Re: Re: [ncl-talk] NCL vs. GrADS plotting performance

From: Jan Ploski <Jan.Ploski_at_nyahnyahspammersnyahnyah>
Date: Fri, 25 Jul 2008 17:56:48 +0200

"Paulo Henrique" <paulohsm_at_gmail.com> schrieb am 07/25/2008 12:55:37 PM:

> Hi Jan,
>
> I have used Ferret (http://ferret.wrc.noaa.gov/Ferret/) to produce
> real-time weather forecast images (some examples at http://www.
> funceme.br/rams/). Time to process images hasn't been problem to me.
>
> Recently I started using NCL to do the same job, and I would like to
> know if you are interested in testing and comparing Ferret's performance
too.
>
> I don't know any existing performance comparison between this
> packages, but I want to express my availability if you have some
> idea on how to test and compare this packages.

Hi Paulo,

I have reconstructed the case which makes me worry about NCL's perfromance
and uploaded all the relevant files here:

https://bi.offis.de/ncl_perf/

I guess that one possibility for an ad-hoc performance comparison would be
to check how much time Ferret needs to create an equivalent of
frame1-nice.jpg based on the same input data (see README.txt there). I
can't really say that "this NCL plot is slow"; the very same script
applied to another input may run in under a second whereas on this
particular data it takes over a minute. I also haven't excluded the
possibility that slight changes to the NCL script (such as modifying the
number of steps in the color scale) could make it a lot faster.

However, when the intent is to use such scripts to generate many plots
automatically, one cannot tweak them manually here and there based on any
particular input case. Of course, even if I could find out that Ferret is
a lot faster on this case, I still wouldn't know whether it does not
suffer from similar unpredictable performance on other data. That's why I
am eager to hear "been there, done that" experience. Out of curiosity,
what advantages do you see of NCL over Ferret or vice versa?

Best regards,
Jan Ploski

--
Dipl.-Inform. (FH) Jan Ploski
OFFIS
FuE Bereich Energie | R&D Division Energy
Escherweg 2  - 26121 Oldenburg - Germany
Phone/Fax: +49 441 9722 - 184 / 202
E-Mail: Jan.Ploski_at_offis.de
URL: http://www.offis.de
_______________________________________________
ncl-talk mailing list
ncl-talk_at_ucar.edu
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Fri Jul 25 2008 - 09:56:48 MDT

This archive was generated by hypermail 2.2.0 : Fri Jul 25 2008 - 11:11:16 MDT