[TWEAK](SIYAH/REDPILL/N.E.A.K.)MagicConfig for VC & NOW for ExTweaks(GB/ICS)

Status
Not open for further replies.
Search This thread

droidphile

Senior Member
Jul 19, 2011
1,119
7,103
Bangalore
Samsung Galaxy S20
No. smartassv2 has.. great governor:D and lulzactive have
SmartassV2 does not have a screen-off profile. (Smartass has). It has only a "sleep_ideal_freq" to which the governor ramp down CPU aggressively when screen is turned off.
If some task demands more power during screen off, the frequency can go all the way upto maximum frequency depending on the load/demand.

Lulzactive has screen-off profile and the max-freq for scree-off state is configurable as well. :)
 
  • Like
Reactions: It_ler and mekhonte

Entropy512

Senior Recognized Developer
Aug 31, 2007
14,088
25,086
Owego, NY
Yes, i have the same opinion.
Second graph in this thread by Ezekeel, shows the difference between lower and higher frequency steps' battery consumption with and without LPA/deep idle. As it occurs, higher frequency supports race-to-idle better there :)

Again, "100 mhz bad compared to 200 mhz" depends on we're using lpa or not. (sched_mc=2 or 3).

Anyhow, since only first core can hit lpa, screen-off profiles with: different governor, hard limiting frequency, switching to single core mode, etc actually does a little harm than good by delaying for race-to-idle.

Yup - thanks to stuff like idle states and multiple cores, many of the old "rules of thumb" have become FAR more complicated.

In my opinion - hard-limiting to 200 is bad. I hard-limit to 500 as a safety measure - voltage increases a decent amount at 800, and if something drives more than a blip at 800, something is wrong.

Deciding when to turn on/off the second core is much tougher - having it on gets you "done quicker" in theory - but turning it on takes time, and some tasks run less efficiently when spread across cores. Also it may be better to hit 800 MHz than to run both cores at 500.
 
S

Suarez7

Guest
SmartassV2 does not have a screen-off profile. (Smartass has). It has only a "sleep_ideal_freq" to which the governor ramp down CPU aggressively when screen is turned off.
If some task demands more power during screen off, the frequency can go all the way upto maximum frequency depending on the load/demand.

Lulzactive has screen-off profile and the max-freq for scree-off state is configurable as well. :)

That is a weird one as i always thought smartassv2 had that profile already implemented in it. Honestly i prefer smartassv2 compare to lulz cause phone sleeps better and when in idle does not lose as much batter as lulz would. Thanks for explaining friend

Sent from my GT-I9100 using xda premium
 

droidphile

Senior Member
Jul 19, 2011
1,119
7,103
Bangalore
Samsung Galaxy S20
Yup - thanks to stuff like idle states and multiple cores, many of the old "rules of thumb" have become FAR more complicated.

In my opinion - hard-limiting to 200 is bad. I hard-limit to 500 as a safety measure - voltage increases a decent amount at 800, and if something drives more than a blip at 800, something is wrong.

Deciding when to turn on/off the second core is much tougher - having it on gets you "done quicker" in theory - but turning it on takes time, and some tasks run less efficiently when spread across cores. Also it may be better to hit 800 MHz than to run both cores at 500.
Yup, 'rules of thumb' are no longer valid. It's the world where "ondemand" can save juice, "conservative" can conserve nothing, and so on, with more and more sophisticated tunables in the user space.

800 has to be the 'weirdest' frequency. Some devices enters deep sleep at 800 (when user configures so), device need to be @800 for a few ms when awaken, voltage required considerably increase @800, and so on. With voodoo louder, equalizer and powerampp contributing to music-play in screen-off state, CPU spends most of the time at 100/200 and negligibly little at 500. So 500 is the ideal-high screen-off cap freq. (When device hits deep sleep, this is however overriden and cpu set to 800 or 1000 for the req time anyways)

Using hotplugging and single core mode is sometimes funny if you think of a task that cause low load on cpu0.(assuming low UP thresh & high LOW thresh for hotplugging) The time taken to switch cpu1 plus time to turn it off (when low threshold is found) plus the time taken to do the task may be more than the time taken for cpu0 in single core mode to do the same task.
 
  • Like
Reactions: CamoGeko

XDA mark

Senior Member
Dec 30, 2009
1,600
449
Rotterdam (NL) & Timisoara (RO)
I wanted to post feedback for geko: raising all your UV settings with 50mV has kept my device all day running without a single problem. Maybe I will try to go down 25 mV again to get even closer to your setup. To bad my device can't handle your config, but like this it's close ;)
 

CamoGeko

Retired Forum Moderator
Kudos to what geko has done for the many once-newbies on here - he has provided a stable platform on which newbies can test, and to improve on. I for one, greatly appreciate his help. :D

Exactly the point of this config. :)

Thanks for your understanding mate.

+1 now i don't have to feel bad about bugging you for your latest config. i can just refer to this page. nice! :)

Yeah, for me it's no more sending out 50 PMs per day lol :p

I wanted to post feedback for geko: raising all your UV settings with 50mV has kept my device all day running without a single problem. Maybe I will try to go down 25 mV again to get even closer to your setup. To bad my device can't handle your config, but like this it's close ;)

Dude, so my suggestion is working fine. I'd say leave it for a bit longer then if everything else is fine, start reducing each step by 25mV. Remember to give a bit of time to run it each step at reduced voltage before going onto the next frequency.

