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

[Valery P. <pi...@is...>]

Dear Alan,

I'm sorry for delay with reply.
The problem is not in internal plplot interface speed.
I issue is that python read ascii files much slower than yorick. The problem 
is seen on the large data set (say > 1Mb).
However I observe that for binary files there is no such problems.
Situation is opposite, reading binary data file in python is  easy than with 
yorick.  I will switch to python interface completely  after this discovery.

Happy Holidays and Happy New Year!

Valery




Alan W. Irwin wrote:
> 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
> __________________________
>
> -------------------------------------------------------------------------
> Take Surveys. Earn Cash. Influence the Future of IT
> Join SourceForge.net's Techsay panel and you'll get the chance to share
> your opinions on IT & business topics through brief surveys - and earn cash
> http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
> _______________________________________________
> Plplot-general mailing list
> Plp...@li...
> https://lists.sourceforge.net/lists/listinfo/plplot-general