Message-Id: <cover.1741091349.git.hongyan.xia2@arm.com>
Date: Tue,  4 Mar 2025 14:23:07 +0000
From: Hongyan Xia <hongyan.xia2@....com>
To: Ingo Molnar <mingo@...hat.com>,
	Peter Zijlstra <peterz@...radead.org>,
	Vincent Guittot <vincent.guittot@...aro.org>,
	Dietmar Eggemann <dietmar.eggemann@....com>
Cc: Morten Rasmussen <morten.rasmussen@....com>,
	Lukasz Luba <lukasz.luba@....com>,
	Christian Loehle <christian.loehle@....com>,
	Pierre Gondois <pierre.gondois@....com>,
	linux-kernel@...r.kernel.org
Subject: [PATCH v2 0/8] uclamp sum aggregation

This series gives an alternative implementation that addresses some of
the problems in uclamp max aggregation. Sum aggregation mostly gives:

1. Simplicity. Sum aggregation implements uclamp with less than half of
   code than max aggregation.
2. Effectiveness. Sum aggregation shows better uclamp effectiveness,
   either in benchmark scores or sometimes in energy efficiency.

The key idea of sum aggregation is fairly simple. Each task has a
util_avg_bias, which is obtained by:

    util_avg_bias = clamp(util_avg, uclamp_min, uclamp_max) - util_avg;

If a CPU has N tasks, p1, p2, p3... pN, then we sum the biases up and
obtain a rq total bias:

    rq_bias = util_avg_bias1 + util_avg_bias2... + util_avg_biasN;

Then we use the biased rq utilization rq_util + rq_bias to select OPP
and to schedule tasks.

PATCH BREAKDOWN:

Patch 1/6 reverts a patch that accommodate uclamp_max tasks under max
aggregation. This patch is not needed and creates other problems for sum
aggregation. It is discussed elsewhere that this patch will be improved
and there may not be the need to revert it in the future.

Patch 2, 3 and 4 implement sum aggregation.

Patch 5 and 6 remove max aggregation.

Patch 7 applies PELT decay on negative util_avg_bias. This improves
energy efficiency and task placement, but is not strictly necessary.

Patch 8 addresses sum aggregation under-utilization problem.

TESTING:

Two notebooks are shared at

https://nbviewer.org/github/honxia02/notebooks/blob/aac12d9becae2b2fe4690cbb672439fd884ede30/whitebox/max.ipynb
https://nbviewer.org/github/honxia02/notebooks/blob/aac12d9becae2b2fe4690cbb672439fd884ede30/whitebox/sum-offset.ipynb

The experiments done in notebooks are on Arm Juno r2 board. CPU0-3 are
little cores with capacity of 383. CPU4-5 are big cores. The rt-app
profiles used for these experiments are included in the notebooks.

Scenario 1: Scheduling 4 tasks with UCLAMP_MAX at 110.

The scheduling decisions are plotted in Out[11]. Both max and sum
aggregation understand the UCLAMP_MAX hint and schedule all 4 tasks on
the little cluster. Max aggregation sometimes schedule 2 tasks on 1 CPU,
and this is the reason why sum aggregation reverts the 1st commit.

Scenario 2: Scheduling 4 tasks with UCLAMP_MIN and UCLAMP_MAX at a value
slightly above the capacity of the little CPU.

Results are in Out[17]. The purpose is to use UCLAMP_MIN to place tasks
on the big core. Both max and sum aggregation handle this correctly.

Scenario 3: Task A is a task with a small utilization pinned to CPU4.
Task B is an always-running task pinned to CPU5, but UCLAMP_MAX capped
at 300. After a while, task A is then pinned to CPU5, joining B.

Results are in Out[23]. Max aggregation sees a frequency spike at
873.64s. When zoomed in, one can see square-wave-like utilization values
because of A periodically going to sleep. When A wakes up, its default
UCLAMP_MAX of 1024 will uncap B and reach the highest CPU frequency.
When A sleeps, B's UCLAMP_MAX will be in effect and will reduce rq
utilization. This happens repeatedly, hence the square wave. In
contrast, sum aggregation sees a normal increase in utilization when A
joins B, without any square-wave behavior.

Scenario 4: 4 always-running tasks with UCLAMP_MAX of 110 pinned to the
little PD (CPU0-3). 4 same tasks pinned to the big PD (CPU4-5).
After a while, remove the CPU pinning of the 4 tasks on the big PD.

Results are in Out[29]. After unpinning, max aggregation moves all 8
tasks to the little cluster, but schedules 5 tasks on CPU0 and 1 each on
CPU1-3. In contrast, sum aggregation schedules 2 on each little CPU
after unpinning, which is the desired balanced task placement.

EVALUATION:

We backport patches to GKI kernel v6.1 on Pixel 9 and run Android
benchmarks.

Speedometer:

We run Speedometer 2.1 on Chrome v131 to test ADPF/uclamp effectiveness.
Because sum aggregation does not circumvent the 25% OPP margin, we
reduce uclamp values to 80% to be fair.

|   score   | score |   %    | CPU power % |
|    max    | 192.4 |        |             |
|  sum_0.8  | 230.8 | +19.96 |   +31.54    |
| sum_tuned | 201.8 |  +4.89 |    -0.41    |

We see a consistant higher score and higher average power consumption.
Note that a higher score also means a reduction in run-time, total
energy increase for sum_0.8 is only 9.65%.

We then reduce uclamp values so that power consumption is roughly
the same. If we do so, then sum aggregation achieves slightly better
scores, shown in the sum_tuned row.

UIBench:

|   score   | jank percentage |   %    | CPU power (mW) |   %   |
|    max    |     0.115%      |        |     158.1      |       |
|  sum_0.8  |     0.129%      | +11.96 |     154.9      | -4.19 |

UIBench on Pixel 9 by default already has a low enough jank percentage.
Moving to sum aggregation gives slightly higher jank percentage and
lower power consumption.

--- 
Changed in v2:
- Completely separate uclamp component from PELT and util_est.
- Separate util_est_uclamp into an individual patch.
- Address the under-utilization problem.
- Update Python notebooks to reflect the latest sched/tip.

Hongyan Xia (8):
  Revert "sched/uclamp: Set max_spare_cap_cpu even if max_spare_cap is
    0"
  sched/uclamp: Track a new util_avg_bias signal
  sched/uclamp: Add util_est_uclamp
  sched/fair: Use util biases for utilization and frequency
  sched/uclamp: Remove all uclamp bucket logic
  sched/uclamp: Simplify uclamp_eff_value()
  sched/uclamp: Propagate negative bias
  sched/uclamp: Solve under-utilization problem

 include/linux/sched.h            |   8 +-
 init/Kconfig                     |  32 ---
 kernel/sched/core.c              | 308 ++--------------------
 kernel/sched/cpufreq_schedutil.c |   6 +-
 kernel/sched/debug.c             |   2 +-
 kernel/sched/fair.c              | 430 ++++++++++++++++---------------
 kernel/sched/pelt.c              |  62 +++++
 kernel/sched/rt.c                |   4 -
 kernel/sched/sched.h             | 132 +++-------
 kernel/sched/syscalls.c          |   2 +
 10 files changed, 341 insertions(+), 645 deletions(-)

-- 
2.34.1

Powered by blists - more mailing lists