Good luck. :)
 
  • Like
Reactions: bolek76 and NJGSII

NJGSII

Senior Member
Nov 15, 2011
1,785
451
Howell, NJ
Been running your settings for over 12 hours now. So far everything seems good. Been using the noop/ conservative scheduler governor combo. Put the device through the paces and stability wasn't an issue.

Thanks!

Will report back on battery life tomorrow sometime.
 

CamoGeko

Retired Forum Moderator
Been running your settings for over 12 hours now. So far everything seems good. Been using the noop/ conservative scheduler governor combo. Put the device through the paces and stability wasn't an issue.

Thanks!

Will report back on battery life tomorrow sometime.

Thanks for the feedback mate, sounds like all is working how it should be. :)
 

jermitano

Senior Member
Oct 8, 2011
1,302
884
Manila
at last this thread is growing...cheers mate! btw after testing the andreilux script on defaults ill be trying it tomorrow with your settings...ill see if it makes a diff...currently im on defaults touching nothing from VC...ill try to make a comparison after using up my current set up...

btw...a lot of us (not just me but several buds from our fb thread) are experiencing faster battery drain when wifi is on (on defaults or even when using your cfg)...is there a bug in the latest siyah kernel?
 

NJGSII

Senior Member
Nov 15, 2011
1,785
451
Howell, NJ
btw...a lot of us (not just me but several buds from our fb thread) are experiencing faster battery drain when wifi is on (on defaults or even when using your cfg)...is there a bug in the latest siyah kernel?


You know what, now that you mention it I may have had the same thing happened.

Whlist listening to music today with the screen off and wifi left on I experienced higher drain than normal. I think it was 2% higher than normal. Not much, but it's there.

I'm thinking about reverting to 2.6.9 and retest using the exact same conditions above.
 

maccawire

Member
Dec 20, 2010
44
15
Mullingar
Samsung Galaxy A51
More feedback for you geko. Been using your magic config for about a week now and my phone has been as smooth as a ken doll.
I use bfg and ondemand 200Mhz - 1000Mhz plus all your voltages with the exception 200Mhz I use 875mv as it causes me no issues.
Getting around 48hrs uptime (from 24hrs - 32 hrs) which means the battery saving is great (and if I cared about the environment greener also)

Thanks again.
 
  • Like
Reactions: CamoGeko

jermitano

Senior Member
Oct 8, 2011
1,302
884
Manila
You know what, now that you mention it I may have had the same thing happened.

Whlist listening to music today with the screen off and wifi left on I experienced higher drain than normal. I think it was 2% higher than normal. Not much, but it's there.

I'm thinking about reverting to 2.6.9 and retest using the exact same conditions above.

we actually did a group test on it just to check...3 who are on default and 3 who are on the cfg...same thing...with wifi on the batt drain was higher irregardless of the config and even with the andreilux script...ive tried to ask that on the siyah thread before but i guess gokhan is busy...anyways im just on the draining mode and will try the andreilux+magic cfg combi to see what happens...
 

NJGSII

Senior Member
Nov 15, 2011
1,785
451
Howell, NJ
we actually did a group test on it just to check...3 who are on default and 3 who are on the cfg...same thing...with wifi on the batt drain was higher irregardless of the config and even with the andreilux script...ive tried to ask that on the siyah thread before but i guess gokhan is busy...anyways im just on the draining mode and will try the andreilux+magic cfg combi to see what happens...

I see, I see.

I'm already doing what I said I'll be doing right now. I'll know in about a half hour, what's up.
 

Entropy512

Senior Recognized Developer
Aug 31, 2007
14,088
25,086
Owego, NY
You know what, now that you mention it I may have had the same thing happened.

Whlist listening to music today with the screen off and wifi left on I experienced higher drain than normal. I think it was 2% higher than normal. Not much, but it's there.

I'm thinking about reverting to 2.6.9 and retest using the exact same conditions above.

Details?
If you had Android OS drain increase and deep sleep drop way down when wifi was on (but not before)- it could be due to changes in ARP offloading and broadcast traffic filters. (Actually, since Siyah is only available patched up to .14, the only symptoms of this will be the deep sleep dropping, and lots of wlan_rx and wlan_wake wakelocks in the kernel wakelocks list)

If you have intermittent high drain when connected to wifi, but no indication in any diagnostics of why (deep sleep still high, no high AOS usage on .7 kernels, etc.) - I've seen cases where my drain skyrockets to 2-2.5%/hour despite still maintaining deep sleep in my own kernels ever since finishing the process of patching up to the I9100 update3 codebase. It's intermittent, which makes it frustrating as hell to debug.
 

