1UCLAMPSET(1) User Commands UCLAMPSET(1)
2
6 uclampset - manipulate the utilization clamping attributes of the
7 system or a process
8
10 uclampset [options] [-m uclamp_min] [-M uclamp_max] _command argument
11 uclampset [options] [-m uclamp_min] [-M uclamp_max] -p PID
13
15 uclampset sets or retrieves the utilization clamping attributes of an
16 existing PID, or runs command with the given attributes.
17 Utilization clamping is a new feature added in v5.3. It gives a hint to
19 the scheduler about the allowed range of utilization the task should be
20 operating at.
21 The utilization of the task affects frequency selection and task
23 placement. Only schedutil cpufreq governor understands handling util
24 clamp hints at the time of writing. Consult your kernel docs for
25 further info about other cpufreq governors support.
26 If you’re running on asymmetric heterogeneous system like Arm’s
28 big.LITTLE. Utilization clamping can help bias task placement. If the
29 task is boosted such that util_min value is higher than the little
30 cores' capacity, then the scheduler will do its best to place it on a
31 big core.
32 Similarly, if util_max is smaller than or equal the capacity of the
34 little cores, then the scheduler can still choose to place it there
35 even if the actual utilization of the task is at max.
36 Setting a task’s uclamp_min to a none zero value will effectively boost
38 the task as when it runs it’ll always start from this utilization
39 value.
40 By setting a task’s uclamp_max below 1024, this will effectively cap
42 the task as when it runs it’ll never be able to go above this
43 utilization value.
44 The full utilization range is: [0:1024]. The special value -1 is used
46 to reset to system’s default.
47
49 -m
50 Set util_min value.
51 -M
53 Set util_max value.
54 -a, --all-tasks
56 Set or retrieve the utilization clamping attributes of all the
57 tasks (threads) for a given PID.
58 -p, --pid
60 Operate on an existing PID and do not launch a new task.
61 -s, --system
63 Set or retrieve the system-wide utilization clamping attributes.
64 -R, --reset-on-fork
66 Set SCHED_FLAG_RESET_ON_FORK flag.
67 -v, --verbose
69 Show status information.
70 -V, --version
72 Display version information and exit.
73 -h, --help
75 Display help text and exit.
76
78 The default behavior is to run a new command:
79 uclampset [-m uclamp_min] [-M uclamp_max] command [arguments]
80 You can also retrieve the utilization clamping attributes of an
82 existing task:
83 uclampset -p PID
84 Or set them:
86 uclampset -p PID [-m uclamp_min] [-M uclamp_max]
87 Or control the system-wide attributes:
89 uclampset -s [-m uclamp_min] [-M uclamp_max]
90
92 A user must possess CAP_SYS_NICE to change the scheduling attributes of
93 a process. Any user can retrieve the scheduling information.
94
96 The system wide utilization clamp attributes are there to control the
97 allowed range the tasks can use. By default both uclamp_min and
98 uclamp_max are set to 1024. This means users can set the utilization
99 clamp values for their task across the full range [0:1024].
100 For example:
102 uclampset -s -m 512 -M 700
103 will prevent any task from being boosted higher than 512. And all tasks
105 in the systems are capped to a utilization of 700. Effectively
106 rendering the maximum performance of the system to 700.
107 Consult your kernel docs for the exact expected behavior on that
109 kernel.
110
112 Qais Yousef <qais.yousef@arm.com>
113
115 nice(1), renice(1), taskset(1), sched(7)
116 See sched_setscheduler(2) and sched_setattr(2) for a description of the
118 Linux scheduling scheme.
119
121 For bug reports, use the issue tracker at
122 https://github.com/karelzak/util-linux/issues.
123
125 The uclampset command is part of the util-linux package which can be
126 downloaded from Linux Kernel Archive
127 <https://www.kernel.org/pub/linux/utils/util-linux/>.
128util-linux 2.37.2 2021-06-02 UCLAMPSET(1)