1 CPU frequency and voltage scaling code in the Linux(TM) kernel
4 L i n u x C P U F r e q
6 C P U F r e q G o v e r n o r s
8 - information for users and developers -
11 Dominik Brodowski <linux@brodo.de>
12 some additions and corrections by Nico Golde <nico@ngolde.de>
16 Clock scaling allows you to change the clock speed of the CPUs on the
17 fly. This is a nice method to save battery power, because the lower
18 the clock speed, the less power the CPU consumes.
23 1. What is a CPUFreq Governor?
25 2. Governors In the Linux Kernel
32 3. The Governor Interface in the CPUfreq Core
36 1. What Is A CPUFreq Governor?
37 ==============================
39 Most cpufreq drivers (in fact, all except one, longrun) or even most
40 cpu frequency scaling algorithms only offer the CPU to be set to one
41 frequency. In order to offer dynamic frequency scaling, the cpufreq
42 core must be able to tell these drivers of a "target frequency". So
43 these specific drivers will be transformed to offer a "->target"
44 call instead of the existing "->setpolicy" call. For "longrun", all
45 stays the same, though.
47 How to decide what frequency within the CPUfreq policy should be used?
48 That's done using "cpufreq governors". Two are already in this patch
49 -- they're the already existing "powersave" and "performance" which
50 set the frequency statically to the lowest or highest frequency,
51 respectively. At least two more such governors will be ready for
52 addition in the near future, but likely many more as there are various
53 different theories and models about dynamic frequency scaling
54 around. Using such a generic interface as cpufreq offers to scaling
55 governors, these can be tested extensively, and the best one can be
56 selected for each specific use.
58 Basically, it's the following flow graph:
60 CPU can be set to switch independently | CPU can only be set
61 within specific "limits" | to specific frequencies
64 consists of frequency limits (policy->{min,max})
65 and CPUfreq governor to be used
68 / the cpufreq governor decides
69 / (dynamically or statically)
70 / what target_freq to set within
71 / the limits of policy->{min,max}
74 Using the ->setpolicy call, Using the ->target call,
75 the limits and the the frequency closest
76 "policy" is set. to target_freq is set.
78 is within policy->{min,max}
81 2. Governors In the Linux Kernel
82 ================================
87 The CPUfreq governor "performance" sets the CPU statically to the
88 highest frequency within the borders of scaling_min_freq and
95 The CPUfreq governor "powersave" sets the CPU statically to the
96 lowest frequency within the borders of scaling_min_freq and
103 The CPUfreq governor "userspace" allows the user, or any userspace
104 program running with UID "root", to set the CPU to a specific frequency
105 by making a sysfs file "scaling_setspeed" available in the CPU-device
112 The CPUfreq governor "ondemand" sets the CPU depending on the
113 current usage. To do this the CPU must have the capability to
114 switch the frequency very quickly. There are a number of sysfs file
115 accessible parameters:
117 sampling_rate: measured in uS (10^-6 seconds), this is how often you
118 want the kernel to look at the CPU usage and to make decisions on
119 what to do about the frequency. Typically this is set to values of
120 around '10000' or more. It's default value is (cmp. with users-guide.txt):
121 transition_latency * 1000
122 The lowest value you can set is:
123 transition_latency * 100 or it may get restricted to a value where it
124 makes not sense for the kernel anymore to poll that often which depends
125 on your HZ config variable (HZ=1000: max=20000us, HZ=250: max=5000).
126 Be aware that transition latency is in ns and sampling_rate is in us, so you
127 get the same sysfs value by default.
128 Sampling rate should always get adjusted considering the transition latency
129 To set the sampling rate 750 times as high as the transition latency
130 in the bash (as said, 1000 is default), do:
131 echo `$(($(cat cpuinfo_transition_latency) * 750 / 1000)) \
132 >ondemand/sampling_rate
134 show_sampling_rate_(min|max): THIS INTERFACE IS DEPRECATED, DON'T USE IT.
