Re: [PATCH v5 10/12] sched: get CPU's utilization statistic

From: Vincent Guittot <vincent.guittot@linaro.org>
To: Morten Rasmussen <morten.rasmussen@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>,
	Ingo Molnar <mingo@kernel.org>,
	linux-kernel <linux-kernel@vger.kernel.org>,
	Preeti U Murthy <preeti@linux.vnet.ibm.com>,
	Russell King - ARM Linux <linux@arm.linux.org.uk>,
	LAK <linux-arm-kernel@lists.infradead.org>,
	Rik van Riel <riel@redhat.com>, Mike Galbraith <efault@gmx.de>,
	Nicolas Pitre <nicolas.pitre@linaro.org>,
	"linaro-kernel@lists.linaro.org" <linaro-kernel@lists.linaro.org>,
	Daniel Lezcano <daniel.lezcano@linaro.org>,
	Dietmar Eggemann <Dietmar.Eggemann@arm.com>
Subject: Re: [PATCH v5 10/12] sched: get CPU's utilization statistic
Date: Tue, 16 Sep 2014 15:43:18 -0700	[thread overview]
Message-ID: <CAKfTPtCTxQNQfusJuQhGy7ahDFWjwitQOZHU89O0wyy5ZcNO9A@mail.gmail.com> (raw)
In-Reply-To: <20140915194518.GC28083@e103687>

On 15 September 2014 12:45, Morten Rasmussen <morten.rasmussen@arm.com> wrote:
> On Thu, Sep 11, 2014 at 03:04:44PM +0100, Peter Zijlstra wrote:
>> On Thu, Sep 11, 2014 at 03:07:52PM +0200, Vincent Guittot wrote:
>> > > Also I'm not entirely sure I like the usage, utilization names/metrics.
>> > > I would suggest to reverse them. Call the pure running number
>> > > 'utilization' and this scaled with capacity 'usage' or so.
>> >
>> > ok. i can invert 'usage' and 'utilization', which will give
>> >
>> > s/get_cpu_utilization/get_cpu_usage/
>> > s/sgs->group_utilization/sgs->group_usage/
>
> The confusion will have new dimensions added when we introduce
> scale-invariance too. Then the running number is already scaled by the
> current P-state compute capacity. But I don't have any better
> suggestions.
>
>> > s/cfs.usage_load_avg/cfs.utilization_load_avg/
>
> I don't like using "load" for unweighted metrics. I associate load with
> something that may be weighted by priority like load_avg_contrib, and
> utilization with pure cpu utilization as in how many cycles is spend on
> a particular task. I called it "usage_util_avg" in my own patches, but
> "util_avg" might be better if we agree that utilization == usage.

ok. so i don't have the same definition than you. IMHO, load should be
used for figures that have used the average of the geometric series
used in the per entity load tracking more than the fact that we weight
the figures with priority

>
>> > s/se->avg.usage_avg_contrib/se->avg.utilization_avg_contrib
>
> util_avg_contrib maybe to keep it shorter.
>
>> > s/__update_task_entity_usage/__update_task_entity_utilization
>> > s/__update_entity_usage_avg_contrib/__update_entity_utilization_avg_contrib
>
> Maybe use "util" here as well?

I agree that utilization can be a bit too long but util sounds a bit
too short ans we loose the utilization meaning. so we could use
activity instead of utilization

Nevertheless, the most important is that we find a common definition convention

Would the following proposal be ok ?

s/get_cpu_utilization/get_cpu_usage/
s/sgs->group_utilization/sgs->group_usage/
s/cfs.usage_load_avg/cfs.activity_load_avg/
s/se->avg.usage_avg_contrib/se->avg.activity_avg_contrib
s/__update_task_entity_usage/__update_task_entity_activity
s/__update_entity_usage_avg_contrib/__update_entity_activity_avg_contrib

Vincent
>
> Morten
>