Re: Better control for adaptive performance mode
I would even go a step further on this and have them use the cpufreq modules in the kernel as a source of reference by presenting several options:
A.) Powersave mode (always lowest speed and energy consumption possible.)
B.) Conservative mode (hesitant to switch to a higher/lower speed until load persists long enough to warrant change)
C.) Ondemand mode (always switches up and down quickly dependent on current load)
D.) Performance mode (always on highest speed possible)
E.) Userspace mode (definable criteria on what is acceptable performance)
I'd even argue that some of these variables, like cpufreq, could be controlled in /sys space.