Re: [PATCH v1 0/5] cpuidle: teo: Rework the idle state selection logic
From: Rafael J. Wysocki
Date: Wed Jul 28 2021 - 09:54:45 EST
On Tue, Jul 27, 2021 at 10:06 PM Doug Smythies <dsmythies@xxxxxxxxx> wrote:
> Hi Rafael,
> Further to my reply of 2021.07.04 on this, I have
> continued to work with and test this patch set.
> On 2021.06.02 11:14 Rafael J. Wysocki wrote:
> >This series of patches addresses some theoretical shortcoming in the
> > TEO (Timer Events Oriented) cpuidle governor by reworking its idle
> > state selection logic to some extent.
> > Patches [1-2/5] are introductory cleanups and the substantial changes are
> > made in patches [3-4/5] (please refer to the changelogs of these two
> > patches for details). The last patch only deals with documentation.
> > Even though this work is mostly based on theoretical considerations, it
> > shows a measurable reduction of the number of cases in which the shallowest
> > idle state is selected while it would be more beneficial to select a deeper
> > one or the deepest idle state is selected while it would be more beneficial to
> > select a shallower one, which should be a noticeable improvement.
> I am concentrating in the idle state 0 and 1 area.
> When I disable idle state 0, the expectation is its
> usage will fall to idle state 1. It doesn't.
> CPU: Intel(R) Core(TM) i5-10600K CPU @ 4.10GHz
> HWP: disabled
> CPU frequency scaling driver: intel_pstate, active
> CPU frequency scaling governor: performance.
> Idle configuration: As a COMETLAKE processor, with 4 idle states.
> Sample time for below: 1 minute.
> Workflow: Cross core named pipe token passing, 12 threads.
> Kernel 5.14-rc3: idle: teo governor
> All idle states enabled: PASS
> Processor: 97 watts
> Idle state 0 entries: 811151
> Idle state 1 entries: 140300776
> Idle state 2 entries: 889
> Idle state 3 entries: 8
> Idle state 0 disabled: FAIL <<<<<
> Processor: 96 watts
> Idle state 0 entries: 0
> Idle state 1 entries: 65599283
> Idle state 2 entries: 364399
> Idle state 3 entries: 65112651
This looks odd.
Thanks for the report, I'll take a look at this.