Script(s) work with all roms and kernels and can be imported using new version Kernel Adiutor.
Interactive governor scripts to help save battery and keep a smooth performance. Compared to stock you don't notice any noticeable performance drops and the device also stays cooler. Mostly i reach around 8/9 hours sot using wifi only and low screen brightness around 20/30%. When your device is still new you should be able to reach over 10h sot with very light usage. Its all based and translated from "soniCrons" work who made the "Ghostpepper" script for the nexus 5/6. It will use smooth transitions by defining and using a wider range of CPU speeds.
Important:
Its advised to disable the "performance daemon" service by creating a little init.d scripts using kernel adiutor.
Its called "perfd" and fires up the big cores when touching the screen. I always stop this service to have full control and avoid bugs with governor settings reverting (go_hispeed_load, hispeed_freq, cpu speed,boost) and keep the cpu's more quiet.
You can stop this "perfd" service from within k.a. this way :
1. Press "+" in the init.d section to create a new script.
2. Name it "StopPerfd" and press "ok". When the script opens type "stop perfd" (without quotes) and save.
3. Select the script and when asked to "execute" choose "ok" and finished.
Perfd restarts when rebooting the phone or starts manually by creating another script like above and use "start perfd" instead of "stop perfd". To launch at boot enable the "emulate init.d" switch, but remember it will launch all the scripts you have placed in the init.d folder in the order they appear.
Importing scripts:
You can import profiles with the new kernel adiutor and each profile contains 2 files, one for the "big CPU" and one for the "small CPU". Once imported you can apply them on boot by tapping the 3dot menu on the profile you just created/imported and select "on boot" or launch them manually.
Please don't use the "apply on boot" switch in the CPU section, because it can mess with the settings and may not apply them correctly. Rebooting is back to stock settings.
Universal script (based on Ghostpepper and also works nice using msm-hotplug or cluster-plug)
Small cpu = http://ift.tt/1MNbGj6
Big cpu = http://ift.tt/1MNbGj6
Ghostpepper "translated" (both big and little cores must be online to work best):
Small cpu = http://ift.tt/1MNbGj6
Big cpu = http://ift.tt/1MNbGj6
Custom controls made for new version kernel adiutor :
A new feature using new version of k.a. is that you can now create your own controls and add them in this new "custom control" section.You can add on/off switches and seekbars (sliders) to control system/cpu settings that are not yet supported.
I created some HMP sliders to fine tune the big and little cluster and the way loads are handled between the two. HMP scheduler works best when both big and little cpu's are online. HMP doesnt work when using a msm-hotplug kernel (msm hotplug use smp sheduler). Ex kernel manager already added some of those tunables, but now you can also add them using k.a. and control them in real-time without editing and launching scripts.
They have a minimum effect on performance and battery life, but could be handy to fine-tune the big (performance) and little (power) cluster. HMP tunables are save to use, because the tunables are more like defining suggestions for the system/cpu to take into future calculations. Adviced is to use a cpu monitor tool like "cpu stats" or "floating monitors" to see what it does in the background or you wont notice much difference.
Also made 2 sliders to control the cluster-plug module. You can now control the load threshold up and down.
Works great if you want cluster-plug to act more or less responsive firing up the little cores.
Important:
One little issue when using the "custom control" section is when you leave this section and return later, all the sliders and switches showing the default settings/values again and not the real applied values from last time you visited. It can't read values and only writes them at the moment, but it works and applies them.
The default values you see are not really applied and only start working the moment you touch/change them.
All the defaults values and descriptions are included in the controls and rebooting also resets them.
Save to use and if a control is not supported it just doesn't apply it and nothing changes.
Cluster-plug on/off switch = http://ift.tt/1MNbGj6
Cluster-plug load threshold up = http://ift.tt/1MNbGj6
Cluster_plug load threshold down = http://ift.tt/1MNbGj6
HMP sheduler upmigrate = http://ift.tt/1MNbGj6
HMP scheduler downmigrate = http://ift.tt/1MNbGj6
HMP scheduler small task = http://ift.tt/1MNbGj6
HMP scheduler window stat policy = http://ift.tt/1MNbGj6
HMP scheduler history size = http://ift.tt/1MNbGj6
HMP enable/disable power aware = http://ift.tt/1MNbGj6
Next switches are interactive governor tunables and are useful to achieve lower idle rates and extra smoothness.
Big cores- enable/disable migration = http://ift.tt/1MNbGj6
Big cores- enable/disable use sched load = http://ift.tt/1MNbGj6
Small cores - enable/disable migration = http://ift.tt/1MNbGj6
Small cores - enable/disable use sched load = http://ift.tt/1MNbGj6
Enjoy.
