Date: Wed, 24 Oct 2007 11:34:34 -0400 From: Nick Evans <nevans@talkpoint.com> To: josh.carroll@gmail.com Cc: remy.nonnenmacher@activnetworks.com, Kris Kennaway <kris@freebsd.org>, freebsd-performance@freebsd.org, Bruce Evans <brde@optusnet.com.au> Subject: Re: ULE vs. 4BSD in RELENG_7 Message-ID: <20071024113434.326c3749@pleiades.nextvenue.com> In-Reply-To: <8cb6106e0710240639r20e03ce9w81ed3354338b7395@mail.gmail.com> References: <8cb6106e0710230902x4edf2c8eu2d912d5de1f5d4a2@mail.gmail.com> <471E343C.2040509@FreeBSD.org> <20071024171915.E84143@delplex.bde.org> <8cb6106e0710240639r20e03ce9w81ed3354338b7395@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, 24 Oct 2007 09:39:29 -0400 "Josh Carroll" <josh.carroll@gmail.com> wrote: > > 5-6% is a lot. ULE has some tuning for makeworld in -current, which > > for me reduced it to less than 1% slower than 4BSD (down from 5-10% > > slower), for the case of makeworld -j4 over nfs on a 2-CPU system with > > the sources pre-cached on the server and objects on a local file system, > > and extensive local tuning of makeworld, nfs and network drivers. I > > think the tuning in ULE was mainly for a 2-CPU system, because makeworld > > seemed to be very bad under ULE only with 2 CPUs. Apparently, it is also > > very bad with more CPUs. There are sysctls to modify the ULE tuning. > > I found Kris' document here: > > http://people.freebsd.org/~kris/scaling/mysql.html > > Which also mentions some ULE tuning. The sched_ule man page doesn't > mention what these are. Can someone point me to a document that > explains the sysctl tuning I can try? > > I imagine I would want to tweak one or more of these: > > kern.sched.preemption: 1 > kern.sched.topology: 0 > kern.sched.steal_thresh: 2 > kern.sched.steal_idle: 1 > kern.sched.steal_htt: 1 > kern.sched.balance_interval: 133 > kern.sched.balance: 1 > kern.sched.tryself: 1 > kern.sched.affinity: 3 > kern.sched.pick_pri: 1 > kern.sched.preempt_thresh: 64 > kern.sched.interact: 30 > kern.sched.slice: 13 > > But I'm not sure which to try. I did try setting kern.sched.pick_pri > to 0, but that hasn't helped the ffmpeg workload at least. > > Thanks, > Josh > _______________________________________________ > freebsd-performance@freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-performance > To unsubscribe, send any mail to > "freebsd-performance-unsubscribe@freebsd.org" kern.sched.steal_thresh is/was one of the more effective tuning sysctls. rev 1.205 of sched_ule had a change that was supposed to automatically adjust it based on the number of cores. Is this the same 8 core system as the other thread? In that case the commit dictates steal_thresh should be set to 3. Give that a try. Nick
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20071024113434.326c3749>