Re: fast path cycle muncher (vmstat: make vmstat_updater deferrable again and shut down on idle)

[Christoph Lameter]

On Tue, 26 Jan 2016, Mike Galbraith wrote:

> > 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?

Well if it does not pay it there then it is going to pay for it later when
the vmstat_updater needs to cause a wakeup to fold the data. Also not
folding the vmstat updates increases the time that minor updates are
deferred and thus impacts the accuracy of the global vm counters. That
seems to have been important recently.