N900 battery duration

Andrea Grandi wrote:
> Hi
>
> p.s: my N900 was connected to charger during this test... do I have to
> disconnect it before doing the test?

The charger seems to mess up the numbers at least somewhat. I'll do more
testing with powertop later.

--
Tuomas

Tuomas Kulve wrote:
> The charger seems to mess up the numbers at least somewhat. I'll do more
> testing with powertop later.

Well, I did a quite test right away.

This is Media Player playing 192kbps VBR Ogg/Vorbis without charger:

C# | Ratio | Avg/dura | Frequency | Ratio
--------+--------+----------+-----------+--------+
C0 | 66.3% | | 600 MHz | 16.5% |
C1 | 0.0% | 3.1ms | 550 MHz | 0.0% |
C2 | 7.4% | 3.3ms | 500 MHz | 50.6% |
C3 | 26.3% | 7.6ms | 250 MHz | 32.9% |
C4 | 0.0% | |

And the same with charger:

C# | Ratio | Avg/dura | Frequency | Ratio
--------+--------+----------+-----------+--------+
C0 | 65.6% | | 600 MHz | 15.8% |
C1 | 0.0% | 0.1ms | 550 MHz | 1.3% |
C2 | 8.8% | 2.7ms | 500 MHz | 52.7% |
C3 | 25.7% | 7.5ms | 250 MHz | 30.2% |
C4 | 0.0% | |

No change there. Not sure what I did with an ealier firmware as there I
did saw a clear distinction. Maybe I just messed something else up..

--
Tuomas

When I think back to the N900 hardware lighting talk and the warning
about not fixing the CPU's clock to one of the high frequencies, I guess
that connecting the N900 to a charger should not change the power
settings of the system, i.e. not like with a laptop CPU that's behaving
differently when running on batteries. But that's just a guess, someone
with more knowledge of the ARM Cortex 8 architecture or even better the
N900's architecture (Nokian's, where are you?) should have the right
answer to that.

Am Freitag, den 16.10.2009, 21:59 +0200 schrieb Andrea Grandi:
> Hi
>
> p.s: my N900 was connected to charger during this test... do I have to
> disconnect it before doing the test?
>
> 2009/10/16 Andrea Grandi <a.grandi@gmail.com>:
> > Hi,
> >
> > 2009/10/16 Kai Thomsen <thomsenk@t-online.de>:
> >> Hmm, Skype and apps like that might suck quite a bit of battery,
> >> depending on how chatty they might become in their idle states. Knowing
> >> a bit about Skype from monitoring a large company network, especially
> >> when it does not get a connection or an unreliable connection it tries
> >> to connect to the next Supernode very often, at least once a minute.
> >> Wireshark might help you to find out how "chatty" your N900 is...
> >> If I have some time over the weekend, I could run some tests with
> >> Wireshark and Skype/Foreca/Facebook and whatever else is usually running
> >> on my N900 to see how often these apps actually send traffic in their
> >> idle states.
> >
> > it would be great!
> >
> > Anyway, here is the result of powertop execution on my N900:
> > http://pastebin.ca/1624111
> >
> > I hope it can help you to understand what's happening :)
> >
> > --
> > Andrea Grandi
> > email: a.grandi [AT] gmail [DOT] com
> > website: http://www.andreagrandi.it
> > PGP Key: http://www.andreagrandi.it/pgp_key.asc
> >
>
>
>

OK, I'll do some sniffing over the weekend with a Wifi connection and
post the results on Monday.

Am Freitag, den 16.10.2009, 21:58 +0200 schrieb Andrea Grandi:
> Hi,
>
> 2009/10/16 Kai Thomsen <thomsenk@t-online.de>:
> > Hmm, Skype and apps like that might suck quite a bit of battery,
> > depending on how chatty they might become in their idle states. Knowing
> > a bit about Skype from monitoring a large company network, especially
> > when it does not get a connection or an unreliable connection it tries
> > to connect to the next Supernode very often, at least once a minute.
> > Wireshark might help you to find out how "chatty" your N900 is...
> > If I have some time over the weekend, I could run some tests with
> > Wireshark and Skype/Foreca/Facebook and whatever else is usually running
> > on my N900 to see how often these apps actually send traffic in their
> > idle states.
>
> it would be great!
>
> Anyway, here is the result of powertop execution on my N900:
> http://pastebin.ca/1624111
>
> I hope it can help you to understand what's happening :)
>

Andrea Grandi wrote:
> Anyway, here is the result of powertop execution on my N900:
> http://pastebin.ca/1624111
>
> I hope it can help you to understand what's happening :)

Copy paste from the link above:

