Page 19 of 56 FirstFirst ... 9141516171819202122232429 ... LastLast
Results 361 to 380 of 1118
  1. xalasten's Avatar
    Posts
    26 Posts
    Global Posts
    82 Global Posts
    #361  
    Hello,

    Just a follow up to my post yesterday about the battery drain when using the screenstate setting at 500 and 800. Yesterday I was dropping like a stone. about 9% per hour.

    I switched to "performance" 800 all the time and I am dropping at about 3% per hour. I have not been running the battery monitor as I assume it causes you to drop faster by polling all the time.

    Just wanted to give my experiences.

    Take Care
  2. Leke's Avatar
    Posts
    597 Posts
    Global Posts
    603 Global Posts
    #362  
    I followed the directions on getting uber-kernal i see the icon I launch it and it sits there illuminating. Govna works fine just not uber-kernal
  3. ltlruss's Avatar
    Posts
    180 Posts
    Global Posts
    181 Global Posts
    #363  
    Quote Originally Posted by Leke View Post
    I followed the directions on getting uber-kernal i see the icon I launch it and it sits there illuminating. Govna works fine just not uber-kernal
    You don't need to launch the uber-kernal icon... It's just there so you know that you have it installed.
  4. Leke's Avatar
    Posts
    597 Posts
    Global Posts
    603 Global Posts
    #364  
    Quote Originally Posted by ltlruss View Post
    You don't need to launch the uber-kernal icon... It's just there so you know that you have it installed.
    But my pre is even slower than before now, it never speeds up to 800, even though I set the parameters in gvna
  5. #365  
    New report here:

    I've seen something weird, I've using this kernel for 2 days, in which my Pre processor was between 40-42C almost everytime. It took the battery down fairly quickly. This was using screenstate governor, noop scheduler.

    Yesterday I got sick of it and changed the governor to performance, and stick the min and max speed to 800Mhz. It's running cooler and the battery seems to last a bit longer.

    Could this be related to processes running in the background longer when at 500Mhz than when at 800Mhz? I have AP Mobile, Twee, Weather Dashboard and 2 email accounts, everything configured to connect every 5 minutes (email & twee) and every hour (AP Mobile and Weather). The programs configuration hasn't changed, and now I'm running, at 800Mhz at every moment, and the temp is between 25-30C

    Any ideas????

    Another thing, the load average is insane here, I've had peaks of 15 or more during even minutes!

    We need a way to really monitor what process is hogging the processor (it's LunaSysMgr, I know, but inside it what javascript subroutine is killing it? I don't know).
  6. #366  
    Quote Originally Posted by rwhitby View Post
    So "grep LunaSysMgr /md5sums" will show you what the md5sum should be, and "md5sum /usr/bin/LunaSysMgr" will check what it is.

    -- Rod
    Uber Thanks Rod! I was able to check and see that both of the files you mentioned do match my Doctor to 1.4.1.1

    I would state that the uninstall for the SPK app and its kernel does in fact revert back to the proper files, but inevitably someone may not have started with a freshly doctored Pre or the latest version of SPK and my declaration would be circumstancial at best. Suffice it to say that it worked for me the way it should have.

    I think I am going to run at the default 500 all day and get a rough benchmark based on my use pattern to have a comparison for tomorrow and I will run at a different setting.

    I never monitored battery temp or consumption prior to overclocking, so a benchmark would be good to establish I suppose.

    The ultimate intent of my questions, summary, and logical argument in the previous posts is to help those who feel there are differences in the performance/stability of the two dominant kernel solutions, to have a valid point of technical reference to go and do research to see what might be making the difference (if there is any). It is better for everyone when we are working from good data.

    Thanks again. (the cookies were good too)
  7. #367  
    Quote Originally Posted by NuttyBunny View Post
    New report here:

    I've seen something weird, I've using this kernel for 2 days, in which my Pre processor was between 40-42C almost everytime. It took the battery down fairly quickly. This was using screenstate governor, noop scheduler.

    Yesterday I got sick of it and changed the governor to performance, and stick the min and max speed to 800Mhz. It's running cooler and the battery seems to last a bit longer.

    Could this be related to processes running in the background longer when at 500Mhz than when at 800Mhz? I have AP Mobile, Twee, Weather Dashboard and 2 email accounts, everything configured to connect every 5 minutes (email & twee) and every hour (AP Mobile and Weather). The programs configuration hasn't changed, and now I'm running, at 800Mhz at every moment, and the temp is between 25-30C
    If you have a lot of processes running in the background (ie, still running with the screen off) you may well see better battery life due to the "race to idle" affect. I'm a little surprised at the temperature difference you see, I wouldn't have expected it to be that high at 500. Do you see those temperatures with the stock Palm kernel?

