Re: [PATCH v4 5/5] sched/fair: Track peak per-entity utilization

2016-09-01 Thread Morten Rasmussen
On Thu, Sep 01, 2016 at 12:51:36PM +0100, Patrick Bellasi wrote: > On 31-Aug 11:52, Morten Rasmussen wrote: > > When using PELT (per-entity load tracking) utilization to place tasks at > > wake-up using the decayed utilization (due to sleep) leads to > > under-estimation of true utilization of the

Re: [PATCH v4 5/5] sched/fair: Track peak per-entity utilization

2016-09-01 Thread Patrick Bellasi
On 31-Aug 11:52, Morten Rasmussen wrote: > When using PELT (per-entity load tracking) utilization to place tasks at > wake-up using the decayed utilization (due to sleep) leads to > under-estimation of true utilization of the task. This could mean > putting the task on a cpu with less available cap

[PATCH v4 5/5] sched/fair: Track peak per-entity utilization

2016-08-31 Thread Morten Rasmussen
When using PELT (per-entity load tracking) utilization to place tasks at wake-up using the decayed utilization (due to sleep) leads to under-estimation of true utilization of the task. This could mean putting the task on a cpu with less available capacity than is actually needed. This issue can be