Page 4 of 7 FirstFirst 1234567 LastLast
Results 61 to 80 of 138
Like Tree9Likes
  1. larizzo's Avatar
    Posts
    25 Posts
    Global Posts
    32 Global Posts
    #61  
    Ok I tested, I cannot get the cpu1 to go above 0mhz I ran 3 sunspider tests at teh same time and had the load get up to 4.21. If I'm running warthhog or even your F4 with the turbo mode off it will go down to 0 and go back up to 1.728 on both cores (0 on cpu1 only obviously). BTW I'm not even letting it go to sleep I get this behavior right after boot.

    Also I do see the cpu1 go on and off depending on the load I just don't see it doing anything, also seems to be missing files for it but not sure if cpu0/cpu1 match usually...

    root@rizzopad:/# uptime
    uptime
    12:38:03 up 4 min, 0 users, load average: 4.21, 2.08, 0.88

    root@rizzopad:/# cat /sys/devices/system/cpu/cpu0/online
    cat /sys/devices/system/cpu/cpu0/online
    1
    root@rizzopad:/#
    root@rizzopad:/# ls -l /sys/devices/system/cpu/cpu1/
    ls -l /sys/devices/system/cpu/cpu1/
    drwxr-xr-x 4 root root 0 Sep 12 12:53 cpuidle
    -rw-r--r-- 1 root root 4096 Sep 12 12:52 online
    drwxr-xr-x 2 root root 0 Sep 12 12:53 sleep-stats
    drwxr-xr-x 2 root root 0 Sep 12 12:53 topology
    root@rizzopad:/# ls -la /sys/devices/system/cpu/cpu0
    ls -la /sys/devices/system/cpu/cpu0
    drwxr-xr-x 7 root root 0 Sep 12 12:33 .
    drwxr-xr-x 7 root root 0 Sep 12 12:33 ..
    drwxr-xr-x 4 root root 0 Sep 12 12:33 cpufreq
    drwxr-xr-x 5 root root 0 Sep 12 12:36 cpuidle
    -rw-r--r-- 1 root root 4096 Sep 12 12:34 online
    drwxr-xr-x 2 root root 0 Sep 12 12:33 rq-stats
    drwxr-xr-x 2 root root 0 Sep 12 12:36 sleep-stats
    drwxr-xr-x 2 root root 0 Sep 12 12:36 topology


    EDIT: Did more testing... I can see load using mpstat:
    root@rizzopad:/# mpstat -P ALL
    mpstat -P ALL
    Linux 2.6.35-palm-tenderloin (rizzopad) 09/12/11 _armv7l_
    (2 CPU)

    13:10:43 CPU %usr %nice %sys %iowait %irq %soft %steal %guest
    %idle
    13:10:43 all 22.74 6.52 9.20 3.97 0.00 0.11 0.00 0.00 57.45
    13:10:43 0 31.50 12.17 7.68 1.27 0.00 0.12 0.00 0.00 47.26
    13:10:43 1 14.07 0.93 10.71 6.65 0.00 0.09 0.00 0.00 67.55

    so there is load on cpu1 but there's still missing files and I cannot modify the cpu_freq... Let me knwo if there's anything you want me to try.
    Last edited by larizzo; 09/12/2011 at 12:19 PM.
  2.    #62  
    Well, I cant make this happen AT ALL. I am making changes to code to trying to fix something I cant see nor do I think is broken. I even left this same buid running for almost a day and cant get it to do what you see, sleep, wake, under load, nothing.

    Doing a "/sys/devices/system/cpu/cpu0/online" etc isnt showing me much. is turbo mode on? powersave? etc? govnah profile? what's installed and running? kernel logs? anything useful?

    If I dont get an exact method to reproduce, then I'm going to assume it's some patch or other software installed causing this.
    Live free or DIE!
  3. larizzo's Avatar
    Posts
    25 Posts
    Global Posts
    32 Global Posts
    #63  
    Quote Originally Posted by unixpsycho View Post
    Well, I cant make this happen AT ALL. I am making changes to code to trying to fix something I cant see nor do I think is broken. I even left this same buid running for almost a day and cant get it to do what you see, sleep, wake, under load, nothing.

    Doing a "/sys/devices/system/cpu/cpu0/online" etc isnt showing me much. is turbo mode on? powersave? etc? govnah profile? what's installed and running? kernel logs? anything useful?

