Re: [PATCH RESEND] sched: prefer an idle cpu vs an idle sibling for BALANCE_WAKE

[Rik van Riel <riel@redhat.com>]

On 06/02/2015 01:12 PM, Josef Bacik wrote:

> But now that I re-read your response I think this is even more what you
> were worried about than less.
>
> Basically it comes down to if sd isn't set then we get shit performance.
>   I realize that this magic to find an idle cpu when sd is set is pretty
> heavy handed, but it's obviously helpful in our case.

Ingo and Peter appear to be worried that searching
through too many idle CPUs leads to bad performance.

Your test results seem to indicate that finding an
idle CPU really really helps performance.

There is a policy vs mechanism thing here. Ingo and Peter
are worried about the overhead in the mechanism of finding
an idle CPU.  Your measurements show that the policy of
finding an idle CPU is the correct one.

Can we make the policy change now, and optimize the
mechanism later?

> So let me ask this question.  When do we want to do the heavy handed
> search and when do we not?  With WAKE_AFFINE what is our ultimate goal
> vs the other SD's?  If we don't have an sd that matches our sd_flags
> what should we be doing, should we just go with whatever cpu we're on
> and carry on?  Thanks,
>
> Josef