Received-SPF: pass (google.com: domain of linux-kernel+bounces-87750-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45e3:2400::1 as permitted sender) client-ip=2604:1380:45e3:2400::1;
Precedence: bulk
MIME-Version: 1.0
References: <20240229001806.4158429-1-irogers@google.com> <20240229001806.4158429-2-irogers@google.com>
 <28bc7ea0-ff3e-48a5-a8b3-20cc52728a5f@linux.intel.com>
In-Reply-To: <28bc7ea0-ff3e-48a5-a8b3-20cc52728a5f@linux.intel.com>
From: Ian Rogers <irogers@google.com>
Date: Thu, 29 Feb 2024 17:02:02 -0800
Message-ID: <CAP-5=fVQD91HjfhJ+iX_nhMUJX91ZgJ_havGg=Ya+dMZ67C4yQ@mail.gmail.com>
Subject: Re: [PATCH v1 01/20] perf jevents: Add RAPL metrics for all Intel models
To: "Liang, Kan" <kan.liang@linux.intel.com>
Cc: Perry Taylor <perry.taylor@intel.com>, Samantha Alt <samantha.alt@intel.com>, 
	Caleb Biggers <caleb.biggers@intel.com>, Weilin Wang <weilin.wang@intel.com>, 
	Edward Baker <edward.baker@intel.com>, Andi Kleen <ak@linux.intel.com>, 
	Peter Zijlstra <peterz@infradead.org>, Ingo Molnar <mingo@redhat.com>, 
	Arnaldo Carvalho de Melo <acme@kernel.org>, Namhyung Kim <namhyung@kernel.org>, 
	Mark Rutland <mark.rutland@arm.com>, 
	Alexander Shishkin <alexander.shishkin@linux.intel.com>, Jiri Olsa <jolsa@kernel.org>, 
	Adrian Hunter <adrian.hunter@intel.com>, John Garry <john.g.garry@oracle.com>, 
	Jing Zhang <renyu.zj@linux.alibaba.com>, Thomas Richter <tmricht@linux.ibm.com>, 
	James Clark <james.clark@arm.com>, linux-kernel@vger.kernel.org, 
	linux-perf-users@vger.kernel.org, Stephane Eranian <eranian@google.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Thu, Feb 29, 2024 at 12:59=E2=80=AFPM Liang, Kan <kan.liang@linux.intel.=
com> wrote:
>
>
>
> On 2024-02-28 7:17 p.m., Ian Rogers wrote:
> > Add a 'cpu_power' metric group that computes the power consumption
> > from RAPL events if they are present.
> >
> > Signed-off-by: Ian Rogers <irogers@google.com>
> > ---
> >  tools/perf/pmu-events/intel_metrics.py | 45 ++++++++++++++++++++++++--
> >  1 file changed, 42 insertions(+), 3 deletions(-)
> >
> > diff --git a/tools/perf/pmu-events/intel_metrics.py b/tools/perf/pmu-ev=
ents/intel_metrics.py
> > index 4fbb31c9eccd..5827f555005f 100755
> > --- a/tools/perf/pmu-events/intel_metrics.py
> > +++ b/tools/perf/pmu-events/intel_metrics.py
> > @@ -1,9 +1,10 @@
> >  #!/usr/bin/env python3
> >  # SPDX-License-Identifier: (LGPL-2.1 OR BSD-2-Clause)
> > -from metric import (JsonEncodeMetric, JsonEncodeMetricGroupDescription=
s, LoadEvents,
> > -                    MetricGroup)
> > +from metric import (d_ratio, has_event, Event, JsonEncodeMetric, JsonE=
ncodeMetricGroupDescriptions,
> > +                    LoadEvents, Metric, MetricGroup, Select)
> >  import argparse
> >  import json
> > +import math
> >  import os
> >
> >  parser =3D argparse.ArgumentParser(description=3D"Intel perf json gene=
rator")
> > @@ -14,7 +15,45 @@ args =3D parser.parse_args()
> >  directory =3D f"{os.path.dirname(os.path.realpath(__file__))}/arch/x86=
/{args.model}/"
> >  LoadEvents(directory)
> >
> > -all_metrics =3D MetricGroup("",[])
> > +interval_sec =3D Event("duration_time")
> > +
> > +def Rapl() -> MetricGroup:
> > +  """Processor socket power consumption estimate.
> > +
> > +  Use events from the running average power limit (RAPL) driver.
> > +  """
> > +  # Watts =3D joules/second
> > +  pkg =3D Event("power/energy\-pkg/")
> > +  cond_pkg =3D Select(pkg, has_event(pkg), math.nan)
> > +  cores =3D Event("power/energy\-cores/")
> > +  cond_cores =3D Select(cores, has_event(cores), math.nan)
> > +  ram =3D Event("power/energy\-ram/")
> > +  cond_ram =3D Select(ram, has_event(ram), math.nan)
> > +  gpu =3D Event("power/energy\-gpu/")
> > +  cond_gpu =3D Select(gpu, has_event(gpu), math.nan)
> > +  psys =3D Event("power/energy\-psys/")
> > +  cond_psys =3D Select(psys, has_event(psys), math.nan)
> > +  scale =3D 2.3283064365386962890625e-10
> > +  metrics =3D [
> > +      Metric("cpu_power_pkg", "",
> > +             d_ratio(cond_pkg * scale, interval_sec), "Watts"),
> > +      Metric("cpu_power_cores", "",
> > +             d_ratio(cond_cores * scale, interval_sec), "Watts"),
> > +      Metric("cpu_power_ram", "",
> > +             d_ratio(cond_ram * scale, interval_sec), "Watts"),
> > +      Metric("cpu_power_gpu", "",
> > +             d_ratio(cond_gpu * scale, interval_sec), "Watts"),
> > +      Metric("cpu_power_psys", "",
> > +             d_ratio(cond_psys * scale, interval_sec), "Watts"),
> > +  ]
> > +
> > +  return MetricGroup("cpu_power", metrics,
> > +                     description=3D"Processor socket power consumption=
 estimates")
>
> As far as I know, the RAPL counters are to monitor energy consumption
> across different domains. The scope may not always be a socket. I think
> the description may brings confusions.
> Maybe we just call it "RAPL power consumption estimates", or "Running
> Average Power Limit (RAPL) power consumption estimates".

Ack. Will fix in v2.

Thanks,
Ian

> Thanks,
> Kan
> > +
> > +
> > +all_metrics =3D MetricGroup("", [
> > +    Rapl(),
> > +])
> >
> >  if args.metricgroups:
> >    print(JsonEncodeMetricGroupDescriptions(all_metrics))