Page 30 of 53 FirstFirst ... 20252627282930313233343540 ... LastLast
Results 581 to 600 of 1045
  1. staeit's Avatar
    Posts
    140 Posts
    Global Posts
    144 Global Posts
    #581  
    unixpsycho: You are a genius. I love you.
  2. bjs188's Avatar
    Posts
    188 Posts
    Global Posts
    238 Global Posts
    #582  
    Quote Originally Posted by Xanadu73 View Post

    The battery life isn't *as* minimal as it was with The Kernel That Shall Not Be Mentioned.
    that's too funny. After too much drain from UK/gov, I switched back TKTSNBM. Love the ease of ipk. Looks like I'm gonna try F102 today.

    Thanks Unixpsycho
  3. #583  
    Quote Originally Posted by Xanadu73 View Post
    Personally when the day comes that all these kernels *require* Govnah to keep them at 500/800, is the day I start figuring out how to do all this myself, because I have found that having them "stock locked" at 500/800 gives the best performance and least battery drain. Telling the kernel to do it isn't the same. I don't know why that is, and it makes no sense that I can duplicate it over and over (I had a short thread on that I think you saw).
    Can you point to that thread please? Since it's the same code running, there must be another reason for what you're experiencing, and we'd like to understand what that reason is.

    -- Rod
    WebOS Internals and Preware Founder and Developer
    You may wish to donate by Paypal to donations @ webos-internals.org if you find our work useful.
    All donations go back into development.
    www.webos-internals.org twitter.com/webosinternals facebook.com/webosinternals
  4. #584  
    Quote Originally Posted by rwhitby View Post
    Can you point to that thread please? Since it's the same code running, there must be another reason for what you're experiencing, and we'd like to understand what that reason is.

    -- Rod
    Sorry about the delay. Here is the thread I mentioned. It probably makes one ask more questions than it answers, but it's my findings none-the-less.

    It's a bit on the old side now as it's comparing a slightly "old" UK to The Kernel That Shall Not Be Mentioned (which no longer is available). I can run tests with Psycho's latest kernels, but, the results there are as I mentioned in my post. I hover around <3% *at dead idle* (sitting in my holster for a ~couple hours).



    M.
  5. #585  
    Quote Originally Posted by Xanadu73 View Post
    Sorry about the delay. Here is the thread I mentioned. It probably makes one ask more questions than it answers, but it's my findings none-the-less.

    It's a bit on the old side now as it's comparing a slightly "old" UK to The Kernel That Shall Not Be Mentioned (which no longer is available). I can run tests with Psycho's latest kernels, but, the results there are as I mentioned in my post. I hover around <3% *at dead idle* (sitting in my holster for a ~couple hours).
    You should use the Fixed Speed 800 profile, and then close Govnah and anything else which could be trying to use the Govnah service to read the CPU temperature (e.g. Battery Monitor or any patches that do so).

    Reading the CPU temperature periodically will cause the CPU to wake up, run the govnah service, transfer data across dbus, etc - all of which will suck power.

    -- Rod
    WebOS Internals and Preware Founder and Developer
    You may wish to donate by Paypal to donations @ webos-internals.org if you find our work useful.
    All donations go back into development.
    www.webos-internals.org twitter.com/webosinternals facebook.com/webosinternals
  6. #586  
    Here are some result for testing I've done this weekend.

    My test goal was to observe battery consumption for one day of low usage (week-ends).

    Here are the common premises of my test :
    • F104 Kernel
    • 2 eMail account set to check every 30 minutes
    • Always logged in google talk
    • Screen brightness 19%, Keyboard 27%
    • Weather Icon update every 60 minutes based on GPS
    • Twee running, no notification, manually check from time to time (one of the two application cards opened all day)
    • Reading my RSS feeds (less news items on the weekend)
    • Facebook with notification check every 20 minutes
    • Battery Monitor set to update every minute (the other application card opened all day)
    • I was home for the most part. Pre setup with WIFI Off when sleeping (I have one of those router that constantly wakes up the Pre when connected)
    • Carrier : Bell Canada, Original Pre (non plus)
    • Battery is a Mugen Power 1400mAh


    Day 1 (Saturday) : Userspace Governor (Palm Default Govnah Profile)
    Day 2 (Sunday) : Performance 800 Governor

    My first finding was that the Compcache F104 kernel is quite usable at 500Mhz. While application load time is noticeably slower than running at 800Mhz, the device runs quite smoothly with this kernel... Way better than the sock kernel.

    My average drain per hour was :
    • Day 1 : 2.6%
    • Day 2 : 4.3%


    Note that this was only for 1 day each and I did not measure the total period of time the device was awake (actively used), so the numbers must be taken with a grain of salt.

    It does seem that running at 500Mhz consume less power than running at 800Mhz with the F104 kernel and my Pre. I wil try the test again next week. In the meantime, I'll run this kernel with the Userspace governor as it perform well enough (I may change my mind after a week ;-) ).
  7. #587  
    Quote Originally Posted by ukabu View Post
    My average drain per hour was :
    • Day 1 : 2.6%
    • Day 2 : 4.3%

    Notice that the consumption is 65% more on day two that on day one... And that running at 800Mhz is 60% faster than 500Mhz. So there may be a correlation here.