    Quote Originally Posted by NuttyBunny View Post
    Another thing, the load average is insane here, I've had peaks of 15 or more during even minutes!
    You may see loads that high when the screen is off and the system is running on battery. I saw powerd sending frequent suspend/resume wall messages that would hint that something is failing. I put a Wait For Interrupt instruction in the fail condition of the power suspend code which seems to have had a beneficial effect on powerd (see git.webos-internals.org Git - kernels/patches.git/commit for the patch itself; I put it my last bobafett kernel).

    Quote Originally Posted by NuttyBunny View Post
    We need a way to really monitor what process is hogging the processor (it's LunaSysMgr, I know, but inside it what javascript subroutine is killing it? I don't know).
    Not necessarily. Run top in a novaterm window and watch the load and which process is using the most CPU. I have seen loads of 30+ on my pre when nothing is using any CPU time of any note. You might want to ipkg-opt install powertop as well and run that to see what's actually waking up the processor.

    Cheers, Steve
  8. #368  
    If there is a notification such as an email that has not been swiped away yet will that cause the processor to run at 500 constantly when the screen is off as apposed to being idle? I do have the no blink notification patch so that is not running.
  9. thornev's Avatar
    Posts
    326 Posts
    Global Posts
    329 Global Posts
    #369  
    I just learned something. When using CPUScalerUltimate, if you want scaling to occur, do not select a speed option (via Set Speed Options). I thought both a speed option AND a scaling option (via Scaling Options) were required. But I think what I learned is that selecting a speed option overrides any previously-selected scaling option because my battery life wore down significantly more than yesterday (when I did not have a speed option selected) and there is much more "time in state" in my Max Freq than yesterday. thorne
  10. #370  
    I am a bit confused as to what to do with Govnah. Do I just switch between the different governors for different performence? Can anyone give me a little hint as to what to do?

    Thanks
    There is no spoon
  11. #371  
    an update:
    There is a quick patch for some LCD scaling bugs. LCD brightness still needs to be above a certain number but shouldnt see CPU speed inversion... please test and let me know.

    i'm still working on a real touchstone fix.
    Last edited by unixpsycho; 05/05/2010 at 01:58 PM.
    Live free or DIE!
  12. #372  
    Quote Originally Posted by sbromwich View Post
    If you have a lot of processes running in the background (ie, still running with the screen off) you may well see better battery life due to the "race to idle" affect. I'm a little surprised at the temperature difference you see, I wouldn't have expected it to be that high at 500. Do you see those temperatures with the stock Palm kernel?
    Well, this happens just after bootup, or 1 hour later, or 5 hours later, even if I don't run anything after bootup, it's really weird. No, it doesn't do it with the stock Kernel and no, it didn't do it with your deathstar kernel. And it didn't do it with unixpsycho's kernel (com.unixpsycho.kernel_1.4.1_armv7.ipk) which I didn't left because it didn't have the cpu temp code in

    Quote Originally Posted by sbromwich View Post
    You may see loads that high when the screen is off and the system is running on battery. I saw powerd sending frequent suspend/resume wall messages that would hint that something is failing. I put a Wait For Interrupt instruction in the fail condition of the power suspend code which seems to have had a beneficial effect on powerd (see git.webos-internals.org Git - kernels/patches.git/commit for the patch itself; I put it my last bobafett kernel).
    I may try your kernel to see how it fares against this

    Quote Originally Posted by sbromwich View Post
    Not necessarily. Run top in a novaterm window and watch the load and which process is using the most CPU. I have seen loads of 30+ on my pre when nothing is using any CPU time of any note. You might want to ipkg-opt install powertop as well and run that to see what's actually waking up the processor.
    I'll try powertop. The load can be high when there's a lot of io operations going on, it may not affect cpu, but makes the system sluggish as well, I really don't expect to find anything with powertop tho
  13. #373  
    If you're having problems with uber/govnah take batt out,and it might solve some issues.
  14. #374  
    Quote Originally Posted by sbromwich View Post
    You may see loads that high when the screen is off and the system is running on battery. I saw powerd sending frequent suspend/resume wall messages that would hint that something is failing. I put a Wait For Interrupt instruction in the fail condition of the power suspend code which seems to have had a beneficial effect on powerd.
    That's what was causing high loads for me. Your Wait For Interrupt fix could have benefits beyond powerd (see below).

