1 CPU cooling APIs How To
2 ===================================
4 Written by Amit Daniel Kachhap <amit.kachhap@linaro.org>
8 Copyright (c) 2012 Samsung Electronics Co., Ltd(http://www.samsung.com)
12 The generic cpu cooling(freq clipping) provides registration/unregistration APIs
13 to the caller. The binding of the cooling devices to the trip point is left for
14 the user. The registration APIs returns the cooling device pointer.
18 1.1 cpufreq registration/unregistration APIs
19 1.1.1 struct thermal_cooling_device *cpufreq_cooling_register(
20 struct cpumask *clip_cpus)
22 This interface function registers the cpufreq cooling device with the name
23 "thermal-cpufreq-%x". This api can support multiple instances of cpufreq
26 clip_cpus: cpumask of cpus where the frequency constraints will happen.
28 1.1.2 struct thermal_cooling_device *of_cpufreq_cooling_register(
29 struct cpufreq_policy *policy)
31 This interface function registers the cpufreq cooling device with
32 the name "thermal-cpufreq-%x" linking it with a device tree node, in
33 order to bind it via the thermal DT code. This api can support multiple
34 instances of cpufreq cooling devices.
36 policy: CPUFreq policy.
38 1.1.3 void cpufreq_cooling_unregister(struct thermal_cooling_device *cdev)
40 This interface function unregisters the "thermal-cpufreq-%x" cooling device.
42 cdev: Cooling device pointer which has to be unregistered.
46 The power API registration functions provide a simple power model for
47 CPUs. The current power is calculated as dynamic power (static power isn't
48 supported currently). This power model requires that the operating-points of
49 the CPUs are registered using the kernel's opp library and the
50 `cpufreq_frequency_table` is assigned to the `struct device` of the
51 cpu. If you are using CONFIG_CPUFREQ_DT then the
52 `cpufreq_frequency_table` should already be assigned to the cpu
55 The dynamic power consumption of a processor depends on many factors.
56 For a given processor implementation the primary factors are:
58 - The time the processor spends running, consuming dynamic power, as
59 compared to the time in idle states where dynamic consumption is
60 negligible. Herein we refer to this as 'utilisation'.
61 - The voltage and frequency levels as a result of DVFS. The DVFS
62 level is a dominant factor governing power consumption.
63 - In running time the 'execution' behaviour (instruction types, memory
64 access patterns and so forth) causes, in most cases, a second order
65 variation. In pathological cases this variation can be significant,
66 but typically it is of a much lesser impact than the factors above.
68 A high level dynamic power consumption model may then be represented as:
70 Pdyn = f(run) * Voltage^2 * Frequency * Utilisation
72 f(run) here represents the described execution behaviour and its
73 result has a units of Watts/Hz/Volt^2 (this often expressed in
76 The detailed behaviour for f(run) could be modelled on-line. However,
77 in practice, such an on-line model has dependencies on a number of
78 implementation specific processor support and characterisation
79 factors. Therefore, in initial implementation that contribution is
80 represented as a constant coefficient. This is a simplification
81 consistent with the relative contribution to overall power variation.
83 In this simplified representation our model becomes:
85 Pdyn = Capacitance * Voltage^2 * Frequency * Utilisation
87 Where `capacitance` is a constant that represents an indicative
88 running time dynamic power coefficient in fundamental units of
89 mW/MHz/uVolt^2. Typical values for mobile CPUs might lie in range
90 from 100 to 500. For reference, the approximate values for the SoC in
91 ARM's Juno Development Platform are 530 for the Cortex-A57 cluster and
92 140 for the Cortex-A53 cluster.