lkml.org 
[lkml]   [2021]   [Jul]   [5]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
SubjectRe: [PATCH 3/3] PM: EM: Increase energy calculation precision
From
Date
On 25/06/2021 17:26, Lukasz Luba wrote:
> The Energy Model (EM) provides useful information about device power in
> each performance state to other subsystems like: Energy Aware Scheduler
> (EAS). The energy calculation in EAS does arithmetic operation based on
> the EM em_cpu_energy(). Current implementation of that function uses
> em_perf_state::cost as a pre-computed cost coefficient equal to:
> cost = power * max_frequency / frequency.
> The 'power' is expressed in milli-Watts (or in abstract scale).
>
> There are corner cases then the EAS energy calculation for two Performance
^^^^^^^^^^^^

Again, an easy to understand example to describe in which situation this
change would bring a benefit would help.

> Domains (PDs) return the same value, e.g. 10mW. The EAS compares these
> values to choose smaller one. It might happen that this values are equal
> due to rounding error. In such scenario, we need better precision, e.g.
> 10000 times better. To provide this possibility increase the precision on
> the em_perf_state::cost.
>
> This patch allows to avoid the rounding to milli-Watt errors, which might
> occur in EAS energy estimation for each Performance Domains (PD). The
> rounding error is common for small tasks which have small utilization
> values.

What's the influence of the CPU utilization 'cpu_util_next()' here?

compute_energy()
em_cpu_energy()
return ps->cost * sum_util / scale_cpu
^^^^^^^^
> The rest of the EM code doesn't change, em_perf_state::power is still
> expressed in milli-Watts (or in abstract scale). Thus, all existing
> platforms don't have to change their reported power. The same applies to

Not only existing platforms since there are no changes. So why
highlighting `existing` here.?

> EM clients, like thermal or DTPM (they use em_perf_state::power).
>
> Reported-by: CCJ Yeh <CCj.Yeh@mediatek.com>
> Suggested-by: CCJ Yeh <CCj.Yeh@mediatek.com>
> Signed-off-by: Lukasz Luba <lukasz.luba@arm.com>
> ---
> include/linux/energy_model.h | 5 ++++-
> kernel/power/energy_model.c | 3 ++-
> 2 files changed, 6 insertions(+), 2 deletions(-)
>
> diff --git a/include/linux/energy_model.h b/include/linux/energy_model.h
> index 2016f5a706e0..91037dd57e61 100644
> --- a/include/linux/energy_model.h
> +++ b/include/linux/energy_model.h
> @@ -16,7 +16,10 @@
> * @power: The power consumed at this level (by 1 CPU or by a registered
> * device). It can be a total power: static and dynamic.
> * @cost: The cost coefficient associated with this level, used during
> - * energy calculation. Equal to: power * max_frequency / frequency
> + * energy calculation. Equal to:
> + power * 10000 * max_frequency / frequency
> + * To increase the energy estimation presision use different
> + * scale in this coefficient than in @power field.
> */
> struct em_perf_state {
> unsigned long frequency;
> diff --git a/kernel/power/energy_model.c b/kernel/power/energy_model.c
> index 0f4530b3a8cd..2724f0ac417d 100644
> --- a/kernel/power/energy_model.c
> +++ b/kernel/power/energy_model.c
> @@ -170,7 +170,8 @@ static int em_create_perf_table(struct device *dev, struct em_perf_domain *pd,
> /* Compute the cost of each performance state. */
> fmax = (u64) table[nr_states - 1].frequency;
> for (i = 0; i < nr_states; i++) {
> - table[i].cost = div64_u64(fmax * table[i].power,
> + u64 power_res = (u64)table[i].power * 10000;
> + table[i].cost = div64_u64(fmax * power_res,
> table[i].frequency);
> }
>
>

\
 
 \ /
  Last update: 2021-07-05 14:47    [W:0.733 / U:0.188 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site