NJGSII

Senior Member
Nov 15, 2011
1,785
451
Howell, NJ
Details?
If you had Android OS drain increase and deep sleep drop way down when wifi was on (but not before)- it could be due to changes in ARP offloading and broadcast traffic filters. (Actually, since Siyah is only available patched up to .14, the only symptoms of this will be the deep sleep dropping, and lots of wlan_rx and wlan_wake wakelocks in the kernel wakelocks list)

If you have intermittent high drain when connected to wifi, but no indication in any diagnostics of why (deep sleep still high, no high AOS usage on .7 kernels, etc.) - I've seen cases where my drain skyrockets to 2-2.5%/hour despite still maintaining deep sleep in my own kernels ever since finishing the process of patching up to the I9100 update3 codebase. It's intermittent, which makes it frustrating as hell to debug.
This was when I was listening to music actually. Had screen off for an hour and just using the music app. Wifi was on the whole entire time. on 2.6.9 I noticed only a 2% drain an hour. On 2.6.11 it was 4 % an hour with same exact conditions as above.

Almost done testing it again this time with 2.6.9.

edit: just got done testing and battery life was slightly better with 2.6.9 vs 2.6.11, but when I went to CPU spy I noticed something that was way different.

In 2.6.11 the amount of time the phone was in 200 mhz vs 100 mhz was 36% 34% respectively.

In 2.6.9 it was 1% in 200 mhz, 86% in 100 mhz.

Interesting.
 
Last edited:
  • Like
Reactions: Simba501
Status
Not open for further replies.

Top Liked Posts

  • There are no posts matching your filters.
  • 315
    MagicConfig (previously "mycfg")


    Universal UV tweak to help you easily get an great improvement in battery life, Gingerbread and Ice Cream Sandwich Compatible!
    First step for any Undervolting beginner.


    INITIAL STORY

    Due to receiving endless requests few months back for me to post my "MagicConfig" for SiyahKernel and as I'm typing this right now I still have to send out 20+ PMs every single day on this subject. So I have finally decided to make a thread for it, now it can be discussed and found here without needing to contact me.

    Obviously I only wanted to do this after extensive "real world" testing myself and MagicConfig's initial user's confirmation that it works astonishingly well, since then it has been picked up and loved by many many users more. IT TOOK ME OVER 4 MONTHS TO MAKE SURE THIS CONFIG WOULD BE COMPATIBLE WITH ALMOST ALL SGS2s. HOWEVER, PLEASE READ THE WHOLE POST CAREFULLY BEFORE TRYING TO APPLY THESE SETTINGS.


    Now, what you will find below is not a script of any kind so there is no need to download or flash anything. All you need to do is simply input these values into Voltage Control app which can be downloaded free from the Market.

    I have done my best to test the configuration and to my knowledge it should work on most of Galaxy S2 devices out there. It is designed to bring balance between smoothness and battery life, I have spent over 3 months (since the betas) going through all of SiyahKernel releases to find the perfect combinations of Schedulers and Governors that are included within each release of the Kernel. If you do have issues or if you needed to make alterations of any kind to make it work for you (not including reducing voltages further), please report this here. Thank You.


    ----------​

    UPDATE - 18/06/2012:

    MagicConfig SiyahKernel ICS Tweaks - http://goo.gl/ykgQ0[/B]

    Requirements and description below!

    UV Settings for other kernels can be found in this thread by ram3n:
    http://xdaforums.com/showthread.php?t=1691129

    ----------​

    6+ HOURS BATTERY LIFE IS TRULY POSSIBLE!! NOT JUST A MYTH!!
    http://xdaforums.com/showpost.php?p=21852140&postcount=2

    READ SOME FEEDBACK FROM CURRENT MAGICCONFIG USERS!!
    http://xdaforums.com/showpost.php?p=23081634&postcount=3

    ----------


    * CURRENT PHAZE (MagicConfig ICS):

    MagicConfig v1.0, tested and working with SiyahKernel v3.2.7.1 and v3.3.2!! Earlier versions of this kernel should be suitable as well.

    * REQUIREMENTS:

    You need to be running SiyahKernel (latest is always recommended) and have the eXtweaks application installed.

    SiyahKernel ICS by gokhanmoral development thread is here:
    http://xdaforums.com/showthread.php?t=1555259

    eXtweaks app is here:
    https://play.google.com/store/apps/details?id=com.darekxan.extweaks.app

    * INSTRUCTIONS ON WHAT TO DO NEXT:

    Go to this post and follow the settings there:
    http://goo.gl/ykgQ0


    ----------​

    * CURRENT PHAZE (MagicConfig GB):

    MagicConfig v0.3.2, tested and working with SiyahKernel v2.6.11, SiyahKernel v2.6.12 SiyahKernel v2.6.13, RedPillKernel v1.2, v1.3 and N.E.A.K. 1.3.3x, 1.4x, 1.4.1x and 1.4.2x.

    Have to say, best balance between performance and battery life with this config can be achieved on N.E.A.K. Kernel. Earlier versions of any of these kernels should be suitable as well.

    * REQUIREMENTS:

    Any of the above mentioned kernels, just do a search on XDA and you shall find them.