Interactive governor scripts to help save battery and keep a smooth performance. Compared to stock you don't notice any noticeable performance drops and the device also stays cooler. Mostly i reach around 8/9 hours sot using wifi only and low screen brightness around 20/30%. When your device is still new you should be able to reach over 10h sot with very light usage. Its all based and translated from "soniCrons" work who made the "Ghostpepper" script for the nexus 5/6. It will use smooth transitions by defining and using a wider range of CPU speeds.
Important:
Its advised to disable the "performance daemon" service by creating a little init.d scripts using kernel adiutor.
Its called "perfd" and fires up the big cores when touching the screen. I always stop this service to have full control and avoid bugs with governor settings reverting (go_hispeed_load, hispeed_freq, cpu speed,boost) and keep the cpu's more quiet.
You can stop this "perfd" service from within k.a. this way :
1. Press "+" in the init.d section to create a new script.
2. Name it "StopPerfd" and press "ok". When the script opens type "stop perfd" (without quotes) and save.
3. Select the script and when asked to "execute" choose "ok" and finished.
Perfd restarts when rebooting the phone or starts manually by creating another script like above and use "start perfd" instead of "stop perfd". To launch at boot enable the "emulate init.d" switch, but remember it will launch all the scripts you have placed in the init.d folder in the order they appear.
Importing scripts:
You can import profiles with the new kernel adiutor and each profile contains 2 files, one for the "big CPU" and one for the "small CPU". Once imported you can apply them on boot by tapping the 3dot menu on the profile you just created/imported and select "on boot" or launch them manually.
Please don't use the "apply on boot" switch in the CPU section, because it can mess with the settings and may not apply them correctly. Rebooting is back to stock settings.
Universal script (based on Ghostpepper and also works nice using msm-hotplug or cluster-plug)
Small cpu = http://ift.tt/1MNbGj6
Big cpu = http://ift.tt/1MNbGj6
Ghostpepper "translated" (both big and little cores must be online to work best):
Small cpu = http://ift.tt/1MNbGj6
Big cpu = http://ift.tt/1MNbGj6
Custom controls made for new version kernel adiutor :
A new feature using new version of k.a. is that you can now create your own controls and add them in this new "custom control" section.You can add on/off switches and seekbars (sliders) to control system/cpu settings that are not yet supported.
I created some HMP sliders to fine tune the big and little cluster and the way loads are handled between the two. HMP scheduler works best when both big and little cpu's are online. HMP doesnt work when using a msm-hotplug kernel (msm hotplug use smp sheduler). Ex kernel manager already added some of those tunables, but now you can also add them using k.a. and control them in real-time without editing and launching scripts.
They have a minimum effect on performance and battery life, but could be handy to fine-tune the big (performance) and little (power) cluster. HMP tunables are save to use, because the tunables are more like defining suggestions for the system/cpu to take into future calculations. Adviced is to use a cpu monitor tool like "cpu stats" or "floating monitors" to see what it does in the background or you wont notice much difference.
Also made 2 sliders to control the cluster-plug module. You can now control the load threshold up and down.
Works great if you want cluster-plug to act more or less responsive firing up the little cores.
Important:
One little issue when using the "custom control" section is when you leave this section and return later, all the sliders and switches showing the default settings/values again and not the real applied values from last time you visited. It can't read values and only writes them at the moment, but it works and applies them.
The default values you see are not really applied and only start working the moment you touch/change them.
All the defaults values and descriptions are included in the controls and rebooting also resets them.
Save to use and if a control is not supported it just doesn't apply it and nothing changes.
Cluster-plug on/off switch = http://ift.tt/1MNbGj6
Cluster-plug load threshold up = http://ift.tt/1MNbGj6
Cluster_plug load threshold down = http://ift.tt/1MNbGj6
HMP sheduler upmigrate = http://ift.tt/1MNbGj6
HMP scheduler downmigrate = http://ift.tt/1MNbGj6
HMP scheduler small task = http://ift.tt/1MNbGj6
HMP scheduler window stat policy = http://ift.tt/1MNbGj6
HMP scheduler history size = http://ift.tt/1MNbGj6
HMP enable/disable power aware = http://ift.tt/1MNbGj6
Next switches are interactive governor tunables and are useful to achieve lower idle rates and extra smoothness.
Big cores- enable/disable migration = http://ift.tt/1MNbGj6
Big cores- enable/disable use sched load = http://ift.tt/1MNbGj6
Small cores - enable/disable migration = http://ift.tt/1MNbGj6
Small cores - enable/disable use sched load = http://ift.tt/1MNbGj6
Enjoy.
from xda-developers http://ift.tt/2b3wnw6
via IFTTT
No comments:
Post a Comment