Re: [PATCH v2 0/3] Clarify abstract scale usage for power values in Energy Model, EAS and IPA

From: Doug Anderson <dianders@chromium.org>
To: Quentin Perret <qperret@google.com>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>,
	"Rafael J. Wysocki" <rafael@kernel.org>,
	Lukasz Luba <lukasz.luba@arm.com>,
	"Rafael J. Wysocki" <rjw@rjwysocki.net>,
	Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
	Linux PM <linux-pm@vger.kernel.org>,
	"open list:DOCUMENTATION" <linux-doc@vger.kernel.org>,
	"devicetree@vger.kernel.org" <devicetree@vger.kernel.org>,
	Rob Herring <robh+dt@kernel.org>,
	Amit Kucheria <amitk@kernel.org>,
	Jonathan Corbet <corbet@lwn.net>,
	Dietmar Eggemann <Dietmar.Eggemann@arm.com>,
	Matthias Kaehlcke <mka@chromium.org>,
	"Nayak, Rajendra" <rnayak@codeaurora.org>
Subject: Re: [PATCH v2 0/3] Clarify abstract scale usage for power values in Energy Model, EAS and IPA
Date: Fri, 16 Oct 2020 07:36:03 -0700	[thread overview]
Message-ID: <CAD=FV=VXt_JdYjPdTSJPWpXR2Owz096WhzYrZcQZ8-znmQbkLg@mail.gmail.com> (raw)
In-Reply-To: <20201016130905.GA2426638@google.com>

Hi,

On Fri, Oct 16, 2020 at 6:09 AM Quentin Perret <qperret@google.com> wrote:
>
> On Friday 16 Oct 2020 at 14:50:29 (+0200), Daniel Lezcano wrote:
> > On 16/10/2020 14:18, Quentin Perret wrote:
> > > On Friday 16 Oct 2020 at 13:48:33 (+0200), Daniel Lezcano wrote:
> > >> If the SCMI is returning abstract numbers, the thermal IPA governor will
> > >> use these numbers as a reference to mitigate the temperature at the
> > >> specified sustainable power which is expressed in mW in the DT. So it
> > >> does not work and we can not detect such conflict.
> > >>
> > >> That is why I'm advocating to keep mW for the energy model and make the
> > >> SCMI and DT power numbers incompatible.
> > >
> > > I think it's fair to say SCMI-provided number should only be compared to
> > > other SCMI-provided numbers, so +1 on that. But what I don't understand
> > > is why specifying the EM in mW helps with that?
> >
> > It is already specified in mW. I'm just saying to not add the
> > 'scale'/'abstract'/'bogoWatt' in the documentation.
> >
> > > Can we not let the providers specify the unit?
> >
> > Yes, it is possible but the provider must give the 'unit' and the energy
> > model must store this information along with the "power" numbers, so we
> > can compare apple with apple.
> >
> > Today, the energy model is using the mW unit only and the providers are
> > not telling the 'unit', so both are missing.
> >
> > Because both are missing, it does not make sense to talk about
> > 'abstract' values in the energy model documentation until the above is
> > fixed.
>
> Right, so that sounds like a reasonable way forward with this series.
>
> Lukasz would you be able to re-spin this with a first patch that allows
> the EM provider to specify a unit? And perhaps we could use Doug's idea
> for the sustained power DT binding and allow specifying a unit
> explicitly there too, so we're sure to compare apples with apples.

The one issue that I started with, though, is that I wanted to be able
to specify "sustainable-power" for a board in the device tree.  Unless
you think you'll convince Rob that it's OK to provide a "units"
property in the device tree then just adding a "units" to the API
won't help us because you'll still be stuck mixing/matching with a
value based in mW, right?  ...or are you suggesting that the
board-specific value "sustainable-power" would also have to come from
SCMI?  That would be pretty annoying.

-Doug