From: Quentin Perret <quentin.perret@arm.com>
To: Juri Lelli <juri.lelli@redhat.com>
Cc: peterz@infradead.org, rjw@rjwysocki.net,
gregkh@linuxfoundation.org, linux-kernel@vger.kernel.org,
linux-pm@vger.kernel.org, mingo@redhat.com,
dietmar.eggemann@arm.com, morten.rasmussen@arm.com,
chris.redpath@arm.com, patrick.bellasi@arm.com,
valentin.schneider@arm.com, vincent.guittot@linaro.org,
thara.gopinath@linaro.org, viresh.kumar@linaro.org,
tkjos@google.com, joelaf@google.com, smuckle@google.com,
adharmap@quicinc.com, skannan@quicinc.com,
pkondeti@codeaurora.org, edubezval@gmail.com,
srinivas.pandruvada@linux.intel.com, currojerez@riseup.net,
javi.merino@kernel.org
Subject: Re: [RFC PATCH v3 09/10] sched/fair: Select an energy-efficient CPU on task wake-up
Date: Fri, 8 Jun 2018 17:26:12 +0100 [thread overview]
Message-ID: <20180608162612.GA17720@e108498-lin.cambridge.arm.com> (raw)
In-Reply-To: <20180608115928.GC16089@localhost.localdomain>
On Friday 08 Jun 2018 at 13:59:28 (+0200), Juri Lelli wrote:
> On 08/06/18 12:19, Quentin Perret wrote:
> > On Friday 08 Jun 2018 at 12:24:46 (+0200), Juri Lelli wrote:
> > > Hi,
> > >
> > > On 21/05/18 15:25, Quentin Perret wrote:
> > >
> > > [...]
> > >
> > > > +static int find_energy_efficient_cpu(struct task_struct *p, int prev_cpu)
> > > > +{
> > > > + unsigned long cur_energy, prev_energy, best_energy, cpu_cap, task_util;
> > > > + int cpu, best_energy_cpu = prev_cpu;
> > > > + struct sched_energy_fd *sfd;
> > > > + struct sched_domain *sd;
> > > > +
> > > > + sync_entity_load_avg(&p->se);
> > > > +
> > > > + task_util = task_util_est(p);
> > > > + if (!task_util)
> > > > + return prev_cpu;
> > > > +
> > > > + /*
> > > > + * Energy-aware wake-up happens on the lowest sched_domain starting
> > > > + * from sd_ea spanning over this_cpu and prev_cpu.
> > > > + */
> > > > + sd = rcu_dereference(*this_cpu_ptr(&sd_ea));
> > > > + while (sd && !cpumask_test_cpu(prev_cpu, sched_domain_span(sd)))
> > > > + sd = sd->parent;
> > > > + if (!sd)
> > > > + return -1;
> > >
> > > Shouldn't this be return prev_cpu?
> >
> > Well, you shouldn't be entering this function without an sd_ea pointer,
> > so this case is a sort of bug I think. By returning -1 I think we should
> > end-up picking a CPU using select_fallback_rq(), which sort of makes
> > sense ?
>
> I fear cpumask_test_cpu() and such won't be happy with a -1 arg.
> If it's a recoverable bug, I'd say return prev and WARN_ON_ONCE() ?
Hmmm, yes, prev + WARN_ON_ONCE is probably appropriate here then.
>
> > > > +
> > > > + if (cpumask_test_cpu(prev_cpu, &p->cpus_allowed))
> > > > + prev_energy = best_energy = compute_energy(p, prev_cpu);
> > > > + else
> > > > + prev_energy = best_energy = ULONG_MAX;
> > > > +
> > > > + for_each_freq_domain(sfd) {
> > > > + unsigned long spare_cap, max_spare_cap = 0;
> > > > + int max_spare_cap_cpu = -1;
> > > > + unsigned long util;
> > > > +
> > > > + /* Find the CPU with the max spare cap in the freq. dom. */
> > >
> > > I undestand this being a heuristic to cut some overhead, but shouldn't
> > > the model tell between packing vs. spreading?
> >
> > Ah, that's a very interesting one :-) !
