[PATCH v4 00/18] Introduce runtime modifiable Energy Model
From: Lukasz Luba
Date: Mon Sep 25 2023 - 04:11:25 EST
Hi all,
This patch set adds a new feature which allows to modify Energy Model (EM)
power values at runtime. It will allow to better reflect power model of
a recent SoCs and silicon. Different characteristics of the power usage
can be leveraged and thus better decisions made during task placement in EAS.
It's part of feature set know as Dynamic Energy Model. It has been presented
and discussed recently at OSPM2023 [3]. This patch set implements the 1st
improvement for the EM.
The concepts:
1. The CPU power usage can vary due to the workload that it's running or due
to the temperature of the SoC. The same workload can use more power when the
temperature of the silicon has increased (e.g. due to hot GPU or ISP).
In such situation the EM can be adjusted and reflect the fact of increased
power usage. That power increase is due to static power
(sometimes called simply: leakage). The CPUs in recent SoCs are different.
We have heterogeneous SoCs with 3 (or even 4) different microarchitectures.
They are also built differently with High Performance (HP) cells or
Low Power (LP) cells. They are affected by the temperature increase
differently: HP cells have bigger leakage. The SW model can leverage that
knowledge.
2. It is also possible to change the EM to better reflect the currently
running workload. Usually the EM is derived from some average power values
taken from experiments with benchmark (e.g. Dhrystone). The model derived
from such scenario might not represent properly the workloads usually running
on the device. Therefore, runtime modification of the EM allows to switch to
a different model, when there is a need.
3. The EM can be adjusted after boot, when all the modules are loaded and
more information about the SoC is available e.g. chip binning. This would help
to better reflect the silicon characteristics. Thus, this EM modification
API allows it now. It wasn't possible in the past and the EM had to be
'set in stone'.
Some design details:
The internal mechanisms for the memory allocation are handled internally in the
EM. Kernel modules can just call the new API to update the EM data and the
new memory would be provided and owned by the EM. The EM memory is used by
EAS, which impacts those design decisions. The EM writers are protected by
a mutex. This new runtime modified EM table is protected using RCU mechanism,
which fits the current EAS hot path (which already uses RCU read lock).
The unregister API handles only non-CPU (e.g. GPU, ISP) devices and uses the
same mutex as EM modifiers to make sure the memory is safely freed.
More detailed explanation and background can be found in presentations
during LPC2022 [1][2] or in the documentation patches.
The time cost to update EM for 11 OPPs is listed below. It's roughly
1.5us per 1 OPP while doing this on Little CPU at max frequency (1.8GHz).
More detailed results:
(The 4 CPUs from top are the little (1.8MHz), than 2 Mid (2.2GHz) and
then 2 big (2.8GHz) (while cpu6 didn't run that code)
------------------------------------
Function Hit Time Avg s^2
-------- --- ---- --- ---
em_dev_update_perf_domain 3104 51236.39 us 16.506 us 75.344 us
Function Hit Time Avg s^2
-------- --- ---- --- ---
em_dev_update_perf_domain 1264 20768.15 us 16.430 us 62.257 us
Function Hit Time Avg s^2
-------- --- ---- --- ---
em_dev_update_perf_domain 1166 18632.95 us 15.980 us 70.707 us
Function Hit Time Avg s^2
-------- --- ---- --- ---
em_dev_update_perf_domain 770 12334.43 us 16.018 us 66.337 us
Function Hit Time Avg s^2
-------- --- ---- --- ---
em_dev_update_perf_domain 101 920.613 us 9.114 us 21.380 us
Function Hit Time Avg s^2
-------- --- ---- --- ---
em_dev_update_perf_domain 20 211.830 us 10.591 us 23.998 us
Function Hit Time Avg s^2
-------- --- ---- --- ---
Function Hit Time Avg s^2
-------- --- ---- --- ---
em_dev_update_perf_domain 15 78.085 us 5.205 us 7.444 us
Some test results.
The EM can be updated to fit better the workload type. In the case below the EM
has been updated for the Jankbench test on Pixel6 (running v5.18 w/ mainline backports
for the scheduler bits). The Jankbench was run 10 times for those two configurations,
to get more reliable data.