    -_-_-_-_-_-_-_-_-_-_-

    * GOVERNOR & SCHEDULER MATCH-UP:

    SiyahKernel
    • BFQ / ONDEMAND
    • NOOP / CONSERVATIVE
    • SIO / SMARTASSV2

    RedPillKernel
    • DEADLINE / SLEEPY
    • NOOP / CONSERVATIVE
    • SIO / HYPER

    N.E.A.K. Kernel
    • BFQ / ONDEMAND (some have reported slight lags)
    • NOOP / CONSERVATIVE (with Lionheart tweaks)
    • NOOP / LAZY (best balance so far for me, maybe it sensed my own lazyness lol) (SCHED_MC and EXT4boost enabled)
    *** Please test them out and find which combination works best for you ***


    ----------​

    * CPU:


    • Undervolted Range / 100Mhz - 1400Mhz
      Battery Saver Setting / 200Mhz - 1000Mhz

    + VOLTAGES:

    • 100Mhz - 900mV
    • 200Mhz - 900mV
    • 500Mhz - 950mv
    • 800Mhz - 1000mv
    • 1000Mhz - 1100mV
    • 1200Mhz - 1200mV
    • 1400Mhz - 1300mV
    *** N.E.A.K. Kernel only allows highest frequency of 1200Mhz ***
    *** RedPill Kernel can't handle these voltages on some devices, if you get freezes or reboots please raise all voltages by +25mV ***



    * GPU:


    • 114Mhz low / 267Mhz high
    *** Latest SiyahKernel v2.6.13 doesn't allow step of 114Mhz, please use 133Mhz instead ***


    + VOLTAGES:

    • 850mV low
    • 950mV high

    ----------​

    * OTHER INFO:

    Mainly at the moment I just stick to NOOP + ONDEMAND on N.E.A.K. 2.0.4x, running CyanogenMOD 9 and CPU at 200Mhz - 1000Mhz.

    As I always say though, these voltage settings are a safe stable base and could be tweaked to your liking once you get experienced with undervolting.

    Also can I ask you kindly not to post this config in any other threads, instead if someone asks for it then just direct them to this thread please. Thanks!


    -_-_-_-_-_-_-_-_-_-_-

    * SCREENSHOTS (this is what it should look like after all parameters are set):

    BaKnco6J
    6QcLrs6J
    rkaDyErt
    sOYiLjtb
    vMgOFNjC


    ----------

    * NOTICE:

    That's about it. If you want to save even more battery juice then just reduce the highest CPU frequency down to 1000Mhz or even 800Mhz.

    Also, I found that having the lowest frequency set to 200Mhz improves smoothness quite a bit and doesn't impact battery too much, of course it's also slightly more stable. Potentially, this can conserve even more battery life.

    Do not have multiple CPU / Voltage configuration apps installed at the same time, you will most likely get conflicts.

    If you start having issues (screen going black, freezes or reboots then raise all voltages by +25mV) then everything should work as normal. If it doesn't, then raise by another +25mV.

    Enjoy the smoothness everyone. As always, feedback is greatly welcomed and is very helpful to others. Thank you!
    :)


    ----------​

    * FURTHER INFO:

    I know that time will come when you will want to use Lulzactive governor which can be tweaked to your liking using the Lulzactive app. Here is a thread by Stifler69 which has some excellent settings which you can start with:
    http://xdaforums.com/showthread.php?t=1525048

    If you want to learn more about Overclocking and Undervolting as well as get an explanation on Governors, Schedulers and general terminology please visit this amazing thread by droidphile here:
    http://xdaforums.com/showthread.php?t=1369817

    To know the proper way of adjusting the values yourself, go to this great thread by pikachu01:
    http://xdaforums.com/showthread.php?p=18549017 (press CTRL + F when you're there and type in Undervolting)


    * THANKS:

    *** gokhanmoral (for creating such a wonderful Kernel, you are the man!)
    *** pongster and hyperdroid team (the Kernel is so amazing that it feels like I'm taking a trip to see how far the rabbit hole goes!)
    *** simone201 (this new lightweight kernel you created is so awesome, it's a perfect pairing with CheckROM and my MagicConfig!)
    *** xan (for making Voltage Control so compatible with Siyah and RedPill, that GUI is a treat to use, so easy!)
    *** stifler69 (for being really helpful to others all around the Galaxy S2 forum section and providing good battery saving Lulzactive app settings!)
    *** all Siyah users (those that requested my config in the early stages and reported feedback, you have really made it what it is today!)


    ----------​

    FINAL WORDS, PLEASE READ:

    IF YOU HAVE TESTED THIS CONFIG AND RECEIVED EXCELLENT BATTERY LIFE AND SMOOTH PERFORMANCE LIKE THE OTHER 900+ USERS, THEN ALL I ASK IS THAT YOU RATE THE THREAD 5* STARS AT THE TOP OF THIS PAGE AND CLICK THANKS BELOW THIS POST.

    ALSO IT HELPS NEW USERS IF YOU PUT A LINK TO THIS THREAD IN YOUR SIGNATURE, THEN THEY CAN EASILY FIND IT AND RECEIVE THE SAME IMPROVEMENTS AS YOU ARE GETTING FROM IT.

    SIGNATURE CODE:

    PHP:
    [B][URL="http://goo.gl/eUPmA"]MagicConfig UV Tweaks[/URL][/B]
    86
    CONFIRMED settings for SiyahKernel ICS v3.2.7.1 / v3.3.2 + CM9 and beyond

    Ok, here we go as promised...

