Difference between revisions of "How to configure cpufreqd"

From ThinkWiki
Jump to: navigation, search
(proposal of in-kernel alternative to cpufreqd)
(spelling corrections)
Line 106: Line 106:
 
  profile=hi_boost
 
  profile=hi_boost
  
Remark: in recent Linux kernels there is the ondemand scaling governor. Using that one, you'll probably not need any cpu speed utilities any more. Additionally, you may change between "ondemand" and "performance" or "poewrsave" scaling governors by using ACPI events via acpid. That way, your CPU time probably won't be wasted as cpufreqd and assorted utilities do.
+
Remark: in recent Linux kernels there is the ondemand scaling governor. Using that one, you'll probably not need any cpu speed utilities any more. Additionally, you may change between "ondemand" and "performance" or "powersave" scaling governors by using ACPI events via acpid. That way, your CPU time probably won't get wasted as cpufreqd and assorted utilities do.

Revision as of 14:44, 10 July 2005

In Debian, all you have to do is:

apt-get install cpufreqd

And in most cases it will work out of the box.

This is the default Debian config, but works well out of the box:

# this is a comment
#
# you need: 1 [General] section,
#           1 or more [Profile] sections
#                1 or more [Rule] sections
#
# a section ends at the first blank line
#
# [Rule] sample:
#           [Rule]
#           name=sample_rule
#           ac=on                    # (on/off)
#           battery_interval=0-10
#           cpu_interval=30-60
#           programs=xine,mplayer
#           profile=sample_profile
#
# [Profile] sample:
#           [Profile]
#           name=sample_profile
#           minfreq=10%
#           maxfreq=100%
#           policy=performance
#
# see CPUFREQD.CONF(5) manpage for a complete reference

[General]
pidfile=/var/run/cpufreqd.pid
poll_interval=2
pm_type=acpi #(acpi, apm or pmu)
# Uncomment the following line to enable ACPI workaround (see cpufreqd.conf(5)) # acpi_workaround=1
verbosity=4 #(if you want a minimal logging set to 5)

[Profile]
name=hi_boost
minfreq=66%
maxfreq=100%
policy=performance

[Profile]
name=medium_boost
minfreq=33%
maxfreq=66%
policy=performance

[Profile]
name=lo_boost
minfreq=0%
maxfreq=33%
policy=performance

[Profile]
name=lo_power
minfreq=0%
maxfreq=33%
policy=powersave

# conservative mode when not AC
[Rule]
name=conservative
ac=off                   # (on/off)
battery_interval=0-100
cpu_interval=0-40
cpu_nice_scale=1.5
delay_cycles=3
profile=lo_boost

# need some power
[Rule]
name=lo_cpu_boost
ac=off                   # (on/off)
battery_interval=0-100
cpu_interval=30-80
profile=medium_boost

# need big power (not if battery very low)
[Rule]
name=hi_cpu_boost
ac=off                   # (on/off)
battery_interval=50-100
cpu_interval=70-100
profile=medium_boost

# full power when AC
[Rule]
name=AC_on
ac=on                   # (on/off)
profile=hi_boost

# full power when watching DVDs and not AC:
# this is the last rule and takes less
# precedence with respect to the others
[Rule]
name=dvd_watching
ac=off                   # (on/off)
battery_interval=0-100
programs=xine,totem,vlc,avidemux
cpu_interval=0-100
profile=hi_boost

Remark: in recent Linux kernels there is the ondemand scaling governor. Using that one, you'll probably not need any cpu speed utilities any more. Additionally, you may change between "ondemand" and "performance" or "powersave" scaling governors by using ACPI events via acpid. That way, your CPU time probably won't get wasted as cpufreqd and assorted utilities do.