N900 consumes too much power

Hi,

Don't know if it can help but I had a similar problem. It was caused by a
bug related to auto-completion. My battery lasted only a few hours...

Have a look at this thread:http://talk.maemo.org/showthread.php?p=1020718
It worked for me.


Patrick



On 08/07/2012 11:25 AM, Eero Tamminen wrote:
> Hi,
>
> On 08/07/2012 04:20 AM, ext Cedric Cellier wrote:
>> Of course offline mode is not normal. I suggested it
>> for testing :) So that whatever software is the culprit
>> it can no longer drain the battery very fast. If you can't
>> stand several days in offline mode and without top reporting
>> a process that's heavy on the cpu
>
> It doesn't need to be that heavy.
>
> Even a process that constantly wakes up just at 1s interval,
> (which shows only as 0-1% of CPU usage in top) will
> reduce device idle use-time from one week[1] to a day.
>
> This is when just CPU is used. If also network is used,
> the wakeups can be much rarer and they still ruin the use-time
> even worse. I'm not completely sure of these numbers, but
> if your WLAN accesspoint power management is working, I think
> even <5 minute interval wakeups are bad, with phone networks,
> the network access intervals need to be much longer for device
> to be able to save power.
>
> These kind of wakeup frequencies you don't notice with top, you
> need either strace the programs or use e.g. nethogs[3] utility.
>
> Note also that the weaker the network signal strength is,
> the more power is required.
>
> [1] A week with a full, *new* battery [2], no services installed etc.
>
> [2] Battery capacity worsens with time, with use and higher temperatures
> (check the manufacturing date when buying batteries!), see:
> https://bugs.maemo.org/show_bug.cgi?id=7636#c16
>
> [3] http://packages.debian.org/squeeze/nethogs
>
>
>> then you'll be certain your hardware is faulty.
>
>
> - Eero
>


--

---
VRANCKX Patrick Patrick.Vranckx@uclouvain.be
Responsable des Infrastructures de production du système d'information (LS/SGSI/SIPR)
Responsable des Infrastructures des réseaux du système d'information (LS/SGSI/SRI)
Universite Catholique de Louvain - UCL +32 10 47 38 70

Dne 7.8.2012 08:59, Michael Rösch napsal(a):
> Am 05.08.2012 21:45, schrieb Jan Knutar:
>> On Sunday 05 August 2012, Michael Rösch wrote:
>>
>>> The biggest leap in saving enery is to use smart-reflex.
>> Considering how abnormal his battery use is, I don't think smartreflex is
>> the answer. Considering smartreflex only makes a difference when the CPU
>> is on, and on a properly configured device, the CPU spends most of the
>> time off.
> You're right! Of course he has to find that process(es) that drains the
> battery.
> All information is on the pages that are linked somewhere in the thread.
> He statet that he is always connected to wifi at home, well that is the
> first thing to stop using Autodisconnect for example.
>
> CU Michael
You're right, Michael!

I have started to use AutoDisconnect and I have stopped to be connected
all the time to WiFi.
So this cause is removed. Let's watch my next steps :-)

Pavel

Dne 7.8.2012 09:36, Jan Knutar napsal(a):
> On Sunday 05 August 2012, Pavel Řezníček wrote:
>
>> My Maemo and N900 knowledge is maybe abnormal (abnormally poor). I
>> neearly sure not to have a properly configured device. Now it comes
>> the special moment when I begin to study how to get my N900 properly
>> configured :-)
> Properly configured is something like what the device is like when taken
> out of the retail box after buying it.. Maybe removing the location
> widget and the calendar widget from the desktops.
>
> Then after that there are lots of things you can install and do to make
> it burn battery.
>
> For example, if you turn off wifi power saving and connect to wifi, you'd
> go down from several days of standby, to 6 hours of standby. Also if
> your wifi access point doesn't support wifi powersaving, or has
> broken/buggy wifi powersaving. (Which is probaly why one would turn it off
> on N900) You said it happens also when not on wifi, so I guess that's the
> problem in this case.
>
Thank you, Honzo.
So you suggest me to remove the calendar widget from the desktop? I have
it there… Yes, it consumes about 1% of CPU time…
Yes, location widget! Wow, how did you guess? :-D It goes out of the
window right away. Also about 1% of CPU time.
I have WiFi power saving enabled.