    *** Again though, please be aware that these have not yet been tested thoroughly and some voltages may be too low for some of you... if that happens then please consult me here and I shall guide you. Also, be aware that some of you may have varied performance due to being on different ROMS and having different apps installed. Do not blame me if you don't get exactly the same results as me or someone else. PLEASE READ THE WHOLE POST!! ***

    Also, to avoid being stuck. Please make a Nandroid backup before making any drastic UV changes. I will not be held responsible if your device turns into molten lava and burns a hole in everything it touches. Also, if you do have issues then please report them in this thread and consult with others instead of sending me private messages. Thank you!


    -------

    EXTREME UV VALUES


    REQUIRED:

    eXtweaks app and SiyahKernel ICS (v3.2.7.1 or v3.3.2 recommended)


    SETTINGS (CPU):

    • GENTLE_FAIR_SLEEPERS = On
    • ARCH_POWER = On
    • CPU Hotplug = Default
    • CPU IDLE Mode = AFTR + LPA (default)
    • Smooth Scaling Level = 800Mhz
    • SCHED_MC = 2
    • CPU Undervolting = 100mV (all frequencies)
    • CPU Step Count = 18 (all available)
    • Default CPU Governor = pegasusq
    • Default CPU Scheduler = sio
    • Scaling Max Freq = 1000Mhz
    • Scaling Min Freq = 200Mhz

    All other CPU settings as stock.


    SETTINGS (GPU freq):

    • GPU Freq Step 1 = 40Mhz
    • GPU Freq Step 2 = 100Mhz
    • GPU Freq Step 3 = 200Mhz

    SETTINGS (GPU voltages):

    • GPU Voltage Level 1 = 800mV
    • GPU Voltage Level 2 = 800mV
    • GPU Voltage Level 3 = 900mV

    All other GPU settings as stock.


    SETTINGS (other):

    Screen settings stock. Vibration intensity = 3
    Android logger enabled.

    -------

    LESS AGGRESSIVE UV VALUES


    SETTINGS (CPU):

    • GENTLE_FAIR_SLEEPERS = On
    • ARCH_POWER = On
    • CPU Hotplug = Default
    • CPU IDLE Mode = AFTR + LPA (default)
    • Smooth Scaling Level = 800Mhz
    • SCHED_MC = 2
    • CPU Undervolting = -75mV (if you get reboots or freezes use -50mV)
    • CPU Step Count = 18 (All available)
    • Default CPU Governor = pegasusq
    • Default CPU Scheduler = sio
    • Scaling Max Freq = 1000Mhz
    • Scaling Min Freq = 200Mhz

    All other CPU settings as stock.


    SETTINGS (GPU freq):

    • GPU Freq Step 1 = 40Mhz
    • GPU Freq Step 2 = 100Mhz
    • GPU Freq Step 3 = 200Mhz

    SETTINGS (GPU voltages):

    • GPU Voltage Level 1 = 800mV
    • GPU Voltage Level 2 = 800mV
    • GPU Voltage Level 3 = 950mV

    All other GPU settings as stock.


    SETTINGS (other):

    Screen settings stock. Vibration intensity = 3
    Android logger enabled.

    -------

    RELAXED UV VALUES


    SETTINGS (CPU):

    • GENTLE_FAIR_SLEEPERS = On
    • ARCH_POWER = On
    • CPU Hotplug = Default
    • CPU IDLE Mode = AFTR + LPA (default)
    • Smooth Scaling Level = 800Mhz
    • SCHED_MC = 2
    • CPU Undervolting = -50mV (if you get reboots or freezes, use -25mV)
    • CPU Step Count = 18 (All available)
    • Default CPU Governor = pegasusq
    • Default CPU Scheduler = sio
    • Scaling Max Freq = 1000Mhz
    • Scaling Min Freq = 200Mhz

    All other CPU settings as stock.


    SETTINGS (GPU freq):

    • GPU Freq Step 1 = 66Mhz
    • GPU Freq Step 2 = 133Mhz
    • GPU Freq Step 3 = 267Mhz

    SETTINGS (GPU voltages):

    • GPU Voltage Level 1 = 800mV
    • GPU Voltage Level 2 = 850mV
    • GPU Voltage Level 3 = 950mV

    All other GPU settings as stock.


    SETTINGS (other):

    Screen settings stock. Vibration intensity = 3
    Android logger enabled.


    SETUP SCREENSHOTS (less aggressive UV values) (thanks to juzman for these):

    1-CPU.png


    2-GPU.png


    3-Screen.png


    4-BLN.png


    5-Misc.png



    BATTERY RUN SCREENSHOTS (used less aggressive UV values):

    5609d05b-8e11-550d.jpg


    5609d05b-8df3-6a1c.jpg


    5609d05b-8e04-03e0.jpg


    ------- 30 MINS LATER -------

    5609d05b-91bc-f9e8.jpg


    5609d05b-91cf-3e54.jpg


    5609d05b-91dc-b270.jpg


    That was after a day of medium-heavy use on SiyahKernel v3.2.7.1 and CM9 Nightly 08-06-2012 (flashed in the morning, so didin't even let the ROM settle). Easily 6 hours screen on time guys, I'm sure you will agree that it could easily stretch past 7 hours if I let the ROM and kernel settle. :D

    -------​

    That's it really, one other thing I want to mention is that I do not use auto-brightness (unless I am outside). Usually my brightness meter is at around 40% all the time.

