Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <12275372.O9o76ZdvQC@kreacher>
In-Reply-To: <12275372.O9o76ZdvQC@kreacher>
From:   Doug Smythies <dsmythies@telus.net>
Date:   Mon, 14 Aug 2023 07:51:41 -0700
Message-ID: <CAAYoRsUYrE7gM28VCB0KrP2dCr7NmH2wuDnhT6h1D=S7go6fJA@mail.gmail.com>
Subject: Re: [RFT] [PATCH v2] cpuidle: menu: Skip tick_nohz_get_sleep_length()
 call in some cases
To:     "Rafael J. Wysocki" <rjw@rjwysocki.net>
Cc:     Linux PM <linux-pm@vger.kernel.org>,
        Anna-Maria Behnsen <anna-maria@linutronix.de>,
        Peter Zijlstra <peterz@infradead.org>,
        LKML <linux-kernel@vger.kernel.org>,
        Frederic Weisbecker <frederic@kernel.org>,
        Kajetan Puchalski <kajetan.puchalski@arm.com>,
        Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>,
        Doug Smythies <dsmythies@telus.net>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Thu, Aug 10, 2023 at 11:38=E2=80=AFAM Rafael J. Wysocki <rjw@rjwysocki.n=
et> wrote:
>
> From: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
> Subject: [PATCH] cpuidle: menu: Skip tick_nohz_get_sleep_length() call in=
 some cases
>
> Because the cost of calling tick_nohz_get_sleep_length() may increase
> in the future, reorder the code in menu_select() so it first uses the
> statistics to determine the expected idle duration.  If that value is
> higher than RESIDENCY_THRESHOLD_NS, tick_nohz_get_sleep_length() will
> be called to obtain the time till the closest timer and refine the
> idle duration prediction if necessary.
>
> This causes the governor to always take the full overhead of
> get_typical_interval() with the assumption that the cost will be
> amortized by skipping the tick_nohz_get_sleep_length() call in the
> cases when the predicted idle duration is relatively very small.
>
> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

I ran the same tests as I did for the teo governor changes.
Nothing of significance to report (well, some minor improvements
in records per second at the long interval end of the sleeping
ebizzy test, with a corresponding slight increase in processor
package power).

Tested-by: Doug Smythies <dsmythies@telus.net>

> ---
>
> v1 -> v2: Add missing max check to get_typical_interval().
>

... Doug