Huawei engineer Pengjie Zhang has submitted a patch to the Linux kernel proposing that the CPU Boost mode in the ACPI CPPC CPUFreq driver be enabled by default. The proposed change involves modifying the cppc_cpufreq.c file with 20 lines of new code to initialize the default Boost state.
Currently, the ACPI CPPC driver disables Boost by default when a processor's highest performance level exceeds its nominal performance level. Users are required to manually enable the feature via sysfs. Huawei suggests that if at least one CPU supports Boost, the feature should be automatically enabled for all policies that support it.
Arm engineer and Linux developer Christian Loehle has challenged this proposal. He argues that modern processor power and thermal constraints are too complex to guarantee that Boost will consistently deliver performance without throttling. Loehle suggests that expert users should manage performance limits manually via scaling_max_freq.
Loehle also questioned the practical necessity of Boost given current thermal boundaries and noted that platforms where nominal performance differs from maximum performance can create operational difficulties. The final decision on whether to merge this patch rests with the Linux kernel maintainers.