1. Janky frames percentage
+--------+-----------------+---------------------+-------+-----------+
| metric | variable | kernel | value | perc_diff |
+--------+-----------------+---------------------+-------+-----------+
| gmean | jank_percentage | EM_default | 2.0 | 0.0% |
| gmean | jank_percentage | EM_modified_runtime | 1.3 | -35.33% |
+--------+-----------------+---------------------+-------+-----------+
2. Avg frame render time duration
+--------+---------------------+---------------------+-------+-----------+
| metric | variable | kernel | value | perc_diff |
+--------+---------------------+---------------------+-------+-----------+
| gmean | mean_frame_duration | EM_default | 10.5 | 0.0% |
| gmean | mean_frame_duration | EM_modified_runtime | 9.6 | -8.52% |
+--------+---------------------+---------------------+-------+-----------+
3. Max frame render time duration
+--------+--------------------+---------------------+-------+-----------+
| metric | variable | kernel | value | perc_diff |
+--------+--------------------+---------------------+-------+-----------+
| gmean | max_frame_duration | EM_default | 251.6 | 0.0% |
| gmean | max_frame_duration | EM_modified_runtime | 115.5 | -54.09% |
+--------+--------------------+---------------------+-------+-----------+
4. OS overutilized state percentage (when EAS is not working)
+--------------+---------------------+------+------------+------------+
| metric | wa_path | time | total_time | percentage |
+--------------+---------------------+------+------------+------------+
| overutilized | EM_default | 1.65 | 253.38 | 0.65 |
| overutilized | EM_modified_runtime | 1.4 | 277.5 | 0.51 |
+--------------+---------------------+------+------------+------------+
5. All CPUs (Little+Mid+Big) power values in mW
+------------+--------+---------------------+-------+-----------+
| channel | metric | kernel | value | perc_diff |
+------------+--------+---------------------+-------+-----------+
| CPU | gmean | EM_default | 142.1 | 0.0% |
| CPU | gmean | EM_modified_runtime | 131.8 | -7.27% |
+------------+--------+---------------------+-------+-----------+
Changelog:
v4:
- refactored 2 rcu callbacks into 1 (Dietmar)
- fixed documentation (Dietmar)
- fixed 'run-time' into 'runtime' in all comments (Dietmar)
- fixed comments in patch headers (Diermar)
- fixed build issue in one patch (Dietmar)
- added cost of updating EM in the cover letter (Dietmar)
- added 'performance' field, which alled to optilize further
the pre-calculated 'cost' field and remove division operation
from runtime
- added update function during the boot after new EM is registered
to modify older EMs if needed due to potential CPU capacity change
v3 [6]:
- adjusted inline comments for structure doc (Dietmar)
- extended patch header with infromation that only EAS will use the feature
and it was driving the design (Dietmar)
- changed patch header and put shorter comment (Dietmar)
- moved the 'refactoring' patch that adds 'default_table' before the
introduction of runtime modifiable feature as sugested by Dietmar in
numerous patches v2
- merged documentation patches into one
- added more explenation about the 2 tables design into the Doc (Dietmar)
- removed the CPPC+EM patch for runtime modification
- removed the trace patch, since the trace point would be added after a while
- renamed 'tmp' to 'runtime_table' variable in the unregister function,
to better highlight the memory pointer checks (it is possible after
moving the 'default_table' earlier)
- and added '__rcu' in the unregister function which should calm down
the test bot warning
- renamed 'create' to 'refactor' in the patch header (Dietmar)
v2 [5]:
- solved build warning of unused variable in patch 13/17 when EM is
not compiled in, e.g. on Intel platform for this cpufreq_cooling
- re-based on top of v6.4-rc1
v1:
- implementation can be found here [4]
Regards,
Lukasz Luba
[1] https://lpc.events/event/16/contributions/1341/attachments/955/1873/Dynamic_Energy_Model_to_handle_leakage_power.pdf
[2] https://lpc.events/event/16/contributions/1194/attachments/1114/2139/LPC2022_Energy_model_accuracy.pdf
[3] https://www.youtube.com/watch?v=2C-5uikSbtM&list=PL0fKordpLTjKsBOUcZqnzlHShri4YBL1H
[4] https://lore.kernel.org/lkml/20230314103357.26010-1-lukasz.luba@xxxxxxx/
[5] https://lore.kernel.org/lkml/20230512095743.3393563-1-lukasz.luba@xxxxxxx/
[6] https://lore.kernel.org/lkml/20230721155022.2339982-1-lukasz.luba@xxxxxxx/
Lukasz Luba (18):
PM: EM: Add missing newline for the message log
PM: EM: Refactor em_cpufreq_update_efficiencies() arguments
PM: EM: Find first CPU online while updating OPP efficiency
PM: EM: Refactor em_pd_get_efficient_state() to be more flexible
PM: EM: Refactor a new function em_compute_costs()
PM: EM: Check if the get_cost() callback is present in
em_compute_costs()
PM: EM: Refactor struct em_perf_domain and add default_table
PM: EM: Add update_power() callback for runtime modifications
PM: EM: Introduce runtime modifiable table
PM: EM: Add RCU mechanism which safely cleans the old data
PM: EM: Add runtime update interface to modify EM power
PM: EM: Use runtime modified EM for CPUs energy estimation in EAS
Documentation: EM: Update with runtime modification design
PM: EM: Add performance field to struct em_perf_state
PM: EM: Adjust performance with runtime modification callback
PM: EM: Support late CPUs booting and capacity adjustment
PM: EM: Optimize em_cpu_energy() and remove division
Documentation: EM: Update information about performance field
Documentation/power/energy-model.rst | 149 +++++++++-
drivers/powercap/dtpm_cpu.c | 27 +-
drivers/powercap/dtpm_devfreq.c | 23 +-
drivers/thermal/cpufreq_cooling.c | 24 +-
drivers/thermal/devfreq_cooling.c | 23 +-
include/linux/energy_model.h | 155 ++++++-----
kernel/power/energy_model.c | 400 ++++++++++++++++++++++++---
7 files changed, 651 insertions(+), 150 deletions(-)
--
2.25.1