Re: [PATCH v2] sched/fair: update scale invariance of PELT

From: Vincent Guittot
Date: Fri Apr 14 2017 - 04:48:04 EST


On 13 April 2017 at 20:06, Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
> On Thu, Apr 13, 2017 at 04:59:15PM +0200, Vincent Guittot wrote:
>> On 13 April 2017 at 15:32, Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
>> > On Wed, Apr 12, 2017 at 01:28:58PM +0200, Peter Zijlstra wrote:
>> >
>> >> I still wonder about the whole !running vs !weight thing.,
>> >
>> > Ah, since we use this for both util _and_ load, we need !running &&
>> > !weight, and it so happens that !weight implies !running. Is that it?
>>
>> exactly
>> sorry, I should have started with that
>
> Damn, that just bring me around to wondering why running is the right
> condition to create lost-time.
>
> Because for runnable we want everything that has weight.

I have considered that the waiting time doesn't have to be scaled
unlike the running time of the runnable because waiting is the same
whatever the current capacity

>