From: Seaborne, A. <and...@hp...> - 2007-07-02 11:23:15
|
-------- Original Message -------- > From: Chris Bizer <mailto:ch...@bi...> > Date: 2 July 2007 10:49 >=20 > Hi Christian, >=20 > the benchmarks look really good! >=20 > I'm surprised too. Especially that we seam to beat Andy's new Jena SDB > engine which was especially implemented for speed. But I guess Andy > still has some configuration options in the back-hand which will make > his engine fly past RAP ;-) Finishing the coding being one of those options. SDB does pattern matching at the moment and, of these queries,the onlt = two patterns are { ?s ?p ?o } and {?s ?p ?o1 . ?s ?p ?o2 }. =20 All the queries have LIMIT so that is a significant thing being measured = SDB does not optimize LIMIT (or FILTER) yet. As a quick hack, using a = develop stream for SDB, I get .16s for the 100K triples set on SELECT * = {?s ?p ?o} LIMIT 10 Slightly faster if the DB server and client are = different machines. PostgreSQL. Load time: 100K triples: 5Ktriples/s 200K triples is a bit small side for SDB. Normally, that many triples = will end up in the database cache (fits in Jena mem graph as well) and = .1s is possibly JDBC time. Can SparqlEngineDb hold a subset of Uniprot, say the 100e6 subset of the = full 650e6 triples? We have SPARQL translations of the queries against = Uniprot. (The page is not clear whether this is Jena or SDB. It links to Jena. = SDB is mentioned only incidentally - note: SDB does not ship with the = Jena distribution.) Andy =20 >=20 > I have put links to the benchmark on the RAP project page. >=20 > Keep up the great work! >=20 > Cheers >=20 > Chris >=20 >=20 > -- > Chris Bizer > Freie Universit=E4t Berlin > Phone: +49 30 838 54057 > Mail: ch...@bi... > Web: www.bizer.de >=20 > ----- Original Message ----- > From: "Christian Weiske" <cw...@cw...> > To: "Sebastian Dietzold" <die...@in...>; = "S=F6ren > Auer" <au...@in...>=20 > Cc: <rdf...@li...> > Sent: Monday, July 02, 2007 8:51 AM > Subject: [Rdfapi-php-interest] SparqlEngineDb benchmark results >=20 >=20 > Hello, >=20 >=20 > I made some benchmarks comparing RAP's new SparqlEngineDb against > competitors like ARC, Jena and Redland.=20 > Results can be seen at > = http://cweiske.de/tagebuch/SPARQLer's%20best%20choice:%20SparqlEngineDb.h= tm >=20 > -- > Mit freundlichen Gr=FC=DFen/Regards, > Christian Weiske >=20 > = -------------------------------------------------------------------------= > This SF.net email is sponsored by DB2 Express Download DB2 Express C - > the FREE version of DB2 express and take control of your XML. No > limits. Just data. Click to get it now. =20 > http://sourceforge.net/powerbar/db2/ > _______________________________________________ > Rdfapi-php-interest mailing list > Rdf...@li... > https://lists.sourceforge.net/lists/listinfo/rdfapi-php-interest |