15 Jul
2014
15 Jul
'14
9:53 a.m.
On 15 July 2014 11:32, Morten Rasmussen morten.rasmussen@arm.com wrote:
On Tue, Jul 15, 2014 at 10:27:19AM +0100, Vincent Guittot wrote:
On 11 July 2014 18:13, Morten Rasmussen morten.rasmussen@arm.com wrote:
[snip]
In this example using rq->avg leads to imbalance whereas unweighted load would not. Correct me if I missed anything.
You just miss to take into account how the imbalance is computed
I don't think so. I'm aware that the imbalance is calculated based on the runnable_load_avg of the cpus. But if you pick the wrong cpus to compare to begin with, it doesn't matter.
The average load of your sched_domain is 1 task per CPU so CPU1 will pull only 1 task and not 2
Morten