[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Top 10 for old farts

Pavel Romashkin wrote:
> Paul van Delst wrote:
> > Ahhhhh. When IDL v5 was released, the first thing I did was convert some
> > of my code to use OG. My god was it slow. Also, when I started up
> > Insight I figured that it had to be the showcase of what IDL OG had to
> > offer. That all pretty much turned me off OG - and thus IDL objects - in
> > general. The latter was a bad decision I know, but my "to-do" list kept
> > getting bigger during the 2-week period I spent playing with IDL OG.
> I am not arguing about IDLgr* speed with Paul, or, God forbid, David,
> but I find that with today's computers even object graphics is not that
> slow. I don't use multitudes of graphic objects, but rendering a few
> line plots with zoom, point selection, deletion, calculation, adding and
> removing temporary grPlots, etc. - I notice so little difference
> compared with direct graphics that it is not detrimental to the
> application. I think that once faster machines become more widespread,
> 3D object graphics will find more use. On a Pentium 60 w/o OpenGL, I
> agree, it is intolerable.

Maybe for 99% of cases you're right and I fall in the remaining 1%? :o)

The test I applied was to plot what I usually plotted day-to-day with DG
and OG. What I usually plot are high resolution atmospheric radiance
and/or transmittance spectra - sometimes 100,000's points per spectrum,
with, oh, about 40-100 spectra per atmospheric profile. Plotting this
with DG took a long time (seconds) but plotting in OG - well I could go
and answer the call of nature, return and it would still be in the
process. I'm sure part of it is due to the way I put the plotting code
together - remember this was my learning experience - but after a couple
of days I started thinking that one of the main reasons I bought IDL was
so I could type:

PLOT, x, y

and, shazaam! there was my data on screen. With all the programming
required to get a plot with OG - onscreen or on paper - I may as well
buy a F90 graphics library for my system (non-portable, granted) and do
the number crunching AND the plotting in F90. (I haven't had to write
code to scale data axes for plotting since I linked up an old tektronix
plotter to a VAX terminal back in old day...well, mid-80's)

In this situation, the destination was more important than the journey.


p.s. The system was a Sun dual 250MHz processor with the optional
Creator 3-D graphics accelerator and 512MB of RAM. I don't know if the
graphics extras helped the IDL OG plotting speed or not, and although
it's not a super flash machine, it was pretty zippy. So, given all that,
I couldn't justify using IDL's OG at all.

Maybe things have changed since IDL v5.0 but the first impression has
stuck like sh*t to a blanket. :o)
Paul van Delst           Ph:  (301) 763-8000 x7274
CIMSS @ NOAA/NCEP        Fax: (301) 763-8545
Rm.202, 5200 Auth Rd.    Email: pvandelst@ncep.noaa.gov
Camp Springs MD 20746