    And it does goes in the face (at least for this kernel - which uses the cpu more than a stock kernel to compress the cache pages) of the Run to Idle theory (which I also believe should be true) that faster clock speed should mean lower power consumption. With these, non scientific, result faster clock speed increases consumption mostly linearly.
  8. #588  
    Quote Originally Posted by ukabu View Post
    Day 1 (Saturday) : Userspace Governor (Palm Default Govnah Profile)
    Day 2 (Sunday) : Performance 800 Governor


    My average drain per hour was :
    • Day 1 : 2.6%
    • Day 2 : 4.3%
    It would be interesting to see ScreenState (500/800 - not some variant of the low speed) with UK in comparison to that. I mean, I get ~<3% at dead idle for a couple hours. In actual (moderate) use, it's more like 5-7% or so. Since I'm not currently using UK, my numbers will be meaningless against yours.


    M.
  9. #589  
    so this morning looking on govnah with the F102 installed with the screen on it woud stay at 500mhz instead of bumping up to 800mhz. Now my brightness was set at 11 and it was a dark room.

    Could the brightness be the culprit?

    I tried rebooting, but that didnt do anything. I havent had this happen before, but reading through this thread people suggest adjusting brightness.
  10. #590  
    I've been running F104 for the past couple of days. I've primarily been using the screen state governor set to 250-720. Everything has run great. I began running at 250-800 but noticed considerable drain while I was using my pre. I bumped the screen setting to 250-720 and the battery drain is much better. Drain seems to be much slower too while my pre is off running at 250. I have 1 push email account and 3 im accounts constantly running.
  11. #591  
    Quote Originally Posted by OLD_HATCH View Post
    so this morning looking on govnah with the F102 installed with the screen on it woud stay at 500mhz instead of bumping up to 800mhz. Now my brightness was set at 11 and it was a dark room.
    I woke up to the same thing this morning using F102. But a reboot fixed it for me.
    Twitter: dullgeek
  12. #592  
    Quote Originally Posted by OLD_HATCH View Post
    so this morning looking on govnah with the F102 installed with the screen on it woud stay at 500mhz instead of bumping up to 800mhz. Now my brightness was set at 11 and it was a dark room.

    Could the brightness be the culprit?

    I tried rebooting, but that didnt do anything. I havent had this happen before, but reading through this thread people suggest adjusting brightness.
    If you turn your Pre off, remove the battery, put it back and turn it on, it should rescale to 800. I've had this problem, but it's not the 20% alert as I thought earlier
  13. #593  
    hmmm

    I tried luna reset and complete reboot and it did nothing for me. If it happens again I will try adjusting brightness and pulling battery to see if that changes anything.
  14. #594  
    Quote Originally Posted by OLD_HATCH View Post
    hmmm

    I tried luna reset and complete reboot and it did nothing for me. If it happens again I will try adjusting brightness and pulling battery to see if that changes anything.
    A clean shutdown *then* pull the battery for a minute or so, I assume you mean?

    M.
  15. #595  
    yes
  16. #596  
    Quote Originally Posted by OLD_HATCH View Post
    yes


    :-) Just checkin'. Some people think "pull the battery" doesn't include the clean shutdown FIRST part...


    M.
  17. #597  
    Ive had my share of strange happenings from battery pulls. Besides without a real backup solution its pretty risky actually.
  18. #598  
    To get a sense of the power consumption of those services, you can leave the screen off and the phone idle and either turn on Airplane mode or turn off WiFi and at least disable data. I suspect you will find that the power consumption from when Govnah and Battery Monitor poll the Govnah service is very minimal and probably not noticeable compared with apps/services that create network traffic.

    Quote Originally Posted by rwhitby View Post
    You should use the Fixed Speed 800 profile, and then close Govnah and anything else which could be trying to use the Govnah service to read the CPU temperature (e.g. Battery Monitor or any patches that do so).

    Reading the CPU temperature periodically will cause the CPU to wake up, run the govnah service, transfer data across dbus, etc - all of which will suck power.

    -- Rod
  19. #599  
    Quote Originally Posted by meandmypre View Post
    To get a sense of the power consumption of those services, you can leave the screen off and the phone idle and either turn on Airplane mode or turn off WiFi and at least disable data. I suspect you will find that the power consumption from when Govnah and Battery Monitor poll the Govnah service is very minimal and probably not noticeable compared with apps/services that create network traffic.
    It would be nice to be able to track what uses what and how much. "netstat -t" is of only little help as it's not much of a "histogram". It's really only a "nowogram"


    M.
  20. #600  
    Quote Originally Posted by Xanadu73 View Post
    It would be nice to be able to track what uses what and how much. "netstat -t" is of only little help as it's not much of a "histogram". It's really only a "nowogram"


    M.

    This is actually something that I've meant to look into for a while now (on Desktops/Servers). It requires kernel support for tracking protocol usage and mapping that to the applications using it. It's called "Layer 7 Support". It's not too fine grained. Like I don't think it can single out how much traffic specific applications are using, but it'd be a start, I guess. It requires kernel-level support, though. You game, Psycho? :-)


    M.

Posting Permissions