For processors supporting CPU performance boosting with higher performance states available beyond the base states, Linux allows toggling the boosting on a per-CPU basis. However, a new patch proposed this week would allow per-policy performance boosting where capable.

Rather than toggling the CPU performance boost state for the entire CPU, HiSilicon engineers are floating the concept of per-policy boosting where a subset of the cores could be prioritized for performance boosting while others would not be exposed to the boost state(s). In turn performance sensitive workloads could be prioritized to those boosted cores.

With the request-for-comments patch, a new " local_boost " is added to the CPUFreq sysfs interface for each policy supported by the CPU to allow this more granular control over CPU boosting and to complement the global boost switch.

More details for those interested via the Linux power management mailing list .