Can you run the 2.4 benchmark code against a 2.9 jar? Not sure if this can be done, but we can at least detect where's the change coming from.
On Wed, Aug 5, 2009 at 11:43 PM, Mark Miller <markrmil...@gmail.com> wrote: > Its the indexing it appears. Its very weird. > > With 2.4, during index, CPU keeps dropping to almost 0, and wa hits 20-30% > for long stretches. Not much CPU is being used. > > With 2.9, wa stays under 1% and CPU is pretty much always 100-300%. > > - Mark > > > Shai Erera wrote: > >> Does it measure only indexing, or search as well? I think 2.9 does have >> some good improvements to the search side, especially for sorting. >> >> Maybe it's not "one big thing" that's causing this improvement, but rather >> the aggregation of several "smaller things"? >> >> On Wed, Aug 5, 2009 at 11:35 PM, Michael McCandless < >> luc...@mikemccandless.com <mailto:luc...@mikemccandless.com>> wrote: >> >> Hmm -- looks like 2.9 defaulted to autoCommit=false, but that can't >> explain such a big difference. >> >> Mike >> >> On Wed, Aug 5, 2009 at 4:19 PM, Mark Miller<markrmil...@gmail.com >> <mailto:markrmil...@gmail.com>> wrote: >> > Michael McCandless wrote: >> >> >> >> Was the resulting index size approximately the same? >> >> >> >> >> > >> > lucene 2.9 >> > >> > total 24480 >> > -rw-r--r-- 1 mark mark 18548115 2009-08-05 15:46 _0.fdt >> > -rw-r--r-- 1 mark mark 160004 2009-08-05 15:46 _0.fdx >> > -rw-r--r-- 1 mark mark 47 2009-08-05 15:46 _5m.fnm >> > -rw-r--r-- 1 mark mark 2850871 2009-08-05 15:46 _5m.frq >> > -rw-r--r-- 1 mark mark 20004 2009-08-05 15:46 _5m.nrm >> > -rw-r--r-- 1 mark mark 2419839 2009-08-05 15:46 _5m.prx >> > -rw-r--r-- 1 mark mark 14300 2009-08-05 15:46 _5m.tii >> > -rw-r--r-- 1 mark mark 1026607 2009-08-05 15:46 _5m.tis >> > -rw-r--r-- 1 mark mark 264 2009-08-05 15:46 segments_2 >> > -rw-r--r-- 1 mark mark 20 2009-08-05 15:46 segments.gen >> > >> > >> > lucene 2.4 >> > >> > total 24636 >> > -rw-r--r-- 1 mark mark 18585922 2009-08-05 16:11 _5m.fdt >> > -rw-r--r-- 1 mark mark 160004 2009-08-05 16:11 _5m.fdx >> > -rw-r--r-- 1 mark mark 42 2009-08-05 16:11 _5m.fnm >> > -rw-r--r-- 1 mark mark 2875277 2009-08-05 16:11 _5m.frq >> > -rw-r--r-- 1 mark mark 100004 2009-08-05 16:11 _5m.nrm >> > -rw-r--r-- 1 mark mark 2439839 2009-08-05 16:11 _5m.prx >> > -rw-r--r-- 1 mark mark 14329 2009-08-05 16:11 _5m.tii >> > -rw-r--r-- 1 mark mark 1026623 2009-08-05 16:11 _5m.tis >> > -rw-r--r-- 1 mark mark 59 2009-08-05 16:11 segments_3 >> > -rw-r--r-- 1 mark mark 20 2009-08-05 16:11 segments.gen >> > >> > -- >> > - Mark >> > >> > http://www.lucidimagination.com >> > >> > >> > >> > >> > >> --------------------------------------------------------------------- >> > To unsubscribe, e-mail: java-dev-unsubscr...@lucene.apache.org >> <mailto:java-dev-unsubscr...@lucene.apache.org> >> > For additional commands, e-mail: java-dev-h...@lucene.apache.org >> <mailto:java-dev-h...@lucene.apache.org> >> > >> > >> >> --------------------------------------------------------------------- >> To unsubscribe, e-mail: java-dev-unsubscr...@lucene.apache.org >> <mailto:java-dev-unsubscr...@lucene.apache.org> >> For additional commands, e-mail: java-dev-h...@lucene.apache.org >> <mailto:java-dev-h...@lucene.apache.org> >> >> >> > > -- > - Mark > > http://www.lucidimagination.com > > > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: java-dev-unsubscr...@lucene.apache.org > For additional commands, e-mail: java-dev-h...@lucene.apache.org > >
