Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
From:   Parth Shah <parth@linux.ibm.com>
To:     peterz@infradead.org, mingo@redhat.com
Cc:     linux-kernel@vger.kernel.org, linux-pm@vger.kernel.org,
        patrick.bellasi@arm.com, dietmar.eggemann@arm.com,
        daniel.lezcano@linaro.org, subhra.mazumdar@oracle.com
Subject: [RFC v4 0/8] TurboSched: A scheduler for sustaining Turbo Frequencies for longer durations
Date:   Thu, 25 Jul 2019 12:38:49 +0530
Message-Id: <20190725070857.6639-1-parth@linux.ibm.com>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

This is the 4th version of the patchset to sustain Turbo frequencies for
longer durations.

The previous versions can be found here:
v3: https://lkml.org/lkml/2019/6/25/25
v2: https://lkml.org/lkml/2019/5/15/1258
v1: https://lwn.net/Articles/783959/

The changes in this versions are:
v[3] -> v[4]:
- Based on Patrick Bellasi's comments, removed the use of UCLAMP based
  mechanism to classify tasks as jitter
- Added support to sched_setattr to mark the task as jitter by adding a new
  flag to the existing task_struct->flags attribute. This is decided to not
  have any new variable inside task_struct and thus get rid of size
  bloating.
- No functional changes

v[2] -> v[3]:
- Added a new attribute in task_struct to allow per task jitter
  classification so that scheduler can use this as request to change wakeup
  path for task packing
- Use syscall for jitter classification, removed cgroup based task
  classification
- Use mutex over spinlock to get rid of task sleeping problem
- Changed _Bool->int everywhere
- Split few patches to have arch specific code separate from core scheduler
  code
ToDo:
- Recompute core capacity only during CPU-Hotplug operation
- Remove smt capacity 

v[1] -> v[2]:
- No CPU bound tasks' classification, only jitter tasks are classified from
  the cpu cgroup controller
- Use of Spinlock rather than mutex to count number of jitters in the
  system classified from cgroup
- Architecture specific implementation of Core capacity multiplication
  factor changes dynamically based on the number of active threads in the
  core
- Selection of non idle core in the system is bounded by DIE domain
- Use of UCLAMP mechanism to classify jitter tasks
- Removed "highutil_cpu_mask", and rather uses sd for DIE domain to find
  better fit


Abstract
========

The modern servers allows multiple cores to run at range of frequencies
higher than rated range of frequencies. But the power budget of the system
inhibits sustaining these higher frequencies for longer durations.

However when certain cores are put to idle states, the power can be
effectively channelled to other busy cores, allowing them to sustain the
higher frequency.

One way to achieve this is to pack tasks onto fewer cores keeping others
idle, but it may lead to performance penalty for such tasks and sustaining
higher frequencies proves to be of no benefit. But if one can identify
unimportant low utilization tasks which can be packed on the already active
cores then waking up of new cores can be avoided. Such tasks are short
and/or bursty "jitter tasks" and waking up new core is expensive for such
case.

Current CFS algorithm in kernel scheduler is performance oriented and hence
tries to assign any idle CPU first for the waking up of new tasks. This
policy is perfect for major categories of the workload, but for jitter
tasks, one can save energy by packing them onto the active cores and allow
those cores to run at higher frequencies.

These patch-set tunes the task wake up logic in scheduler to pack
exclusively classified jitter tasks onto busy cores. The work involves the
jitter tasks classifications by using syscall based mechanisms.

In brief, if we can pack jitter tasks on busy cores then we can save power
by keeping other cores idle and allow busier cores to run at turbo
frequencies, patch-set tries to meet this solution in simplest manner.
Though, there are some challenges in implementing it(like smt_capacity,
un-needed arch hooks, etc) and I'm trying to work around that, it would be
great to have a discussion around this patches.


Implementation
==============

These patches uses syscall based mechanism to classify the tasks as jitter.
The task wakeup logic uses this information to pack such tasks onto cores
which are already running busy with CPU intensive tasks. The task packing
is done at `select_task_rq_fair` only so that in case of wrong decision
load balancer may pull the classified jitter tasks for maximizing
performance.

We define a core to be non-idle if it is over 12.5% utilized of its
capacity; the jitters are packed over these cores using First-fit approach.

To demonstrate/benchmark, one can use a synthetic workload generator
`turbo_bench.c`[1] available at
https://github.com/parthsl/tools/blob/master/benchmarks/turbo_bench.c

Following snippet demonstrates the use of TurboSched feature:
```
i=8; ./turbo_bench -t 30 -h $i -n $((i*2)) -j
```
This spawns 2*i total threads: of which i-CPU bound and i-jitter threads.

Current implementation uses only jitter classified tasks to be packed on
the first busy cores, but can be further optimized by getting userspace
input of important tasks and keeping track of such tasks. This leads to
optimized searching of non idle cores and also more accurate as userspace
hints are safer than auto classified busy cores/tasks.


