| Subject | Re: [patch 6/7] sched: use rq->avg_event to resurrect nohz ratelimiting | From | Peter Zijlstra <> | Date | Wed, 23 Nov 2011 12:57:39 +0100 |
| |
On Tue, 2011-11-22 at 15:24 +0100, Mike Galbraith wrote: > Entering nohz at high frequency eats cycles, ratelimit it > just as we do idle_balance(), and for the same reason.
ideally we should have some per-arch information on how costly it actually is to enter/leave nohz.
Also, can you confirm, or otherwise run this by whoemever reported that power-usage regression the last time we tried this?
|