On Monday, December 17, 2012 10:05:29 AM Dang Nguyen wrote:
> I am confused over the results I obtain when trying to measure the
> execution time of regular path queries in a model.
> Essentially, my codes look as follows:
>
> // start of code
> // generate Jena model
> ...
> // generate queryString for regular path query
> ...
>
> // execute the query on the model i = 10 times
>
> for (int i = 0; i < 10; i++){
> long startTime = System.nanoTime(); // start timer
> Query q = QueryFactory.create(queryString);
> QueryExecution qexec= QueryExecutionFactory.create( q, model );
> ResultSet rs= qexec.execSelect();
> long endTime = System.nanoTime(); // end timer
You haven't pulled the results, so the query might still be executing --
in fact it might not have started yet at all.
> long duration = endTime - startTime;
> System.out.println(duration);
> }// end of code
>
> The results I get (in nanosec) are:
> 521328304
> 1606834
> 2182409
> 1745501
> 1310299
> 1244592
> 1240752
> 1256112
> 1242032
> 4153618
>
> What I don't understand here is the first result always seems to be way
> larger than the rest and also the tenth one always seems to be slightly
> different too (from running with different i values).
I'd expect the long first result is the classloading of all the classes
needed to do the SPARQL query.
The long tenth result might be the first garbage collection.
Chris
"Measurement is hard."
--
"I know it was late, but Mountjoy never bothers, /Archer's Goon/
so long as it's the full two thousand words."
Epimorphics Ltd, http://www.epimorphics.com
Registered address: Court Lodge, 105 High Street, Portishead, Bristol BS20 6PT
Epimorphics Ltd. is a limited company registered in England (number 7016688)
