Messages in this thread | | | Date | Mon, 28 Oct 2013 13:22:31 +0100 | From | Peter Zijlstra <> | Subject | Re: [Resend patch v8 0/13] use runnable load in schedule balance |
| |
On Mon, Oct 28, 2013 at 11:25:34AM +0100, Frederic Weisbecker wrote: > 2013/6/28 Peter Zijlstra <peterz@infradead.org>: > > On Fri, Jun 28, 2013 at 03:56:25AM -0700, Paul Turner wrote: > > > >> So this is actually an interesting idea, but don't think of it as > >> overweight. What "cfs_rq->blocked_load_avg / 2" means is actually > >> blocked_load_avg one period from now. This is interesting because it > >> makes the (reasonable) supposition that blocked load is not about to > >> immediately wake, but will continue to decay. > >> > >> Could you try testing the gvr_lb_tip branch at > >> git://git.kernel.org/pub/scm/linux/kernel/git/pjt/sched-tip.git ? > >> > >> It's an extension to your series that tries to improve some of the > >> cpu_load interactions in an alternate way to the above. > >> > >> It seems a little better on one and two-socket machines; but we > >> couldn't reproduce/compare to your best performance results since they > >> were taken on larger machines. > > > > Oh nice.. it does away with the entire cpu_load[] array thing. Just what > > Frederic needs for his NOHZ stuff as well -- he's currently abusing > > LB_BIAS for that. > > Hi guys, > > Is there any updates on the status of this work? I'm getting back on > fixing the cpu_load for full dynticks and this patchset was apparently > taking care of that.
I talked to PJT about this last week, he said Ben was looking (or going to look into) this sometime 'soon' iirc.
| |