    If I dont get an exact method to reproduce, then I'm going to assume it's some patch or other software installed causing this.
    I just want to clarify something since you mentioned is "turbo mode on?". I'm testing with f15c-55 so turbo mode wouldn't be a factor correct?

    I mean it could easily be some other patch or something... I wish I would have tested with my system clean since I did a doctor on Saturday. What I don't get is I can install F4 and but turbo_mode off and I'll get exactly the desired behavior (cpu1 goes from 0-1728). If I put f15c I never see cpu1 got above 0 in govnah and I can't change any of the govnah without getting and error about the file missing. F4 I never have this issue. I'll see If I can test more tonight since I'm at work.
  4.    #64  
    Quote Originally Posted by larizzo View Post
    I just want to clarify something since you mentioned is "turbo mode on?". I'm testing with f15c-55 so turbo mode wouldn't be a factor correct?

    I mean it could easily be some other patch or something... I wish I would have tested with my system clean since I did a doctor on Saturday. What I don't get is I can install F4 and but turbo_mode off and I'll get exactly the desired behavior (cpu1 goes from 0-1728). If I put f15c I never see cpu1 got above 0 in govnah and I can't change any of the govnah without getting and error about the file missing. F4 I never have this issue. I'll see If I can test more tonight since I'm at work.
    Lets not mix F15 in here too much.

    I have a pretty clean TP, barely anything on it. F15 works as it should for days...F4 works as it should for days. Like I mentioned before, I am not seeing these problems. I literally boot the TP, go into Govnah, and look at the graph. Let it go to sleep, wake it up 5-10 mins later, everything still works, run sunspider, both CPUs active. Repeat 5 times during the day. All stil fine. So I dont know what to say.
    Live free or DIE!
  5. larizzo's Avatar
    Posts
    25 Posts
    Global Posts
    32 Global Posts
    #65  
    Quote Originally Posted by unixpsycho View Post
    Lets not mix F15 in here too much.

    I have a pretty clean TP, barely anything on it. F15 works as it should for days...F4 works as it should for days. Like I mentioned before, I am not seeing these problems. I literally boot the TP, go into Govnah, and look at the graph. Let it go to sleep, wake it up 5-10 mins later, everything still works, run sunspider, both CPUs active. Repeat 5 times during the day. All stil fine. So I dont know what to say.
    Ok what I'll do is next time I doctor I'll put f15c on and that's all and I'll install on thing at a time and go from there (in the proper thread). For right now I'm using F4 and I love it works well.

    Can you explain to me what the different options are?
    ls -la /sys/devices/system/cpu/cpufreq/override
    drwxr-xr-x 2 root root 0 Sep 12 13:55 .
    drwxr-xr-x 4 root root 0 Sep 12 13:55 ..
    -rw-r--r-- 1 root root 4096 Sep 12 13:58 power_saver
    -rw-r--r-- 1 root root 4096 Sep 12 13:55 turbo_mode
    -rw-r--r-- 1 root root 4096 Sep 12 13:58 vdd
    -r--r--r-- 1 root root 4096 Sep 12 13:58 vdd_max
    -r--r--r-- 1 root root 4096 Sep 12 13:58 vdd_min

    turbo_mode disables/enables the cpu1 always on.
    power_save I believe disables cpu1 when in sleep?
    what's the other three voltage stuff?
  6.    #66  
    Quote Originally Posted by larizzo View Post
    Ok what I'll do is next time I doctor I'll put f15c on and that's all and I'll install on thing at a time and go from there (in the proper thread). For right now I'm using F4 and I love it works well.

    Can you explain to me what the different options are?
    ls -la /sys/devices/system/cpu/cpufreq/override
    drwxr-xr-x 2 root root 0 Sep 12 13:55 .
    drwxr-xr-x 4 root root 0 Sep 12 13:55 ..
    -rw-r--r-- 1 root root 4096 Sep 12 13:58 power_saver
    -rw-r--r-- 1 root root 4096 Sep 12 13:55 turbo_mode
    -rw-r--r-- 1 root root 4096 Sep 12 13:58 vdd
    -r--r--r-- 1 root root 4096 Sep 12 13:58 vdd_max
    -r--r--r-- 1 root root 4096 Sep 12 13:58 vdd_min

    turbo_mode disables/enables the cpu1 always on.
    power_save I believe disables cpu1 when in sleep?
    what's the other three voltage stuff?
    Uh? I thought you said F4 was broken?