    If you guys could see how these settings go for you, if everything works as it should then the above information will be added to the OP.
    30
    Configurations comparisons, battery run results.

    My recent battery run:


    5609d05b-5d8c-38ba.jpg

    5609d05b-5dfa-3a6d.jpg


    This was on:

    CheckROM V6 + N.E.A.K. 1.4x + MagicConfig v0.3.2 + V6 Supercharger

    As you can see in the screenshot I was on 50% battery at that time, so by the power of addition we can conclude that 6+ hours screen time is very much possible and isn't a myth. :D

    ----------

    Recent kernel comparison report:


    Massive thanks to It_ler for conducting the following comparisons (28-02-2012), please read the whole report carefully to avoid unnecessary questions.[/SIZE][/U][/B][/CENTER]

    Results for SpeedMod / Phenomenal / N.E.A.K. / Siyah / RedPill / stock kernels

    Everything is specific to my phone, my "normal" usage and my home location. Most values are exact results, but some are rounded. I usually have wifi, synch (three email accounts, calendar and news), Google Talk and Skype on.
    For all custom kernels, I was running on Frankenstein XWLA4 ROM (Modem: XXKL1).

    For stock kernels, I was running on the corresponding stock ROM.
    Always using the Samsung's standard / stock 1650mAh battery.
    • For Siyah, as it has many, many configurable settings, I have changed nothing - just default settings.
    • For N.E.A.K. I tested VR schedule, Lulzactive governor, only enabled sched_mc and the rest unchanged defaults.
    • For RedPill kernel, I tested completely default settings (deadline / ondemand) without any change.
    • For RedPill (MC, noop/conservative), I tested geko95gek's MagicConfig v0.3.1 settings with noop / conservative on 200-1200 MHz (no UC or OC, only recommended/safe UV).
    • For RedPill (MC, dealine/sleepy), I tested geko95gek's MagicConfig v0.3.1 settings with deadline / sleepy on 200-1200 MHz (no UC or OC, only recommended/safe UV).
    • For RedPill (sio/conservative) kernel, I tested completely default settings, only changed io sched to sio and governor to conservative.
    • Just tested RedPill rev1.3 with different io sched / governor combinations (and geko95gek's MagicConfig), and added its results.

    Here are my current results:

    Benchmarks (in Performance governor if available)
    Quadrant Standard (3rd run)

    • Phenomenal 4.0: 4780
    • SpeedMod K2-21: 4771
    • N.E.A.K. v1.3.3: 4616
    • Siyah v2.6.12: 4345
    • Stock XILA2 kernel: 4259
    • Stock XWLA4 kernel: 4076
    • Stock XWKL1 kernel: 4036
    • Stock XWKK5 kernel: 3899
    • Stock XWKK2 kernel: 3852

    Quadrant Standard 2.0 (3rd run)

    RedPill v1.3: 3282


    NenaMark2 (best of 3 runs)

    • Stock XILA2 kernel: 54,3
    • N.E.A.K. v1.3.3: 54,2
    • Stock XWLA4 kernel: 54,2
    • Siyah v2.6.12: 54,1
    • SpeedMod K2-21: 53,5
    • RedPill v1.3: 52,7
    • Stock XWKK5 kernel: 47,8
    • Stock XWKK2 kernel: 47,6
    • Phenomenal 4.0: 44,4
    • Stock XWKL1 kernel: 38,1
    • Phoenix v1.27:

    Vellamo (best of 2 runs)

    • Stock XILA2 kernel: 1225
    • SpeedMod K2-21: 1224
    • Siyah v2.6.12: 1221
    • Stock XWLA4 kernel: 1216
    • Phenomenal 4.0: 1214
    • N.E.A.K. v1.3.3: 1213
    • Stock XWKL1 kernel: 1173
    • RedPill v1.3: 1172
    • Stock XWKK5 kernel: 1160
    • Stock XWKK2 kernel: 1080


    Linpack (best of 10 runs): Single / Multi thread

    • Phenomenal 4.0: 57.342 / 102.346
    • SpeedMod K2-21: 57.224 / 98.062
    • Siyah v2.6.12: 57.816 / 95.725
    • N.E.A.K. v1.3.3: 57.499 / 95.454
    • Stock XWLA4 kernel: 57.816 / 93.548
    • Stock XILA2 kernel: 57.302 / 91.667
    • RedPill v1.3: 57.499 / 80.663


    CF-Bench (best of 2 runs): Native / Java / Overall score

    • N.E.A.K. v1.3.3: 11530 / 3623 / 6785
    • Stock XWLA4 kernel: 11396 / 3539 / 6681
    • Phenomenal 4.0: 11522 / 3409 / 6654
    • SpeedMod K2-21: 11428 / 3355 / 6584
    • Siyah v2.6.12: 11413 / 3313 / 6553
    • RedPill v1.3: 11271 / 3229 / 6445
    • Stock XILA2 kernel: 11285 / 3187 / 6426