> >
> > So, with only active costs of the CPUs in the model, we can't really
> > tell what's best between packing or spreading between identical CPUs if
> > the migration of the task doesn't change the OPP request.
> >
> > In a frequency domain, all the "best" CPU candidates for a task are
> > those for which we'll request a low OPP. When there are several CPUs for
> > which the OPP request will be the same, we just don't know which one to
> > pick from an energy standpoint, because we don't have other energy costs
> > (for idle states for ex) to break the tie.
> >
> > With this EM, the interesting thing is that if you assume that OPP
> > requests follow utilization, you are _guaranteed_ that the CPU with
> > the max spare capacity in a freq domain will always be among the best
> > candidates of this freq domain. And since we don't know how to
> > differentiate those candidates, why not using this one ?
> >
> > Yes, it _might_ be better from an energy standpoint to pack small tasks
> > on a CPU in order to let other CPUs go in deeper idle states. But that
> > also hurts your chances to go cluster idle. Which solution is the best ?
> > It depends, and we have no ways to tell with this EM.
> >
> > This approach basically favors cluster-packing, and spreading inside a
> > cluster. That should at least be a good thing for latency, and this is
> > consistent with the idea that most of the energy savings come from the
> > asymmetry of the system, and not so much from breaking the tie between
> > identical CPUs. That's also the reason why EAS is enabled only if your
> > system has SD_ASYM_CPUCAPACITY set, as we already discussed for patch
> > 05/10 :-).
> >
> > Does that make sense ?
>
> Yes, thanks for the explanation. It would probably make sense to copy
> and paste your text above somewhere in comment/doc for future ref.
OK, will do.
Thanks !
Quentin
next prev parent reply other threads:[~2018-06-08 16:26 UTC|newest]
Thread overview: 80+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-05-21 14:24 [RFC PATCH v3 00/10] Energy Aware Scheduling Quentin Perret
2018-05-21 14:24 ` [RFC PATCH v3 01/10] sched: Relocate arch_scale_cpu_capacity Quentin Perret
2018-05-21 14:24 ` [RFC PATCH v3 02/10] sched/cpufreq: Factor out utilization to frequency mapping Quentin Perret
2018-05-21 14:24 ` [RFC PATCH v3 03/10] PM: Introduce an Energy Model management framework Quentin Perret
2018-06-06 13:12 ` Dietmar Eggemann
2018-06-06 14:37 ` Quentin Perret
2018-06-06 15:20 ` Juri Lelli
2018-06-06 15:29 ` Quentin Perret
2018-06-06 16:26 ` Quentin Perret
2018-06-07 15:58 ` Dietmar Eggemann
2018-06-08 13:39 ` Javi Merino
2018-06-08 15:47 ` Quentin Perret
2018-06-09 8:24 ` Javi Merino
2018-06-06 16:47 ` Juri Lelli
2018-06-06 16:59 ` Quentin Perret
2018-06-07 14:44 ` Juri Lelli
2018-06-07 15:19 ` Quentin Perret
2018-06-07 15:55 ` Dietmar Eggemann
2018-06-08 8:25 ` Quentin Perret
2018-06-08 9:36 ` Juri Lelli
2018-06-08 10:31 ` Quentin Perret
2018-06-08 12:39 ` Dietmar Eggemann
2018-06-08 13:11 ` Quentin Perret
2018-06-08 16:39 ` Dietmar Eggemann
2018-06-08 17:02 ` Quentin Perret
2018-06-07 16:04 ` Juri Lelli
2018-06-07 17:31 ` Quentin Perret
2018-06-09 8:13 ` Javi Merino
2018-06-19 11:07 ` Peter Zijlstra
