[kernel hacking] MPdecision

[Guest rederik75]

When talking about kernel hacks, we always hear the "mantra":

disable MPdecision, disable MPdecision, disable MPdecision, disable MPdecision!

I like to understand what I'm doing so I've made some research, coming across this article: http://timos.me/blog/mp-decision-134 by the author of TricksterMOD.

It seems quite a wise argumentation, so I decided to give it a try, enabling it again in trickster, let's see...

 

What do you think about, guys?

[Guest thetonyxx]

you dont understand mpdecision function on y300?

simple, mpdecision active can deactivated second core during low cpu use/iddle

[Guest ZolaIII]

Hot plug function is to power on off CPU cores & how they will go in possible possible states, nothing more.

Problem with a implemented Qualicomms MP decision is; 1) political as its property closed source distributed binary blob & 2) functional as we can't optimize its behavior more.

Lot of people where not sadisfied wait Qualicomms MP decision so now we have a lot of hot plugs written & most of them work worse than MP decision.

One of normal works that is wort of implementing is a in kernel MP decision from Code Aurora. Main point of adjusting it is to get inactive cores to go in "dormant" state instead of power off state when machine is not in a deep sleep mode enabling the CPU cores to get in a operational state much quicker & consume only a fragment more of power (when a core needs to go from power off state to on it will consume more & power & time then to go on from dormant state)...