135 You can use wider ranges now and the general
136 cpuinfo_transition_latency variable (cmp. with user-guide.txt) can be
137 used to obtain exactly the same info:
138 show_sampling_rate_min = transtition_latency * 500 / 1000
139 show_sampling_rate_max = transtition_latency * 500000 / 1000
140 (divided by 1000 is to illustrate that sampling rate is in us and
141 transition latency is exported ns).
143 up_threshold: defines what the average CPU usage between the samplings
144 of 'sampling_rate' needs to be for the kernel to make a decision on
145 whether it should increase the frequency. For example when it is set
146 to its default value of '80' it means that between the checking
147 intervals the CPU needs to be on average more than 80% in use to then
148 decide that the CPU frequency needs to be increased.
150 ignore_nice_load: this parameter takes a value of '0' or '1'. When
151 set to '0' (its default), all processes are counted towards the
152 'cpu utilisation' value. When set to '1', the processes that are
153 run with a 'nice' value will not count (and thus be ignored) in the
154 overall usage calculation. This is useful if you are running a CPU
155 intensive calculation on your laptop that you do not care how long it
156 takes to complete as you can 'nice' it and prevent it from taking part
157 in the deciding process of whether to increase your CPU frequency.
163 The CPUfreq governor "conservative", much like the "ondemand"
164 governor, sets the CPU depending on the current usage. It differs in
165 behaviour in that it gracefully increases and decreases the CPU speed
166 rather than jumping to max speed the moment there is any load on the
167 CPU. This behaviour more suitable in a battery powered environment.
168 The governor is tweaked in the same manner as the "ondemand" governor
169 through sysfs with the addition of:
171 freq_step: this describes what percentage steps the cpu freq should be
172 increased and decreased smoothly by. By default the cpu frequency will
173 increase in 5% chunks of your maximum cpu frequency. You can change this
174 value to anywhere between 0 and 100 where '0' will effectively lock your
175 CPU at a speed regardless of its load whilst '100' will, in theory, make
176 it behave identically to the "ondemand" governor.
178 down_threshold: same as the 'up_threshold' found for the "ondemand"
179 governor but for the opposite direction. For example when set to its
180 default value of '20' it means that if the CPU usage needs to be below
181 20% between samples to have the frequency decreased.
183 3. The Governor Interface in the CPUfreq Core
184 =============================================
186 A new governor must register itself with the CPUfreq core using
187 "cpufreq_register_governor". The struct cpufreq_governor, which has to
188 be passed to that function, must contain the following values:
190 governor->name - A unique name for this governor
191 governor->governor - The governor callback function
192 governor->owner - .THIS_MODULE for the governor module (if
195 The governor->governor callback is called with the current (or to-be-set)
196 cpufreq_policy struct for that CPU, and an unsigned int event. The
197 following events are currently defined:
199 CPUFREQ_GOV_START: This governor shall start its duty for the CPU
201 CPUFREQ_GOV_STOP: This governor shall end its duty for the CPU
203 CPUFREQ_GOV_LIMITS: The limits for CPU policy->cpu have changed to
204 policy->min and policy->max.
206 If you need other "events" externally of your driver, _only_ use the
207 cpufreq_governor_l(unsigned int cpu, unsigned int event) call to the
208 CPUfreq core to ensure proper locking.
211 The CPUfreq governor may call the CPU processor driver using one of
214 int cpufreq_driver_target(struct cpufreq_policy *policy,
215 unsigned int target_freq,
216 unsigned int relation);
218 int __cpufreq_driver_target(struct cpufreq_policy *policy,
219 unsigned int target_freq,
220 unsigned int relation);
222 target_freq must be within policy->min and policy->max, of course.
223 What's the difference between these two functions? When your governor
224 still is in a direct code path of a call to governor->governor, the
225 per-CPU cpufreq lock is still held in the cpufreq core, and there's
226 no need to lock it again (in fact, this would cause a deadlock). So
227 use __cpufreq_driver_target only in these cases. In all other cases
228 (for example, when there's a "daemonized" function that wakes up
229 every second), use cpufreq_driver_target to lock the cpufreq per-CPU
230 lock before the command is passed to the cpufreq processor driver.