Cool, let us know how things work out with the full dataset! Cheers,
/peter neubauer COO and Sales, Neo Technology GTalk: neubauer.peter Skype peter.neubauer Phone +46 704 106975 LinkedIn http://www.linkedin.com/in/neubauer Twitter http://twitter.com/peterneubauer http://www.neo4j.org - Your high performance graph database. http://www.tinkerpop.com - Processing for Internet-scale graphs. http://www.thoughtmade.com - Scandinavias coolest Bring-a-Thing party. On Thu, Mar 25, 2010 at 8:58 PM, Laurent Laborde <kerdez...@gmail.com> wrote: > my new ssd (Intel X25-V, 40GB, 114€) is finally installed. > i continue my collatz number crunching from 90 millions to 100 millions. > Result : i'm cpu-bound again (1 core at 100%) > > when the base was small enough to fit in ram it took ~10s to compute a > block of 10k. > now the db is ~22Go and it took more than a mn to compute a block of > 10k on the raptor 10krpm. > > i moved the base to the ssd, then started the computation again. > I'm back to ~10s per 10k. > > i'm running the windows monitor, while computing i have a mix of : > - 1.5 MB/s read > - 2.5 MB/s write > > The queue length on the ssd is at ~10% (while it was, of course, 100% > on the raptor). > > A side note : > When moving the DB from the raptor to the SSD, the ssd was the > bottlenack, but it make sense : > It's faster to sequentially read from a 10krpm than sequentially write > on a cheap SSD. :) > > conclusion : > TOTALLY ROCKS \o/ > > > -- > Laurent "ker2x" Laborde > Sysadmin & DBA at http://www.over-blog.com/ > _______________________________________________ > Neo mailing list > User@lists.neo4j.org > https://lists.neo4j.org/mailman/listinfo/user > _______________________________________________ Neo mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user
