Problem with load large dataset

[Enrique Vera]

I have a large dataset (the name of the data is ENAHO if you know it). My problem is that I cannot load dataset. When I run for open the dataset, stata show and I receive an error r(909) saying “op. sys. refuses to provide memory”


Also I show them memory settings:

. query memory
---------------------------------------------------------------------------------------------------------------------------------------------
Memory settings
set maxvar 5000 2048-32767; max. vars allowed
set matsize 400 10-11000; max. # vars in models
set niceness 5 0-10
set min_memory 0 0-0
set max_memory . 16m-0 or .
set segmentsize 16m 1m-1g

. set segmentsize 1g
r(909);

. do "C:\Users\TOSHIBA\AppData\Local\Temp\STD04000000.t mp"

. cd "C:\Users\TOSHIBA\Downloads\enaho-marisol"
C:\Users\TOSHIBA\Downloads\enaho-marisol

. use enaho01a-2011-2015-500.dta,clear
op. sys. refuses to provide memory
Stata's data-storage memory manager has already allocated 1536m bytes and it just attempted to allocate another 16m bytes. The
operating system said no. Perhaps you are running another memory-consuming task and the command will work later when the task
completes. Perhaps you are on a multiuser system that is especially busy and the command will work later when activity quiets down.
Perhaps a system administrator has put a limit on what you can allocate; see help memory. Or perhaps that's all the memory your
computer can allocate to Stata.
r(909);

end of do-file


How I to solve this problem?

[Clyde Schechter]

I'm afraid the error message says it all. Look into each of those possibilities. If none of them apply, then you will have to find a different computer with less restrictions on memory allocation to work with this data set.

[Phil Bromiley]

Alternatively, spend a little money on more memory. Memory is cheap.

[Jeph Herrin]

If you don't need all the variables, you can try -use [varlist] using -filename-, which I think reduces memory needs.

[Daniel Feenberg]

Loading only some variables is a possibility already pointed out. You can can also load only some observations (such as by sex or region). You should load a few observations and see if the data are stored compactly. Perhaps they are all doubles but could be floats or ints? Perhaps -compress- would help? Perhaps there are variables stored as long strings that could be recoded? I try to go over all of these at http://www.nber.org/stata/efficient/memory.html