Processor
CPU_DRIVER_OPMODE_ON_AC/BAT
Version 1.6 and newer
CPU_DRIVER_OPMODE_ON_AC=active
CPU_DRIVER_OPMODE_ON_BAT=active
Selects a CPU scaling driver operation mode. Configuration depends on the active driver:
amd-pstate
AMD Zen 2 or newer CPUs provide:
active - requires kernel 6.3 or newer, changes driver name to amd-pstate-epp
passive
guided - requires kernel 6.4 or newer
Since AMD’s kernel documentation is hard to penetrate, here is an attempt to outline the individual modes (without guarantee):
In active mode, the processor selects the operating frequencies autonomously within the limits imposed by the hardware, including turbo boost. The powersave governor can also lead to max frequency. Configuring the min and max frequency (as shown below) is not supported.
With guided mode, you may configure the min and max frequency (see below), and the processor will choose the operating frequency itself in the specified range. Guided is basically active mode with frequency range enforced.
In passive mode, in contrast to guided, the governor dictates the frequencies as configured.
Note
For kernels older than 6.5 amd-pstate must be activated via a kernel boot option; see the driver documentation linked at the bottom.
It should be kept in mind that not all laptop BIOSes allow the activation despite a suitable CPU. Please done not open TLP issues for this.
Some users have observed limited frequency or P-state ranges after switching modes. These are hardware issues or kernel issues, so please do not open a TLP issue for them.
intel_pstate
Intel Core i 2nd gen. (“Sandy Bridge”) or newer Intel CPUs provide these modes:
active
passive
Starting with kernel 5.7, the intel_pstate driver selects passive mode aka intel_cpufreq for CPUs that do not support hardware-managed P-states (HWP), i.e. Intel Core i 5th gen. or older.
acpi-cpufreq
Older AMD/Intel CPUs and other vendors do not support change of driver mode.
CPU_SCALING_GOVERNOR_ON_AC/BAT
CPU_SCALING_GOVERNOR_ON_AC=powersave
CPU_SCALING_GOVERNOR_ON_BAT=powersave
Selects the CPU scaling governor for automatic frequency scaling. Configuration depends on the active driver:
amd-pstate | intel_pstate in active mode
performance
powersave – default
amd-state in passive or guided mode | intel_pstate in passive mode | acpi-cpufreq
conservative
ondemand – default for most distributions
userspace
powersave
performance
schedutil – default for newer kernels and the amd-pstate/intel_pstate drivers in passive mode
Note
Refer to the output of tlp-stat -p to determine the active scaling driver and available governors.
Keep in mind that modularized kernels may not reveal all available governors in the tlp-stat -p output after boot. In this case, configure the desired governor anyway and, after tlp start, check the output of tlp-stat -p again (won’t work in versions 1.4 and 1.5).
Important
Default governors listed above are power efficient for almost all workloads, therefore kernel devs and most distributions have chosen them as such. So before changing the scaling governor please do your research about the advantages and disadvantages of the others (refer to the links at the bottom of this page).
CPU_SCALING_MIN/MAX_FREQ_ON_AC/BAT
CPU_SCALING_MIN_FREQ_ON_AC=0
CPU_SCALING_MAX_FREQ_ON_AC=9999999
CPU_SCALING_MIN_FREQ_ON_BAT=0
CPU_SCALING_MAX_FREQ_ON_BAT=9999999
Set the min/max frequency available for the scaling governor. Possible values depend on your CPU. For available frequencies consult the output of tlp-stat -p.
Hints:
Do not use this setting with the intel_pstate driver in active mode, use CPU_MIN/MAX_PERF_ON_AC/BAT instead
Min/max frequencies must always be specified for both modes i.e. AC and BAT
To enable processor defaults comment all four settings and reboot
Lowering the max frequency on battery power may not conserve power; best results are to be expected from the above mentioned default governors without frequency limits
CPU_ENERGY_PERF_POLICY_ON_AC/BAT
CPU_ENERGY_PERF_POLICY_ON_AC=balance_performance
CPU_ENERGY_PERF_POLICY_ON_BAT=balance_power
Set CPU energy/performance policies (in order of increasing power saving):
performance
balance_performance
default
balance_power
power
Default when unconfigured: balance_performance (AC), balance_power (BAT)
Requirements:
AMD
Version 1.6 and newer
AMD Zen 2 or newer CPU with kernel 6.3 and amd-pstate driver in active mode
Intel
HWP.EPP (hardware-managed P-states): Intel Core i 6th gen. (“Skylake”) or newer CPU with kernel 4.10 and intel_pstate driver in active mode
EPB: Intel Core i 2nd gen. (“Sandy Bridge”) or newer CPU with kernel 5.2 and intel_pstate driver
Note that HWP.EPP and EPB are mutually exclusive. When EPP is available, Intel CPUs will not honor EPB. Only the matching feature will be applied by TLP and shown by tlp-stat -p.
CPU_MIN/MAX_PERF_ON_AC/BAT
CPU_MIN_PERF_ON_AC=0
CPU_MAX_PERF_ON_AC=100
CPU_MIN_PERF_ON_BAT=0
CPU_MAX_PERF_ON_BAT=30
Define the min/max P-state for Intel CPUs. Values are stated as a percentage (0..100%) of the total available processor performance.
Hints:
Requires Intel Core i 2nd gen. (“Sandy Bridge”) or newer CPU with intel_pstate or intel_cpufreq scaling driver
The driver imposes a limit > 0 on the min P-state, see min_perf_pct in the output of tlp-stat -p
This setting is intended to limit the power dissipation of the CPU
CPU_BOOST_ON_AC/BAT
CPU_BOOST_ON_AC=1
CPU_BOOST_ON_BAT=0
Configure CPU “turbo boost” (Intel) or “core performance boost” aka “turbo core” (AMD):
0 - disable
1 - allow
Note
A value of 1 does not activate boosting, it just allows it
For use with the amd-pstate driver, at least kernel 6.11 is required
CPU_HWP_DYN_BOOST_ON_AC/BAT
Version 1.4 and newer
CPU_HWP_DYN_BOOST_ON_AC=1
CPU_HWP_DYN_BOOST_ON_BAT=0
Configure the Intel CPU HWP dynamic boost feature:
0 - disable
1 - enable
Requires: Intel Core i 6th gen. (“Skylake”) or newer CPU with intel_pstate scaling driver in active mode
See also
Settings: Introduction
FAQ: Processor
CPU Performance Scaling – kernel documentation covering scaling governors et al.
amd-pstate CPU Performance Scaling Driver – driver documentation
intel_pstate CPU Performance Scaling Driver – driver documentation
Intel Hardware P-State (HWP) / Intel Speed Shift – a consideration of HWP.EPP
Intel Performance and Energy Bias Hint – EPB documentation
Improvements in CPU frequency management – LWN article covering the schedutil governor
Why EPB is not set when HWP.EPP is available – Arch Linux Forums