    AnTuTu (1st run)

    • N.E.A.K. v1.3.3: 6149
    • Stock XILA2 kernel: 6110
    • Siyah v2.6.12: 6072
    • Stock XWLA4 kernel: 6009
    • SpeedMod K2-21: 4986
    • Phenomenal 4.0: 4966
    • RedPill v1.3: 4832

    Battery stats
    Night use (wifi, synch, Talk and Skype on)

    • RedPill v1.3 (sio/conservative): Battery loss of 0,72% per hour, Awake: 4,8%, Deep sleep: 93%
    • Phenomenal 4.0 Extreme: Battery loss of 0,82% per hour, Awake: 5,6%, Deep sleep: 93%
    • SpeedMod K2-21: Battery loss of 0,86% per hour, Awake: 6,0%, Deep sleep: 93%
    • RedPill v1.3 (MagicConfig, deadline/sleepy): Battery loss of 0,94% per hour, Awake: 6,5%, Deep sleep: 92%
    • XWLA4: Battery loss of 0,95% per hour, Awake: 9,2%, Deep sleep: 90%
    • RedPill v1.3: Battery loss of 1,00% per hour, Awake: 6,3%, Deep sleep: 92%
    • XILA2: Battery loss of 1,09% per hour, Awake: 6,5%, Deep sleep: 91%
    • XWKK5: Battery loss of 1,09% per hour, Awake: 11,6%, Deep sleep: 88%
    • RedPill v1.3 (MagicConfig, noop/conservative): Battery loss of 1,25% per hour, Awake: 8,8%, Deep sleep: 89%
    • N.E.A.K. v1.3.3: Battery loss of 1,28% per hour, Awake: 9,2%, Deep sleep: 90%
    • XWKK2: Battery loss of 1,44% per hour, Awake: 9,9%, Deep sleep: 89%
    • Siyah v2.6.12: Battery loss of 1,52% per hour, Awake: 9,6%, Deep sleep: 89%

    Mixed (day and night) usage (*)

    • RedPill v1.3 (sio/conservative): Battery loss of 1,76% per hour, Wifi: 100%, Awake: 15,8%, Deep sleep: 84%, Screen: 9,7% => loss*(1-screen): 1,59%
    • SpeedMod K2-21: Battery loss of 1,91% per hour, Wifi: 100%, Awake: 16,7%, Deep sleep: 83%, Screen: 10,2% => loss*(1-screen): 1,72%
    • RedPill v1.3: Battery loss of 2,02% per hour, Wifi: 100%, Awake: 16,5%, Deep sleep: 83%, Screen: 10,3% => loss*(1-screen): 1,81%
    • RedPill v1.3 (MagicConfig, noop/conservative): Battery loss of 2,05% per hour, Wifi: 100%, Awake: 18,8%, Deep sleep: 81%, Screen: 10,5% => loss*(1-screen): 1,83%
    • Phenomenal 4.0 Extreme: Battery loss of 2,05% per hour, Wifi: 100%, Awake: 18,6%, Deep sleep: 81%, Screen: 9,8% => loss*(1-screen): 1,85%
    • RedPill v1.3 (MagicConfig, deadline/sleepy): Battery loss of 2,15% per hour, Wifi: 100%, Awake: 20,5%, Deep sleep: 79%, Screen: 10,0% => loss*(1-screen): 1,94%
    • XWLA4: Battery loss of 2,19% per hour, Wifi: 100%, Awake: 19,6%, Deep sleep: 80%, Screen: 10,2% => loss*(1-screen): 1,97%
    • XILA2: Battery loss of 2,42% per hour, Wifi: 100%, Awake: 18,2%, Deep sleep: 81%, Screen: 9,0% => loss*(1-screen): 2,20%
    • XWKK2: Battery loss of 2,37% per hour, Wifi: 81,9%, Awake: 17,4%, Deep sleep: 82%, Screen: 5,9% => loss*(1-screen): 2,23%
    • XWKK5: Battery loss of 2,58% per hour, Wifi: 96,1%, Awake: 19,5%, Deep sleep: 80%, Screen: 9,0% => loss*(1-screen): 2,35%
    • Siyah v2.6.12: Battery loss of 2,94% per hour, Wifi: 100%, Awake: 23,3%, Deep sleep: 76%, Screen: 10,5% => loss*(1-screen): 2,63%
    • N.E.A.K. v1.3.3: Battery loss of 3,05% per hour, Wifi: 100%, Awake: 24,3%, Deep sleep: 75%, Screen: 11,4% => loss*(1-screen): 2,70%

    (*) In order to get a value, which allows to compare test results with different screen on times for mixed usage battery tests, I tried to „normalized“ the battery loss per hour with the following formula:
    loss*(1-screen): battery loss in % (per hours) * (100% - screen on time in %)
    So, it would be the battery loss in % (per hour) for the screen off time.
    That is how a kernel with the same battery loss per hour will get a better/lower value, if it has more screen on time. (And the other way round.)
    RedPill rev1.3 has got best battery savings with sio/conservative on my phone, even without any additional UC/UC, etc.
    Subjectively, it feels a little bit laggy/not as smooth as in other combinations (e.g. sio/hyper).

