How can I generate multicore load on Stata/MP?

Question

I'm working on monitoring CPU and memory usage of Stata/MP (the multicore version of Stata/SE) but I am not a Stata programmer (more of a Perl guy).

Can anyone post some code that utilizes a public data set to generate enough load on Stata/MP such that four CPU cores are utilized (or even maxed out) for several minutes or so?

If you can provide me with a .do file and a .dta file (or whatever I may need for this), I think I can take it from there. Thanks in advance!

Diogenes

This question, because it does not deal with statistical analysis, is not really appropriate on the stats site. You would likely get a fast and authoritative answer from the Stata user's listserve at stata.com/statalist . In the meantime let's migrate it to the relevant Stack Exchange forum. – Comedown

Answer 1

This should do it:

sysuse auto
expand 10000
bootstrap: logistic foreign price-gear_ratio

Gigot

Answer 2

// Clear memory before each run
// http://www.stata.com/help.cgi?clear
clear all

// Allocate plenty of memory
// http://www.stata.com/help.cgi?memory
set memory 1024m

// Load data set: 1978 Automobile Data
// (Use "sysuse dir" to list other data sets)
// http://www.stata.com/help.cgi?sysuse
sysuse auto

// Duplicate observations
// http://www.stata.com/help.cgi?expand
expand 10000

// Bootstrap sampling and estimation
// http://www.stata.com/help.cgi?bootstrap
// Generate high load using example from bootstrap documentation
bootstrap: regress mpg weight gear foreign
// Generate even higher load and for a longer period
//bootstrap: logistic foreign price-gear_ratio

This answer is based on an earlier answer, but I added some comments, some setup, and an extra bootstrap command that generates less load. You can put this into a file called "load.do" and execute it in Stata with File -> Do. Click the Break button to stop execution.

Please feel free to merge this into the earlier answer if that is more appropriate.

Diogenes