Pavel

Dne 7.8.2012 09:03, Michael Rösch napsal(a):
> Am 05.08.2012 20:25, schrieb Pavel Řezníček:
>
>> Now I don't exactly know what smart-reflex is for now. I surely can
>> search for it but if you could explain it also for the others, I'd be
>> happy :-)
> You can look it up here: http://wiki.maemo.org/Smartreflex
>
>> OK, I'm not sure if I have the kernel-power installed for now. I'll
>> definitely install it. Last time the installation ran without problems.
>> CSSU. This is still a mystery for me. It's a sure choice then. I notice
>> I am a bit noob :-D
>>
> Open a Terminal an type
>
> uname -a
>
> Then the prompt should be
>
> Linux $HOSTNAME 2.6.28.10-power50 #1 PREEMPT Sun Mar 18 20:10:56 EET
> 2012 armv7l GNU/Linux
>
> HTH & CU
>
> Michael

Thank you for the link, Michael.

Here follows my uname output:

~ $ uname -a
Linux Nokia-N900 2.6.28.10-power50 #1 PREEMPT Sun Mar 18 20:10:56 EET 2012 armv7l GNU/Linux

As you see, I just installed the Power Kernel (the day before yesterday)
;-)

Pavel

Dne 7.8.2012 03:20, Cedric Cellier napsal(a):
> ----- Message d'origine -----
>> The first really kicked me! I've never used the offline mode because I
>> want people to reach me on my phone number.
>> So I wasn't really right I don't use GPRS/2.5G/3G. I am connected to my
>> cell operator all the time even when not connected to their data
>> services! I just only use the call and SMS services.
>> To be clear: Is it normal to be connected to this basic services all the
>> time with an N900 device?
>> Or did you, Rixed, recommend me to go offline only for testing purposes?
> Of course offline mode is not normal. I suggested it
> for testing :) So that whatever software is the culprit it can no longer drain the battery very fast. If you can't stand several days in offline mode and without top reporting a process that's heavy on the cpu then you'll be certain your harware is faulty.
>
>
Dear Cedric,
I've tried to go in offline mode for an afternoon and a night. And yes,
my N900's battery lasted all the time, about 18 hours, still having
approx. 64.5% of energy. That's a great difference! I don't remember my
device to stay on for so long off the charger ;-)

Pavel

Dne 7.8.2012 15:10, Patrick Vranckx napsal(a):
>
> Hi,
>
> Don't know if it can help but I had a similar problem. It was caused
> by a
> bug related to auto-completion. My battery lasted only a few hours...
>
> Have a look at this thread:http://talk.maemo.org/showthread.php?p=1020718
> It worked for me.
>
>
> Patrick
>

BINGO, Patrick!

Thanks you a hundred thousand times, that was exactly my problem!
I already wanted to write about hildon-input-method crashing and
consuming about 90-96% of CPU time.
Now you were faster than my thoughts ;-)
So is it all about broken autocompletion dictionaries? Doh!

Here is the solution from your link (for the conveniency of others
falling into the same issue):

1) Run xterminal and "sudo gainroot"
2) "cd /home/user/.osso/"
3) "mv dictionaries olddics"
4) Restart your phone

How do they get broken? (I don't expect you to answer this question.)
/usr/bin/hildon-input-method is really the worst battery killer I have
noticed.
I'm watching my battery using BatteryGraph continuously.
Now I'm going to bed to move my broken dictionaries out of the way ;-)
Maybe we should raise a bug:

1. The dictionaries shouldn't get broken by autocompletion software itself.
2. If the dictionaries get broken somehow, the hildon-input-method
program should not crash or behave in such a buggy way that it
consumes all available CPU time.

Should I raise the bug or is anyone here to take the initiative? Could
you recommend me where to open it? Should this bug's fix be implemented
by the CSSU team?

