Re: st: Best machine to build for running STATA

[Stas Kolenikov <skolenik@gmail.com>]

On Mon, Feb 22, 2010 at 6:50 PM, Dana Chandler <dchandler@gmail.com> wrote:

> Re: Memory allocation... Christopher Baum mentions that for a 13 GB
> data set, 24GB of RAM would be recommended. Are there any rules of
> thumb people use in terms of how much memory a system should have to
> comfortably do analysis on a dataset of a given size? Also, how much
> memory should be allocated to a dataset. If you have a 250mb dataset,
> is allocating 1gig overkill, can this be harmful?
>

It was not Kit, it was me. I thought that for a 13Gb data set, 16Gb won't be
enough (if you give Windows 1-2Gb), and 24Gb seemed like the next "nice"
number. Most Stata commands and data sets have a certain overhead, and for
more complicated commands, you would want to have 30-50% additional memory
(assuming your data set is already efficient: if you work with 20 variables
out of 1000, you may not notice these overheads). Whether allocating extra
memory is harmful or not very much depends on the memory management habits
of your OS. Windows Task Manager shows the amount of memory actually used by
the software, and when I

set memory 200m
use [data set of 10m]

it only shows the actual amount of 10Mb or so used by Stata.

Personally, most of my work in Stata are simulations, plus some -gllamm-
estimation that takes quite a bit of time. So I can advocate Stata MP, but I
am not entirely qualified to judge the fine details of hardware requirements
for handling large data sets and specific architectures that would
supposedly work better.

-- 
Stas Kolenikov, also found at http://stas.kolenikov.name
Small print: I use this email account for mailing lists only.
*
*   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/