On Monday 15 March 2004 18:17, Brian F. Feldman wrote: > Peter Schultz <pmes_at_bis.midco.net> wrote: > > Taku YAMAMOTO wrote: > > > Unfortunately, due to over-optimization in sched_switch(), SCHED_ULE > > > doesn't give reasonable CPU time to the threads which are using > > > scheduler activation. > > > > > > Detailed analisis is described in my previous message posted to > > > current_at_: "SCHED_ULE sometimes puts P_SA processes into ksq_next > > > unnecessarily" <20040213063139.71298ea9.taku_at_cent.saitama-u.ac.jp> > > > or > > > http://docs.freebsd.org/cgi/mid.cgi?20040213063139.71298ea9.taku > > > , which didn't get broader audience :( > > > > > > Until the problem is fully addressed, I will propose following patch > > > to be applied. (the least intrusive one attached in the former message) > > > > This patch improves interactivity under heavy load very much. > > My system is VERY well-behaved using this change. (It's 2xSMP). Here too on an SMP system. This is purely subjective though - I haven't done any actual benchmarking. Any chance of this patch getting in the tree? Arjan
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:37:47 UTC