Re: st: Stata graphics speed

[David Kantor <kantor.d@att.net>]

I believe that Uli is correct, and that the graph commands also make 
heavy use of object-oriented techniques, which were added to Stata 
just for the purpose of rewriting -graph-.

I find that when I want to see a simple graph without having to wait, 
I use -gr7-, which is the older (Stata 7 and earlier) version of 
-graph-. Of course, this is a step backwards in technology, and lacks 
many features found in the current -graph- (and the syntax is 
different), but it is much faster that the current -graph-; the graph 
will appear nearly immediately.  See -help gr7-, and a version 7 user 
manual if you can find one.

I usually use graphs for data exploration purposes, in which case I 
want to see them immediately; so I use -gr7-.  If I want to draw a 
graph for a report, I would use -graph-.

I hope this helps.


At 06:35 AM 6/12/2006, you wrote:

> I think it's because much of Stata graph-programs are written as Ado 
> code. The
> programs have to be loaded into memory when used the first time, i.e. Stata
> have to look for the files. Graphs will take even longer if Stata is
> installed on a slow network.
>
> This is also just a guess.
>
> Uli
>
> Allan Reese (Cefas) wrote:
> > Friedrich Huebler <huebler@rocketmail.com> wrote 9 Jun 2006
> > Subject: Re: st: Video card and Stata speed
> >
> > As an example, a fairly detailed map of the world drawn with -tmap-
> > from SSC takes about one minute to render on my computer. ...
> > Other Stata graphs like scatter plots or bar charts are *redrawn* [my
> > emphasis] instantaneously and virtually nothing would be gained from a new
> > video card.
> > -----
> >
> > Something I've never understood is why graphs take a long time to appear in
> > the first place.  Commands like "gen y=_n" happen immediately, even with
> > quite large numbers of observations.  (I could just notice the delay with
> > obs=1 millon). In contrast, a simple "scatter y x" with a few observations
> > takes several seconds before anything appears.  That sometimes applies even
> > when what comes up is a syntax error.  Can anyone explain?  I've guessed
> > that the delay is while Windows allocates memory for the graph window, but
> > that is just a guess.
> >
> > Allan
> [...]
*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/