st: mem usage and Stata version

[Will Hauser <whauseriii@gmail.com>]

I am working with a largish dataset.  It is 1,405,050 cases (544,758 kb) and
I am merging in some additional info from other sources.  The final variable
count will be somewhere in the neighborhood of 200 and probably 1.3 million
cases (some attrition due to missing data).  As it is I am running into
memory problems and while these could be addressed by trimming and
compressing the dataset at this point the problem will only get worse as I
move forward and should become insurmountable.   

I am working on a pc with 3 gigs memory (upgradable to 4) with a copy of
Stata 10 intercooled.  The platform is 64bit but I don't know if the Stata
version is 64 or 32bit.  I've read in the archives that Stata version 11 may
allow me to devote more memory to the program but only if it is 64 bit
(32bit and I would be worse off than I am now).  I'm willing to move to a
new version of Stata or have the pc upgraded although the new version of
Stata will probably be at my expense which is not altogether great since I
am a poor graduate student.  I'm wondering what the best course of action
is.  If I upgrade the memory on the pc will Windows 7 let me devote memory
to it?  Or is it my version of Stata that is the problem?  Maybe it's both.
How can I tell?

FYI, the error I get when merging the files is r(902) "no room to add more
variables because of width...."
The set mem command allows me to devote up to 1000m to Stata which is not
enough.  To demand more (i.e. 1001m) results in the "op. sys. refuses to
provide memory" error.

So what is the limiting factor in my arrangement?  Thank you all for any
help and input you can provide.

Will Hauser
Ph.D. Candidate
FSU College of Criminology



*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/