Fuduntu 2013.1 battery life - What gives?

Discussion in 'all things UNIX' started by fewt, Apr 5, 2013.

Thread Status:
Not open for further replies.
  1. fewt

    fewt Registered Member

    Joined:
    May 5, 2011
    Posts:
    14
    As I mentioned I read the review and I had a lot of questions. I forwarded the link, and we decided to talk about it here. Probably a really good place so we can capture information about the system and how we manage / test / improve battery life at Fuduntu.

    I dug into the specs of the machine a bit and found that the system ships with a 6 cell battery. I'm curious about the state of the battery itself, we can get some info from:

    Code:
    cat /sys/class/power_supply/BAT0/uevent
    I would also like to see if ktune is aware of all of the devices in your laptop, so if you wouldn't mind install powertop and run a report, send that to me at andrew- -at- -fuduntu- -dot- -org

    Code:
    sudo yum -y install powertop
    (unplug)
    sudo powertop --html=dv6-2130ej.html
    I noticed in your review that you tested using gnome-power-manager and by leaving the system idle for 5 minutes. I was wondering if you rebooted and tested from a fresh boot as processes running in the background can skew.

    Also, gnome-power-manager isn't the most accurate method to test. Using powertop can glean far more accurate results.

    If you would, retest from a fresh reboot (for each platform) and to test:

    Code:
    (unplug)
    sudo powertop
    Allow it to run for 15 minutes uninterrupted. That will give you a baseline measurement of the system idle power usage.

    After that, charge it back up and re-run the test unplugged doing the things that you would normally do. Capture the power use during those functions after 15 minutes of normal use.

    Measuring idle is a good indicator of the baseline, but it is more important to measure the baseline plus usage and then working through any processes that are causing tons of wakeups or racing the CPU. You can compare the output in use to your baseline to figure out what the potential gain could be as you remove IOs and CPU race conditions.

    At Fuduntu, we went through great lengths to reduce disk IOs allowing platters to sleep longer. We've also tuned the kernel to have a less aggressive swap policy, and built a tool called ktune to reconfigure devices and the kernel to go into low power or power friendly states when unplugged.

    You may not see much of a gain when leaving it idle for 5 minutes, but you may when using it over a period of time while unplugged.

    In addition you have an nVidia adapter, I don't know how functional power management is in the proprietary driver so that may be draining. Is that laptop optimus capable? We do support optimus.

    Finally, if you aren't getting the best battery life and your battery isn't charging 100% you may need to deep cycle it. It's not very common anymore, but I've spoken with a few battery manufacturers who recommend it in cases where it isn't charging completely.

    If that's the case you can deep cycle by killing the gnome-power-manager process, unplugging, and burning through the battery allowing it to drain completely until it will no longer power on.

    Then charge it for 14 hours (no less), and repeat two more times.

    In a few cases we find that you can't improve battery life any, but I think in this case there is a great chance that we can. :)

    Feel free to reach out to us when you find problems like this, we would be happy to look into them with you and try to resolve them. We are nearly always available in #fuduntu on freenode, at Fuduntu Forum, or by email at team- -at- -fuduntu- -dot- -org.
     
    Last edited: Apr 5, 2013
  2. Mrkvonic

    Mrkvonic Linux Systems Expert

    Joined:
    May 9, 2005
    Posts:
    8,693
    I'll definitely do this, can take a bit time.
    It's also an interesting exercise in power measurement.
    Mrk
     
  3. fewt

    fewt Registered Member

    Joined:
    May 5, 2011
    Posts:
    14
    Awesome, no rush. :cool:
     
Loading...
Thread Status:
Not open for further replies.