Rick Pietrobon ([email protected]) wrote
> I just installed Stata 7.0 for Windows in a pentium IV computer with 4 GIG
> of RAM and, to my surprise, the program is now running really slow.
> Besides Stata, another program called N5 was the only one having the same
> problem, which was solved upgrading it to the next version. According to
> the vendor, N5 (which is a program for qualitative analysis) couldn't
> handle computers with more than 1GIG.
>
> Is anybody working on computers with similar RAM capabilities running into
> similar problems in Stata. I contacted Stata support, but they never got
> back to me with an answer to the problem.
Our technical support database and mailserver records show that responses to
Rick regarding this issue were sent by Kevin Crow on November 14 (three
separate emails) and November 22 (one email). Our mailserver logs
show that the emails were successfully transmitted to a relay named
"gibson.acpub.duke.edu" for delivery to "[email protected]". I apologize
that the emails seem not to have reached Rick, but we received no
bounces nor indication that the mail did not reach its final destination.
Stata has no problems with computers with large amounts of memory,
although on 32-bit computers running Windows, individual applications
are restricted to accessing no more than a theoretical 2 gigabytes of
memory. In practice, the most memory Windows will give to a single
process is less than 2 gigabytes. Stata can use up to the maximum
amount of memory Windows will provide with no speed problems. The
Stata code is robust to even larger memory allocations under 64-bit
operating systems. We know of sites allocating more than 10 GB of
memory to Stata regularly.
To diagnose Rick's problem, some more information will be needed.
What is the processor speed of the Pentium IV? How much memory is being
allocated to Stata? How large is the dataset that is being analyzed?
Some common speed problems can be caused by allocating more memory
to Stata than is needed for the problem at hand. If a dataset is
only 10 MB in size, allocating a gigabyte of memory to Stata is
overkill, and depending on the configuration of the computer can
cause Windows to start using virtual memory (even if more RAM is
available on the computer).
The records of our technical support email exchanges with Rick show
that he reported the slowness starting as soon as the Results window
needed to scroll. It would be interesting for Rick to run some
experiments with "lightweight" commands that produce a lot of output,
such as -list-. For example,
. set more off
. set rmsg on
. set memory 2m
. use c:\stata\auto
. list
-set rmsg on- will cause Stata to report how long each subsequent
command takes to execute. I added the -set memory 5m- command to
set Stata's memory allocation to 2 megabytes just to make sure
that an unnecessarily large amount of memory had not been allocated
to Stata.
Slow scrolling can sometimes be caused by video card issues. It would
be helpful to know Rick's screen resolution and color depth settings.
There is one setting in Stata related to the number of lines "remembered"
by the Results window: -set scrollbufsize-. -help scrollbufsize- provides
information on it. In case this setting has been changed on Rick's
computer, he may wish to change it to the default:
. set scrollbufsize 32000
Note that this setting does not take effect until the NEXT session of
Stata; exit Stata and restart it.
We can continue to help Rick diagnose the problem through technical
support. Rick--since for some reason emails appear to be having
problems making it through to your duke.edu account, if you have
an alternate account we could cc: the emails to as a backup, please
let us know.
--Alan
([email protected])
*
* 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/
