schedutils/uclampset.1.adoc

   1 //po4a: entry man manual
   2 ////
   3 uclampset(1) manpage
   4
   5 Copyright (C) 2020-2021 Qais Yousef <qyousef@layalina.io>
   6 Copyright (C) 2020-2021 Arm Ltd
   7
   8 This is free documentation; you can redistribute it and/or
   9 modify it under the terms of the GNU General Public License,
  10 version 2, as published by the Free Software Foundation.
  11
  12 The GNU General Public License's references to "object code"
  13 and "executables" are to be interpreted as the output of any
  14 document formatting or typesetting system, including
  15 intermediate and printed output.
  16
  17 This manual is distributed in the hope that it will be useful,
  18 but WITHOUT ANY WARRANTY; without even the implied warranty of
  19 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  20 GNU General Public License for more details.
  21
  22 You should have received a copy of the GNU General Public License along
  23 with this program; if not, write to the Free Software Foundation, Inc.,
  24 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
  25 ////
  26 = uclampset(1)
  27 :doctype: manpage
  28 :man manual: User Commands
  29 :man source: util-linux {release-version}
  30 :page-layout: base
  31 :command: uclampset
  32 :colon: :
  33
  34 == NAME
  35
  36 uclampset -
  37 manipulate the utilization clamping attributes of the system or a process
  38
  39 == SYNOPSIS
  40
  41 *uclampset* [options] [*-m* _uclamp_min_] [*-M* _uclamp_max] _command argument_
  42
  43 *uclampset* [options] [*-m* _uclamp_min_] [*-M* _uclamp_max_] *-p* _PID_
  44
  45 == DESCRIPTION
  46
  47 *uclampset* sets or retrieves the utilization clamping attributes of an existing _PID_, or runs _command_ with the given attributes.
  48
  49 Utilization clamping is a new feature added in v5.3. It gives a hint to the scheduler about the allowed range of utilization the task should be operating at.
  50
  51 The utilization of the task affects frequency selection and task placement. Only schedutil cpufreq governor understands handling util clamp hints at the time of writing. Consult your kernel docs for further info about other cpufreq governors support.
  52
  53 If you're running on asymmetric heterogeneous system like Arm's big.LITTLE. Utilization clamping can help bias task placement. If the task is boosted such that _util_min_ value is higher than the little cores' capacity, then the scheduler will do its best to place it on a big core.
  54
  55 Similarly, if _util_max_ is smaller than or equal the capacity of the little cores, then the scheduler can still choose to place it there even if the actual utilization of the task is at max.
  56
  57 Setting a task's _uclamp_min_ to a none zero value will effectively boost the task as when it runs it'll always start from this utilization value.
  58
  59 By setting a task's _uclamp_max_ below 1024, this will effectively cap the task as when it runs it'll never be able to go above this utilization value.
  60
  61 The full utilization range is: [0:1024]. The special value -1 is used to reset to system's default.
  62
  63 Consult latest kernel documentation for more details:
  64
  65         https://kernel.org/doc/html/latest/scheduler/sched-util-clamp.html
  66
  67 == OPTIONS
  68
  69 *-m*::
  70 Set _util_min_ value.
  71
  72 *-M*::
  73 Set _util_max_ value.
  74
  75 *-a*, *--all-tasks*::
  76 Set or retrieve the utilization clamping attributes of all the tasks (threads) for a given PID.
  77
  78 *-p*, *--pid*::
  79 Operate on an existing PID and do not launch a new task.
  80
  81 *-s*, *--system*::
  82 Set or retrieve the system-wide utilization clamping attributes.
  83
  84 *-R*, *--reset-on-fork*::
  85 Set *SCHED_FLAG_RESET_ON_FORK* flag.
  86
  87 *-v*, *--verbose*::
  88 Show status information.
  89
  90 include::man-common/help-version.adoc[]
  91
  92 == USAGE
  93
  94 //TRANSLATORS: Keep {colon} untranslated.
  95 The default behavior is to run a new command{colon}::
  96 *uclampset* _[-m uclamp_min]_ _[-M uclamp_max]_ _command_ [_arguments_]
  97
  98 //TRANSLATORS: Keep {colon} untranslated.
  99 You can also retrieve the utilization clamping attributes of an existing task{colon}::
 100 *uclampset -p* _PID_
 101
 102 //TRANSLATORS: Keep {colon} untranslated.
 103 Or set them{colon}::
 104 *uclampset -p* _PID_ _[-m uclamp_min]_ _[-M uclamp_max]_
 105
 106 //TRANSLATORS: Keep {colon} untranslated.
 107 Or control the system-wide attributes{colon}::
 108 *uclampset -s* _[-m uclamp_min]_ _[-M uclamp_max]_
 109
 110 == PERMISSIONS
 111
 112 A user must possess *CAP_SYS_NICE* to change the scheduling attributes of a process. Any user can retrieve the scheduling information.
 113
 114 == NOTES
 115
 116 The system wide utilization clamp attributes are there to control the _allowed_ range the tasks can use. By default both _uclamp_min_ and _uclamp_max_ are set to 1024. This means users can set the utilization clamp values for their task across the full range [0:1024].
 117
 118 //TRANSLATORS: Keep {colon} untranslated.
 119 For example{colon}::
 120 *uclampset -s* `-m 512` `-M 700`
 121
 122 will prevent any task from being boosted higher than 512. And all tasks in the systems are capped to a utilization of 700. Effectively rendering the maximum performance of the system to 700.
 123
 124 Consult your kernel docs for the exact expected behavior on that kernel.
 125
 126 == AUTHORS
 127
 128 mailto:qyousef@layalina.io[Qais Yousef]
 129
 130 == SEE ALSO
 131
 132 *nice*(1),
 133 *renice*(1),
 134 *taskset*(1),
 135 *sched*(7)
 136
 137 See *sched_setscheduler*(2) and *sched_setattr*(2) for a description of the Linux scheduling scheme.
 138
 139 include::man-common/bugreports.adoc[]
 140
 141 include::man-common/footer.adoc[]
 142
 143 ifdef::translation[]
 144 include::man-common/translation.adoc[]
 145 endif::[]