of course, Numeric has now itself been
succeeded by the similar NumPy.  Note, I plan to switch our PLplot Python
examples from Numeric array processing to NumPy array processing as soon as
NumPy becomes widely available for the mainstream Linux distros.

Even though I have personally switched from using yPlot (The Yorick
interface to PLplot) to Python/Numeric for my own research plots, I will
continue to try and minimally maintain yPlot for those who 
continue to use it.  Glad to hear yPlot is still working for you, Valery.

Alan
__________________________
Alan W. Irwin

Astronomical research affiliation with Department of Physics and Astronomy,
University of Victoria (astrowww.phys.uvic.ca).

Programming affiliations with the FreeEOS equation-of-state implementation
for stellar interiors (freeeos.sf.net); PLplot scientific plotting software
package (plplot.org); the Yorick front-end to PLplot (yplot.sf.net); the
Loads of Linux Links project (loll.sf.net); and the Linux Brochure Project
(lbproject.sf.net).
__________________________

Linux-powered Science
__________________________