Many thanks for you all, guys! I'd be helpless without you. This is what
I really love about open source communities 8-)

Pavel

Dne 6.8.2012 17:04, Paul Hartman napsal(a):
> On Sun, Aug 5, 2012 at 3:52 AM, Pavel Řezníček <cigydd@gmail.com> wrote:
>> Dear comaemoers,
>>
>> I have recently read an article, a review of Nokia N900. The reviewer stated
>> that his device lasts 3 days on battery when only sometimes calling or
>> writing SMS.
>> This is what I hardly can believe because my N900 lasts only 3, max. 6 hours
>> on battery and then the battery gets discharged.
>> Was the reviewer kidding or is something wrong with my mobile computer?
>> I have recently bought a new battery and the effect is minimal.
>> My device was in repair service already and they changed the motherboard. It
>> helped a bit but I never made it to that 3 days on battery.
>> Any ideas? :-)
> Using Scud battery + CSSU + kernel-power with smartreflex enabled,
> usually but not always connected to either 3.5G cellular internet or
> wifi, with email refresh every 15 minutes, no IM or voip accounts, no
> wifi autoconnect, brightness set to 2nd lowest setting, usually it
> will last me about a day and a half.
>
> If I'm listening to streaming music then battery dies much faster, a
> time measured in hours not in days.
>
> Some installed programs/widgets may be consuming CPU, daemons running
> in the background. HAM auto-update check can be a killer if you have a
> lot of repos enabled. If you are familiar with standard linux tools
> you can look in top and similar and find out what's running.
>
> BatteryGraph package is great, it will show when your battery usage is
> occurring, whether you are connected to gprs or wifi at the time.
> Maybe you can figure out if battery loss is constant or if there's a
> particular activity you did that is causing rapid drain.
>
> I also find that battery drain is affected by environmental
> conditions. If the N900 is in my pocket all day, or I'm in a place
> with poor signal, battery drains faster than if it is sitting on my
> desk with full bars.
>
> N900 in offline mode will last a week, but is not as fun. :)
Thank you, Paul!

As I already announced on another place in this thread,

1. I use BatteryGraph now a lot. I monitor the power consumption
thoroughly.
2. I have kernel-power-50 and CSSU installed.
3. I'm going to enable smartreflex.
4. I'm going to remove the calendar and location widgets from the
desktop (even though I loved to have the calendar close to my eyes…).
5. I started to use AutoDisconnect.
6. I reduced the WiFi power from 100 mW to 10 mW.
7. The biggest battery killer was hildon-input-method having corrupt
autocompletion dictionaries. I asked here if we should raise a bug
against this situation.

Pavel

On Thu, Aug 9, 2012 at 7:21 PM, Pavel Řezníček <cigydd@gmail.com> wrote:
> I'm going to remove the calendar and location widgets from the desktop (even
> though I loved to have the calendar close to my eyes…).

I use the "Calendar Home Widget" by Nicolai Hess (package name:
cal-home-widget) which is a replacement for stock calendar widget, and
more configurable as well (can change size, choose which view to open
calendar, filter events to hide, etc). It doesn't give me any battery
problems so far. :)

On Thu, Aug 9, 2012 at 7:07 PM, Pavel Řezníček <cigydd@gmail.com> wrote:
> Should I raise the bug or is anyone here to take the initiative? Could you
> recommend me where to open it? Should this bug's fix be implemented by the
> CSSU team?

A bug report already exists:
https://bugs.maemo.org/show_bug.cgi?id=8657

Dne 10.8.2012 16:47, Paul Hartman napsal(a):
> On Thu, Aug 9, 2012 at 7:07 PM, Pavel Řezníček <cigydd@gmail.com> wrote:
>> Should I raise the bug or is anyone here to take the initiative? Could you
>> recommend me where to open it? Should this bug's fix be implemented by the
>> CSSU team?
> A bug report already exists:
> https://bugs.maemo.org/show_bug.cgi?id=8657
Great, Paul! Thank you for your info ;-)

Pavel