Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: Re: list in stata8


From   ariley@stata.com (Alan Riley)
To   statalist@hsphsun2.harvard.edu
Subject   st: Re: list in stata8
Date   Mon, 24 Feb 2003 11:37:15 -0600

Steven A. Melnyk (melnyk@msu.edu) and a couple of others commented
on the speed of -list- in Stata 8:
> I second this comment.  I have used STATA for a long time and have always 
> liked the speed.  The new STATA amazes me because it is so 
> slo-o-o-w.  Nowhere is this more evident than in the list command.  It is 
> also evident in the pwcorr.  I was doing a series of pwcorr analyses and 
> usually STATA would rip through them -- so fast that you could not 
> follow.  Not this one -- it was so slow that I could see each result.  So 
> far, with the significant decrease in speed, I must admit that I am not 
> impressed with V8.0.  It seems to be a step back, not a step forward.

Stata 8 overall is faster than Stata 7.  In addition to many other
optimizations, we made significant changes to the way programs are
parsed to increase speed, and under the Macintosh and Unix operating
systems, significantly increased the scrolling speed of the Results
window (the Windows version already benefitted from this under
Stata 7).

-list-, however, can appear to be slower since it by default
goes to more work up front to try to produce a nice listing.
The use of a simple option provides the old speed of list.

On January 15, the same question arose on Statalist and was
answered by me.  I will attach the original message and my response
below:

<------------begin prior question and response---------------------->
Fred Wolfe (fwolfe@arthritis-research.org) wrote
> Using  Windows 2000. In stata 7 with 313,000 obs, -list- is instantaneous. 
> In Stata 8, list, clean takes 2 seconds on my main machine and 8 seconds on 
> a laptop to begin to display. Is there a way to speed it up?

The quick answer is that Fred should use

   . list, nocompress

because he has a large dataset and wants output to start as
quickly as possible.

By default, -list- examines the data and decides on the width needed
to display each variable.  It does so to try to make the output of
-list- as nice as possible.  A display format on a variable might be
set to %9.0g, but if that variable contains only single digit integers
as values, there is no point in using 9 columns of screen width to
list it.

Examining the data can take a few seconds for large datasets
like Fred has.  In these cases, the -nocompress- option may be
used with -list- to tell it to just use the display formats on
the variables rather than trying to be smart and examine every
value in every variable.  Output will start instantaneously.
<------------end prior question and response---------------------->


Steven also mentioned that he thought -pwcorr- is slower in Stata 8.
Using the latest update of Stata 8, I typed

   . which pwcorr
   /usr/local/stata8/ado/base/p/pwcorr.ado
   *! version 3.0.8  13sep2000

and noted that pwcorr.ado is still the same as it was in Stata 7.
We just did some quick tests here and do not see that it is
slower than Stata 7.  If Steven would like to email me privately
with a do-file demonstrating what he is experiencing under Stata 8
and Stata 7 with the same data, we will investigate the issue.


--Alan
(ariley@stata.com)
*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index