Received-SPF: pass (google.com: domain of linux-kernel+bounces-174143-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.199.223 as permitted sender) client-ip=147.75.199.223;
Precedence: bulk
MIME-Version: 1.0
References: <20240508060427.417-1-ravi.bangoria@amd.com>
In-Reply-To: <20240508060427.417-1-ravi.bangoria@amd.com>
From: Namhyung Kim <namhyung@kernel.org>
Date: Wed, 8 May 2024 22:08:09 -0700
Message-ID: <CAM9d7cgFKcHXeTXxax7GRCK__0U3HUnG3Ls09GpnD8FipyJk1A@mail.gmail.com>
Subject: Re: [RFC 0/4] perf sched: Introduce schedstat tool
To: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: peterz@infradead.org, mingo@redhat.com, acme@kernel.org, 
	irogers@google.com, swapnil.sapkal@amd.com, mark.rutland@arm.com, 
	alexander.shishkin@linux.intel.com, jolsa@kernel.org, rostedt@goodmis.org, 
	vincent.guittot@linaro.org, bristot@redhat.com, adrian.hunter@intel.com, 
	james.clark@arm.com, kan.liang@linux.intel.com, gautham.shenoy@amd.com, 
	kprateek.nayak@amd.com, juri.lelli@redhat.com, yangjihong@bytedance.com, 
	linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, 
	santosh.shukla@amd.com, ananth.narayan@amd.com, sandipan.das@amd.com
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

Hi Ravi,

On Tue, May 7, 2024 at 11:05=E2=80=AFPM Ravi Bangoria <ravi.bangoria@amd.co=
m> wrote:
>
> MOTIVATION
> ----------
>
> Existing `perf sched` is quite exhaustive and provides lot of insights
> into scheduler behavior but it quickly becomes impractical to use for
> long running or scheduler intensive workload. For ex, `perf sched record`
> has ~7.77% overhead on hackbench (with 25 groups each running 700K loops
> on a 2-socket 128 Cores 256 Threads 3rd Generation EPYC Server), and it
> generates huge 56G perf.data for which perf takes ~137 mins to prepare
> and write it to disk [1].

Right, this is painful.

>
> Unlike `perf sched record`, which hooks onto set of scheduler tracepoints
> and generates samples on a tracepoint hit, `perf sched schedstat record`
> takes snapshot of the /proc/schedstat file before and after the workload,
> i.e. there is zero interference on workload run. Also, it takes very
> minimal time to parse /proc/schedstat, convert it into perf samples and
> save those samples into perf.data file. Result perf.data file is much
> smaller. So, overall `perf sched schedstat record` is much more light-
> weight compare to `perf sched record`.

Nice work!

>
> We, internally at AMD, have been using this (a variant of this, known as
> "sched-scoreboard"[2]) and found it to be very useful to analyse impact
> of any scheduler code changes[3][4].
>
> Please note that, this is not a replacement of perf sched record/report.
> The intended users of the new tool are scheduler developers, not regular
> users.

Great, I think it's very useful.

>
> USAGE
> -----
>
>   # perf sched schedstat record
>   # perf sched schedstat report

Hmm. I think we can remove the duplication in 'sched'. :)
Given you are thinking of taskstat, how about making it
'cpustat' instead?

Also I think it'd be easier if you also provide 'live' mode so that
users can skip record + report steps and run the workload
directly like uftrace does. :)

Something like this

  # perf sched cpustat  myworkload
  (result here ...)

Thanks,
Namhyung

>
> Note: Although perf schedstat tool supports workload profiling syntax
> (i.e. -- <workload> ), the recorded profile is still systemwide since
> the /proc/schedstat is a systemwide file.