Result
======

The patch-set proves to be useful for the system and the workload where
frequency boost is found to be useful than packing tasks into cores. IBM
POWER 9 system shows the benefit for a workload can be up to 13%.

                Performance benefit of TurboSched w.r.t. CFS 
   +--+--+--+--+--+--+-+--+--+--+--+--+--+--+--+--+--+--+-+--+--+--+--+--+
   |  +  +  +  +  +  + +  +  +  +  +  +  +  +  +  +  +  + +  +  +  +  +  |
15 +-+                                  Performance benefit in %       +-+
   |                         **                                          |
   |                         ** **                                       |
10 +-+                       ** ** **                                  +-+
   |                         ** ** **                                    |
   |                         ** ** **                                    |
 5 +-+                 ** ** ** ** **    **                            +-+
   |                   ** ** ** ** ** ** ** **                           |
   |                   ** ** ** ** ** ** ** ** ** **                     |
   |                 * ** ** ** ** ** ** ** ** ** ** ** *                |
 0 +-+** ** ** ** ** * ** ** ** ** ** ** ** ** ** ** ** * ** ** ** ** **-+
   |  ** ** ** **                                                        |
   |  **                                                                 |
-5 +-+                                                                 +-+
   |  +  +  +  +  +  + +  +  +  +  +  +  +  +  +  +  +  + +  +  +  +  +  |
   +--+--+--+--+--+--+-+--+--+--+--+--+--+--+--+--+--+--+-+--+--+--+--+--+
      2  3  4  5  6  7 8  9 10 11 12 13 14 15 16 17 18 1920 21 22 23 24   
                           No. of workload threads                        


                      Frequency benefit of TurboSched w.r.t. CFS
   +--+--+--+--+--+--+-+--+--+--+--+--+--+--+--+--+--+--+-+--+--+--+--+--+
   |  +  +  +  +  +  + +  +  +  +  +  +  +  +  +  +  +  + +  +  +  +  +  |
15 +-+                                    Frequency benefit in %       +-+
   |                         **                                          |
   |                         **                                          |
10 +-+            **         **                                        +-+
   |              **         ** **                                       |
   |        **    ** * **    ** **                                       |
 5 +-+      ** ** ** * ** ** ** **                                     +-+
   |     ** ** ** ** * ** ** ** **    **                                 |
   |  ** ** ** ** ** * ** ** ** ** ** **                                 |
   |  ** ** ** ** ** * ** ** ** ** ** ** ** ** **                        |
 0 +-+** ** ** ** ** * ** ** ** ** ** ** ** ** ** ** ** * ** ** ** ** **-+
   |                                                                     |
   |                                                                     |
-5 +-+                                                                 +-+
   |  +  +  +  +  +  + +  +  +  +  +  +  +  +  +  +  +  + +  +  +  +  +  |
   +--+--+--+--+--+--+-+--+--+--+--+--+--+--+--+--+--+--+-+--+--+--+--+--+
      2  3  4  5  6  7 8  9 10 11 12 13 14 15 16 17 18 1920 21 22 23 24   
                             No. of workload threads                      

 
These numbers are w.r.t. `turbo_bench.c` multi-threaded test benchmark
which can create two kinds of tasks: CPU bound (High Utilization) and
Jitters (Low Utilization). N in X-axis represents N-CPU bound and N-Jitter
tasks spawned.


Series organization
==============
- Patches [01-03]: Jitter tasks classification using syscall
- Patches [04-05]: Defines Core Capacity to limit task packing
- Patches [06-08]: Tune CFS task wakeup logic to pack tasks onto busy
  cores

Series can be applied on the top of tip/sched/core at
commit af24bde8df20 ("sched/uclamp: Add uclamp support to energy_compute()")


Parth Shah (8):
  sched/core: Add manual jitter classification using sched_setattr
    syscall
  sched: Introduce switch to enable TurboSched mode
  sched/core: Update turbo_sched count only when required
  sched/fair: Define core capacity to limit task packing
  powerpc: Define Core Capacity for POWER systems
  sched/fair: Tune task wake-up logic to pack jitter tasks
  sched/fair: Bound non idle core search within LLC domain
  powerpc: Set turbo domain to NUMA node for task packing

 arch/powerpc/include/asm/topology.h |   7 ++
 arch/powerpc/kernel/smp.c           |  38 ++++++++
 include/linux/sched.h               |   1 +
 include/uapi/linux/sched.h          |   4 +-
 kernel/sched/core.c                 |  39 ++++++++
 kernel/sched/fair.c                 | 135 +++++++++++++++++++++++++++-
 kernel/sched/sched.h                |   9 ++
 7 files changed, 231 insertions(+), 2 deletions(-)

-- 
2.17.1