    Quote Originally Posted by sbromwich View Post
    Not necessarily. Run top in a novaterm window and watch the load and which process is using the most CPU. I have seen loads of 30+ on my pre when nothing is using any CPU time of any note. You might want to ipkg-opt install powertop as well and run that to see what's actually waking up the processor.
    Same here again. For the time being, a simple workaround is to always run 'Nodoze' which silences the powerd activity. Probably not the most efficient, but it returns the load avg's to normal. And crond is happy again.

    This also seemed to be the trigger to the 'screenstate / inverse max-min' issue, where the CPU would jump to max when the screen went off. Nodoze running (or on USB charger): screenstate happy. On battery (w/o Nodoze): screenstate inverted.

    Quote Originally Posted by unixpsycho View Post
    an update:
    There is a quick patch for some LCD scaling bugs. LCD brightness still needs to be above a certain number but shouldnt see CPU speed inversion... please test and let me know.

    i'm still working on a real touchstone fix.
    Marco:
    It's looking good. With new UberK (1.4.1-16) and 'Nodoze' *not* running (so that the powerd msgs return), screenstate is behaving correctly. Thanx!!
  15. #375  
    Looking forward to the touchstone fix. Once that happens I am trying uber and govnah
    *Patchers STILL make this phone Perfect!
    Palm Devices Currently Owned: Pre, Pre Plus, Pre 2, Pre 3, Pixi, Veer, Touchpads, and too many touchstones!
  16. #376  
    I was having the touchstone problem with the screenstate setting sticking at 500mhz after pulling off the touchstone. After the update it jumps to 800 like its supposed to. So for me this was a touchstone fix, well done!
  17. #377  
    Quote Originally Posted by Blipstein View Post
    I am a bit confused as to what to do with Govnah. Do I just switch between the different governors for different performence? Can anyone give me a little hint as to what to do?

    Thanks
    Thats exactly what you do! Its the beauty of this setup, you can mess around with the settings until you find something that fits your usage needs. Play around with the settings until you find one that gives you what you are looking for whether it be performance, battery life, or a combination of the 2.
  18. kalex's Avatar
    Posts
    253 Posts
    Global Posts
    728 Global Posts
    #378  
    Rod - reading this thread, you seemed to have nailed the reason why i was having one result with SPK and one result with Uber. it looks like LunaSysMgr is not what it suppose to be, which means that SPK didn't uninstall properly. Do you happen to know if there is a package that restores this to original condition like you have a package for the original Pre Kernel?

    here is what i get from md5 stuff:
    Code:
    root@palm-webos-device: grep LunaSysMgr /md5sums
      ce8ab50242283ea66592d5380041de2d *./usr/bin/LunaSysMgr
      194acce643388c8778d689fb259e8752 *./etc/event.d/LunaSysMgr
    root@palm-webos-device: md5sum /usr/bin/LunaSysMgr
      ca6a3981c382b01a559d7186742f1fe5  /usr/bin/LunaSysMgr
    root@palm-webos-device: md5sum /etc/event.d/LunaSysMgr
      cbc4e1c8da94d7c106dd3356d5455714  /etc/event.d/LunaSysMgr
  19. PreGk's Avatar
    Posts
    441 Posts
    Global Posts
    631 Global Posts
    #379  
    Having big problems with Uber Kernel and the camera app. Anything over 600Mhz causes my camera to lock up,stutter, etc. Any fix for this? It didn't happen on the old 720Mhz.sh when I was using that.
  20.    #380  
    Quote Originally Posted by kalex View Post
    Rod - reading this thread, you seemed to have nailed the reason why i was having one result with SPK and one result with Uber. it looks like LunaSysMgr is not what it suppose to be, which means that SPK didn't uninstall properly. Do you happen to know if there is a package that restores this to original condition like you have a package for the original Pre Kernel?
    Unfortunately, because LunaSysMgr is a proprietary Palm binary, as opposed to the GPL kernel, we cannot legally distribute that file - so we cannot create a recovery package for it, nor can we include it in the existing recovery package.

    You really need to take this incorrect uninstallation issue up with the maintainer of the SPK. This is yet another objective technical issue that I strongly campaigned about over a month ago both in private and in public, but was unfortunately ignored, and now you suffer the consequences.

    -- 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

Posting Permissions