Re: fast path cycle muncher (vmstat: make vmstat_updater deferrable again and shut down on idle)
From: Mike Galbraith
Date: Tue Jan 26 2016 - 14:09:54 EST
On Tue, 2016-01-26 at 12:22 -0600, Christoph Lameter wrote:
> On Tue, 26 Jan 2016, Mike Galbraith wrote:
> > On Tue, 2016-01-26 at 10:26 -0600, Christoph Lameter wrote:
> > > On Tue, 26 Jan 2016, Mike Galbraith wrote:
> > >
> > > > > Why would the deferring cause this overhead?
> > > >
> > > > Because we schedule to idle cores aggressively, thus we may pop
> > > > in and
> > > > out of idle at high frequency.
> > >
> > > Whats the point of going idle if you have things to do soon?
> > When a task schedules off, how do you know it'll be back at all,
> > much
> > less soon?
> Ok so you are running an artificial benchmark that always gets the
> system running again when it decides to go idle?
The benchmark does not alter the cycle expenditure per event. The real
world will pay the toll less frequently than the artificial benchmark,
yes, but it will pay nonetheless, and for most, needlessly. Or?