aboutsummaryrefslogtreecommitdiffstats
path: root/Documentation/cpu-freq
diff options
context:
space:
mode:
Diffstat (limited to 'Documentation/cpu-freq')
-rw-r--r--Documentation/cpu-freq/governors.txt26
-rw-r--r--Documentation/cpu-freq/user-guide.txt12
2 files changed, 34 insertions, 4 deletions
diff --git a/Documentation/cpu-freq/governors.txt b/Documentation/cpu-freq/governors.txt
index 5b0cfa67aff9..ce73f3eb5ddb 100644
--- a/Documentation/cpu-freq/governors.txt
+++ b/Documentation/cpu-freq/governors.txt
@@ -117,10 +117,28 @@ accessible parameters:
117sampling_rate: measured in uS (10^-6 seconds), this is how often you 117sampling_rate: measured in uS (10^-6 seconds), this is how often you
118want the kernel to look at the CPU usage and to make decisions on 118want the kernel to look at the CPU usage and to make decisions on
119what to do about the frequency. Typically this is set to values of 119what to do about the frequency. Typically this is set to values of
120around '10000' or more. 120around '10000' or more. It's default value is (cmp. with users-guide.txt):
121 121transition_latency * 1000
122show_sampling_rate_(min|max): the minimum and maximum sampling rates 122The lowest value you can set is:
123available that you may set 'sampling_rate' to. 123transition_latency * 100 or it may get restricted to a value where it
124makes not sense for the kernel anymore to poll that often which depends
125on your HZ config variable (HZ=1000: max=20000us, HZ=250: max=5000).
126Be aware that transition latency is in ns and sampling_rate is in us, so you
127get the same sysfs value by default.
128Sampling rate should always get adjusted considering the transition latency
129To set the sampling rate 750 times as high as the transition latency
130in the bash (as said, 1000 is default), do:
131echo `$(($(cat cpuinfo_transition_latency) * 750 / 1000)) \
132 >ondemand/sampling_rate
133
134show_sampling_rate_(min|max): THIS INTERFACE IS DEPRECATED, DON'T USE IT.
135You can use wider ranges now and the general
136cpuinfo_transition_latency variable (cmp. with user-guide.txt) can be
137used to obtain exactly the same info:
138show_sampling_rate_min = transtition_latency * 500 / 1000
139show_sampling_rate_max = transtition_latency * 500000 / 1000
140(divided by 1000 is to illustrate that sampling rate is in us and
141transition latency is exported ns).
124 142
125up_threshold: defines what the average CPU usage between the samplings 143up_threshold: defines what the average CPU usage between the samplings
126of 'sampling_rate' needs to be for the kernel to make a decision on 144of 'sampling_rate' needs to be for the kernel to make a decision on
diff --git a/Documentation/cpu-freq/user-guide.txt b/Documentation/cpu-freq/user-guide.txt
index 917918f84fc7..75f41193f3e1 100644
--- a/Documentation/cpu-freq/user-guide.txt
+++ b/Documentation/cpu-freq/user-guide.txt
@@ -152,6 +152,18 @@ cpuinfo_min_freq : this file shows the minimum operating
152 frequency the processor can run at(in kHz) 152 frequency the processor can run at(in kHz)
153cpuinfo_max_freq : this file shows the maximum operating 153cpuinfo_max_freq : this file shows the maximum operating
154 frequency the processor can run at(in kHz) 154 frequency the processor can run at(in kHz)
155cpuinfo_transition_latency The time it takes on this CPU to
156 switch between two frequencies in nano
157 seconds. If unknown or known to be
158 that high that the driver does not
159 work with the ondemand governor, -1
160 (CPUFREQ_ETERNAL) will be returned.
161 Using this information can be useful
162 to choose an appropriate polling
163 frequency for a kernel governor or
164 userspace daemon. Make sure to not
165 switch the frequency too often
166 resulting in performance loss.
155scaling_driver : this file shows what cpufreq driver is 167scaling_driver : this file shows what cpufreq driver is
156 used to set the frequency on this CPU 168 used to set the frequency on this CPU
157 169