    So, my first impression of this kernel was proven: Very stable kernel which is highly configurable to fit every demand - for performance and battery life.

    Thanks again pongster for that great kernel.
    And thanks to geko95gek for his MagicConfig which gave me enough confidence to test out some UV, io sched and governor changes.

    Looking forward to test new versions of that kernel ...

    ----------

    It_ler has finally opened his own thread with more up to date results... you can follow it here - xdaforums.com/showthread.php?t=1682765

    5
    Configuration user feedback.

    Just some feedback received from users of MagicConfig:

    From boromxed

    "Hi,

    This is to inform you that your settings are excellent.
    Have tested them on Siyah 2.6.7 /2.6.8 and 2.6.9 with Omega B&B 10.2 rom and, it works wonderfull on all.
    Very smooth and very little battery drain.
    For information battery drain during night time over an 8 hours period (Wifi/Data/Background Data on off) is less than 2%.
    I even got 0% during one night (battery level at 35% at 12 pm and still at 35% at 08:00 am);

    Congratulations, mate"

    ---

    From itsedstech

    "So i tested out your config, and oh my do i get amazing battery life! whenever i go to school id have it drain down to about 92% after my first class which was roughly 1 hour and 35 minutes, and thats without touching it, this time it only went down to 98% and that was with music playing too some of the time! Thanks man! i read about your 0.3.1 config, whens that coming out? lol"

    ---

    From blk15

    "Hey Geko,

    Used your settings back in 2.5.2 and battery life was AMAZING.

    Wondering whether you have v3 settings for 2.6.1 out

    Cheers"

    Great to see you've created a thread on your configs geko!!!

    Been using your configs since around v1 and it has always provided me with the best battery life & performance. Keep up the good work :)

    ---

    From jermitano
    jermitano said:
    ...btw im using ur config right now on 2.6.9...and with the kernel settling in im experiencing phenomenal batt life...2 hrs watching mkv film/screen on 6hrs/3g-h+ 8hrs/20 mins call/6hrs wifi/download game data (almost 1gb)...18hrs and still have 40% juice on it...

    ---

    From maccawire
    More feedback for you geko. Been using your magic config for about a week now and my phone has been as smooth as a ken doll.
    I use bfg and ondemand 200Mhz - 1000Mhz plus all your voltages with the exception 200Mhz I use 875mv as it causes me no issues.
    Getting around 48hrs uptime (from 24hrs - 32 hrs) which means the battery saving is great (and if I cared about the environment greener also)

    Thanks again.

    ---

    From tonysac
    Since yesterday i'm using last siyah with settings from first post on noop-conservative combo with only difference that I set max to 1200..indeed those values make a huge difference on battery drain and also phone is stable like a rock!(omega 10.2 ROM)
    Thanks a lot :thumbup:

    ---

    From dotter2000
    Just would like to say thanx and i found this thread most useful, i was using siyah kernels before i went over to trying ics roms and i got amazing battery life out of my phone from following thid thread...

    ---

    From ole12
    Dude.

    3 days on phone, and 5hour display time is sick.
    Thank you so much. :-D

    Sent from my GT-I9100 using xda premium

    ---

    From tetrakinitis
    Geko thanks for the settings !! Everything is perfect + smooth !!! CheckROM V4 + Siyah Kernel 2.6.7 + Lionheart Governor (deadline+tweaked conservative).......

    Sent from my SmoothCriminal using xda premium

    ---

    From aceofclubs
    Did exactly what you suggested with siyah and voltage control.

    Made a huge difference. About 1.5hrs more screen on time.

    Previously used cfroot KK5 and setcpu to UV/UC and by 4.5 to 5 hours I'd be looking for a charger.

    Thanks a million for this. I have smoother, better performance all round and what is a record for battery performance for me.

    I'll be sticking to this setup for sometime now. Unless neak really is better.

    724938d6-611d-0f18.jpg
    724938d6-6137-6218.jpg

    ---

    From bade90
    All is working good now, unbelievable battery life. Many thanks geko, you have earned my "thanks" and a mention in my signature :D

    ---

    From aceofclubs
    Top thread gek. Best starting point for UV I've come across.

    Thanks for spending months of your time nutting out a safe and thoroughly tested point for us to start at.

    From your research it's easy as to move up or down to find the best settings and governors for our kernel and specific device.

    I've found my personal favourite setup and finally can stop chasing performance and battery.

    That's it for me with GB. Nothing left to do. If only it came out of the box like this.

    Sent from my GT-I9100 using XDA App


    ----------

    More to come....
    5
    Hi everyone. I just want to share this script I created. It has both andreilux's gpu settings and magicconfig by geko. Though the undervolt in the script is higher so it is a bit more aggressive and if it runs fine on your phone. It will give you the best battery life on Siyah 2.6.11. Goodluck

    Just flash the zip like usual.

    Some Notes:
    It is recommended that you remove boot settings especially if you have the start up service boot by voltage control. Though this script will execute later than boot vc if you have it. It is still better that you just remove any boot settings done on vc.