    powersave will re-enable the default webOS powersaving profile. Disabling turbomode will essentially do the same thing. vdd is where you can set custom voltages for each freq.
    Live free or DIE!
  7. larizzo's Avatar
    Posts
    25 Posts
    Global Posts
    32 Global Posts
    #67  
    Quote Originally Posted by unixpsycho View Post
    Uh? I thought you said F4 was broken?
    hahah sorry I'm LOL! I was using f15c before then I started having this weird issues posted above so I tried f4 and it worked will for me so I use that one. I installed f15c-55 to test but still having sames issues so went back to f4. That's my story and I'm sticking to it!

    For the record f4 has always worked well, f15c has been the one that gave me all the issues above. I thought that was clear that I was testing with f15c above. Now that the f15c thread is back open I'll post in there if I do any future testing.
  8.    #68  
    Quote Originally Posted by larizzo View Post
    hahah sorry I'm LOL! I was using f15c before then I started having this weird issues posted above so I tried f4 and it worked will for me so I use that one. I installed f15c-55 to test but still having sames issues so went back to f4. That's my story and I'm sticking to it!

    For the record f4 has always worked well, f15c has been the one that gave me all the issues above. I thought that was clear that I was testing with f15c above. Now that the f15c thread is back open I'll post in there if I do any future testing.
    LOL. Maybe it's me getting threads confused.

    Ok, i'll reboot my brain and head over to the F15 thread.
    Live free or DIE!
  9. #69  
    Quote Originally Posted by unixpsycho View Post
    Let me know if -56 fixes this.
    yes, F4-v56 cpu0/cpu1 issue resolved.

    of note individual users' mileage may vary, but for me all of the F4 and F15C kernels run stable and buttery smooth, regardless of any quirkiness reported by Govnah.

    thanks.
  10. zoomy942's Avatar
    Posts
    38 Posts
    Global Posts
    490 Global Posts
    #70  
    So I did some reading.

    The f15 kernel lets you use speeds up to 1.9 and uses ondemand as it should be. Meaning it uses the first core primarily then uses the second if need be. At 1.9 the second core may not get used much. When idle, the second core is off and the first is at 192. The f4 kernel uses both cores all the time, either at 192 or whatever the top speed is.

    Im curious about battery life. The f15 one goes faster but uses one core primarily, and sits at 192 with the second core off. I set my f4 to be 1.5 max, with both cores running and 192 at idle, with both cores running.

    By that thinking, im guessing the f15 with one core off most of the time will be better battery life.

    I'm trying f4 at 1.5 for now.

    /toomuchcoffee
  11. #71  
    I just want to cautiously step in here to say that the few times I ran into issues with cpu1 seemed to have been directly related to the order of installing the new kernel and govanah, and whether or not there were residual profiles remaining (this with both the F4 kernel & f15 kernel.) I've taken to always removing all profiles from Govanah and then uninstalling it before changing kernels. My only other comment is that I've found that the battery life seems lower when using the F4, compared to the F15, running each at 1.7.

    Just 2 cents, or FWIW. Thanks for all the hard work!
  12. zoomy942's Avatar
    Posts
    38 Posts
    Global Posts
    490 Global Posts
    #72  
    Quote Originally Posted by kayphoonstar View Post
    I just want to cautiously step in here to say that the few times I ran into issues with cpu1 seemed to have been directly related to the order of installing the new kernel and govanah, and whether or not there were residual profiles remaining (this with both the F4 kernel & f15 kernel.) I've taken to always removing all profiles from Govanah and then uninstalling it before changing kernels. My only other comment is that I've found that the battery life seems lower when using the F4, compared to the F15, running each at 1.7.

    Just 2 cents, or FWIW. Thanks for all the hard work!
    its lower because the two cores are on all the time. With other kernels they only power up the second one when it needs it
  13.    #73  
    I'm looking for some feedback on the sysfs entries, like voltage tweaking, powersaver, etc.

