> Eric Lowe wrote: > > If you're comparing against 10, make sure you have > the latest patches > > installed (particuarly the KU patch). There was a > lot of magic needed to > > get this platform to run well, and some of it is > only in the more recent > > kernel updates. > > > > - Eric > > > > > In addition, various driver fixes to allow the > relatively slow > individual Niagara CPUs to cooperatively handle the > interrupt > load from the Ethernet driver have gone into Nevada. > > Keep in mind that this benchmark apparently involves > serving up a > single 100 byte file of all NULLs to clients as fast > as possible; > all data is cached. > > > - Bart
Hi is there any further investigation about this performance test? What can be result of turning disk cache on as suggested in one of comments? Sorry for asking but unfortunatelly I have no access to Tx000.... Also, how it can be that Linux is running better on T1 than Solaris, which should (theoretically) work better on larger number of CPUs (cores)? Kind regards Jan Wiktorowicz This message posted from opensolaris.org _______________________________________________ perf-discuss mailing list perf-discuss@opensolaris.org
