Re: [RFC PATCH v3 2/2] cpuidle: teo: Introduce util-awareness

[Doug Smythies]

On Mon, Oct 31, 2022 at 11:24 PM Doug Smythies <dsmythies@xxxxxxxxx> wrote:
>
> Hi Kajetan,
>
> On Mon, Oct 31, 2022 at 5:14 AM Kajetan Puchalski
> <kajetan.puchalski@xxxxxxx> wrote:
>
> ... [delete some]...
>
> >  /**
> >   * teo_update - Update CPU metrics after wakeup.
> >   * @drv: cpuidle driver containing state data.
> > @@ -303,7 +359,9 @@ static int teo_select(struct cpuidle_driver *drv, struct cpuidle_device *dev,
> >         int i;
> >
> >         if (dev->last_state_idx >= 0) {
> > -               teo_update(drv, dev);
> > +               /* don't update metrics if the cpu was utilized during the last sleep */
> > +               if (!cpu_data->utilized)
> > +                       teo_update(drv, dev);
> >                 dev->last_state_idx = -1;
> >         }
>
> Ignoring the metrics is not the correct thing to do.
> Depending on the workflow, it can severely bias the idle states deeper
> than they should be because most of the needed information to select
> the appropriate shallow state is tossed out.
>
> Example 1:
> 2 pairs of ping pongs = 4 threads
> Parameters chosen such that idle state 2 would be a most used state.

Sorry, typo, I meant idle state 1 would be most used.

> CPU frequency governor: Schedutil.
> CPU frequency scaling driver: intel_cpufreq.
> HWP: Disabled
> Processor: i5-10600K (6 cores 12 cpus).
> Kernel: 6.1-rc3
> Run length: 1e8 cycles
> Idle governor:
> teo: 11.73 uSecs/loop ; idle state 1 ~3.5e6 exits/sec
> menu: 12.1 uSecs/loop ; idle state 1 ~3.3e6 exits/sec
> util-v3: 15.2 uSecs/loop ; idle state 1 ~200 exits/sec
> util-v4: 11.63 uSecs/loop ; idle state 1 ~3.5e6 exits/sec
>
> Where util-v4 is the same as this patch (util-v3) with the above code reverted.
>
> Note: less time per loop is better.
>
> Example 2: Same but parameters selected such that idle state 0 would
> be a most used idle state.
> Run Length: 4e8 cycles
> Idle governor:
> teo: 3.1 uSecs/loop ; idle state 0 ~1.2e6 exits/sec
> menu:  3.1 uSecs/loop ; idle state 0 ~1.3e6 exits/sec
> util-v3: 5.1 uSecs/loop ; idle state 0 ~4 exits/sec
> util-v4: ? uSecs/loop ; idle state 0 ~1.2e6 exits/sec (partial result)

util-v4: 3.15 uSecs/loop ; idle state 0 ~1.2e6 exits/sec

For the above tests we do not expect teo-util to have much impact.

For completeness:

Test 3: Same but parameters selected such that idle states 2 and 3
would be used the most.
Run Length: 3.42e5 cycles
CPU frequency scaling governor: schedutil.
CPU frequency scaling driver: intel_cpufreq.
Idle governor:
teo: 4005 uSecs/loop ; IS 2: 1917 IS 3: 107.4 exits/sec
menu: 3113 uSecs/loop ;  IS 2: 1020 IS 3: 1576 exits/sec
util-v3: 3457 uSecs/loop ;  IS 2: 1139 IS 3: 1000 exits/sec
util-v4: 3526 uSecs/loop ;  IS 2: 2029 IS 3: 109 exits/sec

Now, things are very noisy with the schedutil governor, so...

Test 4: Same as test 3 except for frequency scaling governor.
Run Length: 3.42e5 cycles
CPU frequency scaling governor: performance.
CPU frequency scaling driver: intel_pstate.
Idle governor:
teo: 2688 uSecs/loop ; IS 2: 2489 IS 3: 16 exits/sec
menu: 2596 uSecs/loop ;  IS 2: 865 IS 3: 2005 exits/sec
util-v3: 2766 uSecs/loop ;  IS 2: 1049 IS 3: 1394 exits/sec
util-v4: 2756 uSecs/loop ;  IS 2: 2440 IS 3: 24 exits/sec

... Doug