    Are they working properly?
    Live free or DIE!
  14. #74  
    Quote Originally Posted by unixpsycho View Post
    feedback on the sysfs entries, like voltage tweaking, powersaver, etc.
    F4-v73, don't think power_saver module working properly. here's the data: power_saver test - Pastebin.com

    basically CPU % time in deep sleep = 0% during 10 minute standby, regardless of whether power_saver enabled or disabled.

    in comparison, my TP with F15C-v74 10 minute standby would produce 9 minutes of CPU deep sleep with power_saver enabled.

    the turbo_mode module works. haven't looked into voltage tweaking.

    thanks.
  15. hhy2k's Avatar
    Posts
    10 Posts
    Global Posts
    14 Global Posts
    #75  
    Using 384-1.836 profile, ondemandtcl, nothing else changed.

    Very stable, zero issues so far.

    Lithium score is worse than the previous F4. It went from 88 to 96 (tested multiple times, profile has no effect here). The device doesn't feel slower on the other hand.

    Restarting the Touchpad sets CPU to a lower speed at 1.7 GHz. I assume it is a built in fail safe.

    Thank you for all your efforts.

    (I use 384 MHz, because using 192 caused spikes in the idling TP on one of the cpus and it seemed in govnah, draw more juice from the battery than idling it at 384 all the time.)
  16.    #76  
    Quote Originally Posted by maxwintrobe View Post
    F4-v73, don't think power_saver module working properly. here's the data: power_saver test - Pastebin.com

    basically CPU % time in deep sleep = 0% during 10 minute standby, regardless of whether power_saver enabled or disabled.

    in comparison, my TP with F15C-v74 10 minute standby would produce 9 minutes of CPU deep sleep with power_saver enabled.

    the turbo_mode module works. haven't looked into voltage tweaking.

    thanks.
    I dont see any stats of sleep states in that pastebin. looking at the time_in_state means nothing when compared to actual CPU sleep modes. time_in_state does not increment when the CPU is actually sleeping/suspend.

    Enabling powersaver also turns off turbomode.

    F15 and F4 use the exact same override module. so having opposite effects is hard to believe.
    Live free or DIE!
  17.    #77  
    Quote Originally Posted by hhy2k View Post
    Using 384-1.836 profile, ondemandtcl, nothing else changed.

    Very stable, zero issues so far.

    Lithium score is worse than the previous F4. It went from 88 to 96 (tested multiple times, profile has no effect here). The device doesn't feel slower on the other hand.

    Restarting the Touchpad sets CPU to a lower speed at 1.7 GHz. I assume it is a built in fail safe.

    Thank you for all your efforts.

    (I use 384 MHz, because using 192 caused spikes in the idling TP on one of the cpus and it seemed in govnah, draw more juice from the battery than idling it at 384 all the time.)
    I noticed the lower scores as well, but I contributed that towards background jobs. But the patches were suppose to smooth out the load. I'll have to take more of a look at this...
    Live free or DIE!
  18. hhy2k's Avatar
    Posts
    10 Posts
    Global Posts
    14 Global Posts
    #78  
    I found a serious glitch. Text is degrading in the browser with scrolling. You scroll a few lines a couple of times and the pixels are getting misaligned. This does not happen in the Original kernel and warthog-68 1.7. Also, I don't remember this being an issue with the previous version of F4., but there is no way for me to test that (I do only beta and it's not in the feed). The pixel mismatch clears right away after a pinch to zoom, but they get misaligned with scrolling after a few steps.

    Setup: min384, max 1.835 everything else is untouched F4
  19.    #79  
    Quote Originally Posted by hhy2k View Post
    I found a serious glitch. Text is degrading in the browser with scrolling. You scroll a few lines a couple of times and the pixels are getting misaligned. This does not happen in the Original kernel and warthog-68 1.7. Also, I don't remember this being an issue with the previous version of F4., but there is no way for me to test that (I do only beta and it's not in the feed). The pixel mismatch clears right away after a pinch to zoom, but they get misaligned with scrolling after a few steps.

    Setup: min384, max 1.835 everything else is untouched F4
    Need more info... please post output of below commands from CLI. Also a web page and steps to reproduce this.

    cat /proc/version
    dmesg
    find /sys/devices/system/cpu/cpufreq/override/* -print -exec cat {} \;
    find /sys/devices/system/cpu/cpu?/cpufreq/* -print -exec cat {} \;
    Live free or DIE!
  20. #80  
    For what it's worth, testing @ speedtest.net with the cpu at 1.7 sio ondemand seemed to induce "configuration load error" on the flash app on occasion that didn't happen at 1.5 using the same app. Milage may vary, but I thought it was worth mentioning.
Page 4 of 7 FirstFirst 1234567 LastLast

Posting Permissions