2018-06-19 12:35 ` Quentin Perret
2018-06-19 11:31 ` Peter Zijlstra
2018-06-19 12:40 ` Quentin Perret
2018-06-19 11:34 ` Peter Zijlstra
2018-06-19 12:58 ` Quentin Perret
2018-06-19 13:23 ` Peter Zijlstra
2018-06-19 13:38 ` Quentin Perret
2018-06-19 14:16 ` Peter Zijlstra
2018-06-19 14:21 ` Peter Zijlstra
2018-06-19 14:30 ` Peter Zijlstra
2018-06-19 14:23 ` Quentin Perret
2018-05-21 14:24 ` [RFC PATCH v3 04/10] PM / EM: Expose the Energy Model in sysfs Quentin Perret
2018-06-19 12:16 ` Peter Zijlstra
2018-06-19 13:06 ` Quentin Perret
2018-05-21 14:25 ` [RFC PATCH v3 05/10] sched/topology: Reference the Energy Model of CPUs when available Quentin Perret
2018-06-07 14:44 ` Juri Lelli
2018-06-07 16:02 ` Quentin Perret
2018-06-07 16:29 ` Juri Lelli
2018-06-07 17:26 ` Quentin Perret
2018-06-19 12:26 ` Peter Zijlstra
2018-06-19 13:24 ` Quentin Perret
2018-06-19 16:20 ` Peter Zijlstra
2018-06-19 17:13 ` Quentin Perret
2018-06-19 18:42 ` Peter Zijlstra
2018-06-20 7:58 ` Quentin Perret
2018-05-21 14:25 ` [RFC PATCH v3 06/10] sched: Add over-utilization/tipping point indicator Quentin Perret
2018-06-19 7:01 ` Pavan Kondeti
2018-06-19 10:26 ` Dietmar Eggemann
2018-05-21 14:25 ` [RFC PATCH v3 07/10] sched/fair: Introduce an energy estimation helper function Quentin Perret
2018-06-08 10:30 ` Juri Lelli
2018-06-19 9:51 ` Pavan Kondeti
2018-06-19 9:53 ` Quentin Perret
2018-05-21 14:25 ` [RFC PATCH v3 08/10] sched: Lowest energy aware balancing sched_domain level pointer Quentin Perret
2018-05-21 14:25 ` [RFC PATCH v3 09/10] sched/fair: Select an energy-efficient CPU on task wake-up Quentin Perret
2018-06-08 10:24 ` Juri Lelli
2018-06-08 11:19 ` Quentin Perret
2018-06-08 11:59 ` Juri Lelli
2018-06-08 16:26 ` Quentin Perret [this message]
2018-06-19 5:06 ` Pavan Kondeti
2018-06-19 7:57 ` Quentin Perret
2018-06-19 8:41 ` Pavan Kondeti
2018-05-21 14:25 ` [RFC PATCH v3 10/10] arch_topology: Start Energy Aware Scheduling Quentin Perret
2018-06-19 9:18 ` Pavan Kondeti
2018-06-19 9:40 ` Quentin Perret
2018-06-19 9:47 ` Juri Lelli
2018-06-19 10:02 ` Quentin Perret
2018-06-19 10:19 ` Juri Lelli
2018-06-19 10:25 ` Quentin Perret
2018-06-19 10:31 ` Juri Lelli
2018-06-19 10:49 ` Quentin Perret
2018-06-01 9:29 ` [RFC PATCH v3 00/10] " Quentin Perret
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20180608162612.GA17720@e108498-lin.cambridge.arm.com \
--to=quentin.perret@arm.com \
--cc=adharmap@quicinc.com \
--cc=chris.redpath@arm.com \
--cc=currojerez@riseup.net \
--cc=dietmar.eggemann@arm.com \
--cc=edubezval@gmail.com \
--cc=gregkh@linuxfoundation.org \
--cc=javi.merino@kernel.org \
--cc=joelaf@google.com \
--cc=juri.lelli@redhat.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-pm@vger.kernel.org \
--cc=mingo@redhat.com \
--cc=morten.rasmussen@arm.com \
--cc=patrick.bellasi@arm.com \
--cc=peterz@infradead.org \
--cc=pkondeti@codeaurora.org \
--cc=rjw@rjwysocki.net \
--cc=skannan@quicinc.com \
--cc=smuckle@google.com \
--cc=srinivas.pandruvada@linux.intel.com \
--cc=thara.gopinath@linaro.org \
--cc=tkjos@google.com \
--cc=valentin.schneider@arm.com \
--cc=vincent.guittot@linaro.org \
--cc=viresh.kumar@linaro.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).