[KERNEL] [2012-01-12] V2.0 FM Kernel based on 2.6.35.14 [Beta]

Search This thread

squicky

Senior Member
Mar 11, 2012
661
200
@zylor

How did you tweaked it to achieve such a long screen on time? What are your lulzactive settings, min and max cpu rate? Is the 200Mhz really better than 100Mhz for battery life in our SGN?
 

Top Liked Posts

  • There are no posts matching your filters.
  • 63
    Hello,

    I want to publish the first public beta of my Kernel. It is based on 2.6.35.14 and should reduce the Android OS Battery usage. The Kernel is based on CF Root Kernel.

    As always: You are responsible for your phone. Don't blame me, if something goes wrong and you loose the photos from grandmas birthday or brick your phone. I have warned you! The Kernel is in very early stage and has been tested only by me so far.

    Features:
    • 2.6.35.14
    • Root
    • CWM5
    • Compatible to Chainfires CWM App
    • Reduced Debugging and Logging
    • Interactive Governor enabled (but not set as default. Use SetCPU).

    NEW Version 1.2:
    - Based on KJ6 Sources
    - optimized compiler flags
    - bug fixing

    NEW Version 1.3:
    - OC/UV Support (100 - 1700 MHz, SetCPU and Voltage Control)
    - additional governors (ondemand, ondemandx, interactive, interactivex, smartass, smartassv2, performance, powersave, conservative, Lagfree, Lazy, Savagedzen, Lulzactive)
    - Tegrak 2nd core Support
    - I/O schedulers: BFQv3, SIO, CFQ, deadline, VR, noop
    - some tweaks and bugfixes

    NEW Version 1.3.1:
    - Removed some Optimizations in order to fix the Camera Bug

    NEW Version 1.4:
    - added 50 MHz
    - Fixed Bug which caused removal of CWM Manager
    - EFS Backup reworked
    - intellidemand govenor added
    - added Version Number in Version String
    - Fixed Camera Bug finally
    - Deactivated GPU Undervolting
    - Some tweaks added again, which were removed for 1.3.1

    NEW Version 1.5:
    - Optimizations in Schedulers and Governors
    - Maybe Bluetooth Streaming Bugfix

    NEW Version 1.6:
    - Voodoo Louder support
    - Optimized schedulers
    - new MALI drivers
    - more logging and debugging (due to voodoo louder)

    NEW Version 2.0:
    - complete reworked cpufreq (incl. Abyss Smooth Scaling support), no more 50 MHz support
    - Lulzactive V2 inkl. Tegrak support
    - xz compression
    - optimized configuration
    - tun, ntfs, fuse support
    - amazing performance improvement

    The configuration is more or less Samsung Stock (except debugging, logging and the governor). Maybe I will optimize it in the future. The CWM App from Chainfire is not included, but should work on this kernel. I suggest to first flash a CF Kernel in order to root your phone and get the CWM App. After that you can flash this kernel. I recommend the CWM Manager App for flashing activities.

    Please report bugs. I had a problem with endless USB Port Suspend / Resume cycles during testing, which caused the phone not to enter the deep sleep. Maybe the reason were to optimistic compiler options (it did not occur in the last 24 hours).

    I would also appreciate, if you compare battery lifetime with Stock kernel. The battery statistic tells, the Android OS Usage is reduced, but results this really in a longer battery lifetime? Please observe and report.


    For the Developers: The Sourcecode.
    https://github.com/riversource/Galaxy-Note-Kernel
    6
    V2.0 online in first post.
    4
    Hello,

    I want to publish the first public beta of my Kernel. It is based on 2.6.35.14 and should reduce the Android OS Battery usage.
    It does not reduce Android OS battery usage at all - 2.6.35.12 only hides it.

    http://git.kernel.org/?p=linux/kern...it;h=a3fe22ee824895aafdc1b788e19c081a2e6dd9da is the one that makes Android OS battery usage disappear from Settings->About->Battery Usage (How do I know? git bisect is your friend!)

    As you can see, that is only an accounting/reporting patch - it doesn't change the amount of time spent in interrupt handlers during suspend/resume at all, nor does it affect the amount of battery used during suspend/resume.

    Best example, one can cause frequent resume/suspend cycles which shoot AOS through the roof as follows from a Linux machine:
    Code:
    ping -i 5 <wifi ip address of the device>
    You will see that whether you are on .7 or .14, battery drain will be the same, but the percentage of AOS usage reported will be MUCH lower on 2.6.35.12 and above.

    Edit: Also, as far as implementing 100 MHz and 50 MHz, I strongly suggest you read section 6 of the paper on idle states (clock/power gating) that Ezekeel links here - http://xdaforums.com/showthread.php?t=1431105&page=9
    4
    V1.4 is online in first post.
    4
    Read this sticky for very good explanations on most if not all governors and schedulers.

    http://xdaforums.com/showthread.php?t=1369817