Linux 6.6's cpupower Utility Enables New AMD P-State Features

Linux's cpupower utility lives within the Linux kernel source tree for reading and tuning various CPU power settings rather than poking at sysfs files directly or other means of adjusting your processor power-related tunables. With the upcoming Linux 6.6 kernel cycle the cpupower utility is adding support for adjusting new AMD P-State driver features.

Among the new cpupower capabilities that will be found in the Linux 6.6+ source tree include:

- Support for changing the AMD P-State mode. The AMD P-State driver allows dynamically changing its mode of operation between the passive, active, guided autonomous modes of operation. The new "amd-pstate-mode" option for the CPU Power utility will allow adjusting the AMD P-State mode rather than having to deal with the driver's sysfs interface directly. This option can also be set at boot-time using the amd_pstate= option on existing kernels.

- More general, there is also the "turbo-boost" option with the cpupower utility for processors exposing the cpufreq/boost sysfs interface as seen on Intel, AMD, and some Arm processors. Via cpupower set --turbo-boost 1 can activate the CPU frequency boosting mode or "0" for disabling the boost mode if desired.

- For benefiting both AMD and Intel P-State drivers, there is Energy Performance Preference (EPP) value change support. This allows adjusting the EPP value from the cpupower utility for indicating your energy/performance preference that in turn ultimately influences the CPU frequency scaling behavior. This is exposed via a new "epp" option for cpupower.

These feature additions to cpupower were submitted on Monday to the Linux power management subsystem's "-next" branch ahead of the Linux 6.6 merge window opening around the end of August. Again, these features can be dealt with directly right now on current Linux kernel versions via sysfs interfaces while AMD added these options to cpupower to make it easier on end-users and if wanting to set the desired values via the cpupower systemd service or similar handling. Excitingly it's with the current Linux 6.5 cycle where AMD P-State EPP "active" is the new default for Ryzen CPUs.