C# | Ratio | Avg/dura | Frequency | Ratio
--------+--------+----------+-----------+--------+
C0 | 27.3% | | 600 MHz | 19.1% |
C1 | 0.0% | 0.2ms | 550 MHz | 0.0% |
C2 | 8.8% | 3.0ms | 500 MHz | 21.8% |
C3 | 63.7% | 34.2ms | 250 MHz | 59.1% |
C4 | 0.1% | 6.0ms |


Was that an idle device? It really doesn't look like that. Here's a same
set from my idle device:

C# | Ratio | Avg/dura | Frequency | Ratio
--------+--------+----------+-----------+--------+
C0 | 0.7% | | 600 MHz | 0.0% |
C1 | 0.0% | | 550 MHz | 0.0% |
C2 | 0.2% | 1.2ms | 500 MHz | 0.0% |
C3 | 99.2% | 211.0ms | 250 MHz | 100.0% |
C4 | 0.0% | |

C-state is the sleep state of the CPU (afaik) and the frequency is the
Mhz the cpufreq chooses the run the device based on the cpu load. I
don't know what it takes to get to C4.

--
Tuomas

Hi,

2009/10/16 Tuomas Kulve <tuomas@kulve.fi>:
> Was that an idle device? It really doesn't look like that. Here's a same
> set from my idle device:

no, it wasn't idle: I was connected to SSH and executed command from
my PC, wifi was ON, connected to Skype, GTalk, SIP account.
Facebook, Forecast widgets ON.

--
Andrea Grandi
email: a.grandi [AT] gmail [DOT] com
website: http://www.andreagrandi.it
PGP Key: http://www.andreagrandi.it/pgp_key.asc

hi,
sorry for topposting, owa sucks big time :-(

pmtrackerdaemon is a tool for internal usage which monitors the system and crashes when idle states have not been reached for a while.
it is not part of the images released.
it intefers with powertop, so powertop tries to remove it.
running powertop a second time should work.
powertop is not the best tool to track problems: the best way is to start from a clean slate, install apps one by one and check them first with top and strace.
the powertop can be used.
a less precise but simpler way is to enable rd mode, let the keyboard open and observe the backlight leds.
the less they are on, the better.

the reference should be an idle device, free from 3rd party applications, especially untested stuff coming out from the plain extras repository.

igor
________________________________________
From: maemo-users-bounces@maemo.org [maemo-users-bounces@maemo.org] On Behalf Of ext Tuomas Kulve [tuomas@kulve.fi]
Sent: 16 October 2009 18:44
To: maemo-users@maemo.org
Subject: Re: powertop (was: N900 battery duration)

Andrea Grandi wrote:
> # powertop
> Powertop 1.13.3
> status: Unknown job: pmtrackerdaemon
> Mounting debugfs...FAILED

I get the same.

Any idea what's that about?

--
Tuomas

Andrea Grandi wrote:
> Hi,
>
> 2009/10/16 Tuomas Kulve <tuomas@kulve.fi>:
>> Was that an idle device? It really doesn't look like that. Here's a same
>> set from my idle device:
>
> no, it wasn't idle: I was connected to SSH and executed command from
> my PC, wifi was ON, connected to Skype, GTalk, SIP account.
> Facebook, Forecast widgets ON.

That still counts as an idle device :)

Or at least should. But obviously something is running actively there
all the time.

Have you tried running "top" if that shows something?

--
Tuomas

Tuomas Kulve wrote:

> I don't know what it takes to get to C4.

This is now run on a 41-10 release without SIM, without WLAN, without
charger, with only ogg-support and sshd installed (default widgets on
the desktop), and display blanked via the lock button:

C# | Ratio | Avg/dura | Frequency | Ratio
--------+--------+----------+-----------+--------+
C0 | 0.3% | | 600 MHz | 0.0% |
C1 | 0.0% | | 550 MHz | 0.0% |
C2 | 0.1% | 1.1ms | 500 MHz | 0.0% |
C3 | 0.2% | 15.2ms | 250 MHz | 100.0% |
C4 | 99.5% | 2297.0ms |

--
Tuomas

On Sat, Oct 17, 2009 at 12:34, Tuomas Kulve <tuomas@kulve.fi> wrote:
>
> This is now run on a 41-10 release without SIM, without WLAN, without
> charger, with only ogg-support and sshd installed (default widgets on
> the desktop), and display blanked via the lock button:
>
>     C4 |  99.5% | 2297.0ms |

Bet it'll last for *aaaages* :-)

Cheers,

Andrew

--
Andrew Flegg -- mailto:andrew@bleb.org | http://www.bleb.org/