[ltp] Thinkpad Power Consumption Roundup [TPCR]

Volker Krüger linux-thinkpad@linux-thinkpad.org
Wed, 08 Apr 2009 18:24:46 +0200


Hi all,

Thanks for Ferdinand to spawn this thread.

I believe that the time of many of us is rather limited, and unless
someone is a wizard in writing such scripts (which I am not), it would
take too long time.
What I had in mind was the following:

we need to identify where everyone hides his powertuning tricks.
I, for one, have a) put most of my settings into /etc/rc.local and b)
done some changes to the config file of the laptop-mode tools, (c) 
changed some module parameters and stored them in a file
/etc/modprobe.d/modules.con.local and (d) have a very small set of
adapted config files (e.g. xorg.conf) where the parameters cannot be moved.

one first step could be to identify the smartest places to put these
adaptations so that we have some consistency and thus can exchange more
deterministically parameter settings. Furthermore, it becomes evident
whether it matters, WHERE the parameters are set (i couldn't image that
it matters, but who knows).

Once we have that, it should be possible to compare the different files
and everyone can imitate the other.

I have updated my system to the newest ubuntu, which gives the
possibility to compare absolute values with others running the same OS.
For different OSs, we can at least get a local change in consumption
when parameters are changed (positive/negative gradients).

Another question is: what programs are running? Gnome?kde? something
else? What goodies are running? I.e. OS setup. I think I have pretty
much the standard ubuntu with compiz, gnome and the standard goodies.

how to measure the consumption? powertop?

finally: what laptop do we have?

questions that I would like to answer:
--why does Ferdinand's laptop consume so little? can we identify the
differences in software? is it due to the more efficient processor in
the x301 compared to the SL9400 in my x200s?
--Is it possible to compare my laptop with Christophs laptop which seem
to be rather similar? Can we perhaps identify that we have a completely
different setup but still consume the same amount of power?

opinions?

and last but not least:
how do you guys suggest we should publish our settings? publish them on
this thread? It's not going to be much anyway.

does it make sense to start with the "simple" stuff and then, when we
have reached some optimum, we could look into more advanced stuff like
undervoting the processor (advanced in the sense that a) we need to
patch the kernel and b) it seems not to be obvious how it will work for
Core Duo processors).

best,
Volker


Yves Dorfsman wrote:
> axp wrote:
>
>> This thread is dedicated to gather some ideas on how to measure a power
>> consumption for X200[s] and X30[0-1] .
>>
>>
>> Which metric to take into account?
>> Definition of an usable system (as Volker pointed out, turning off WLAN
>> and Display brightness at lowest value may not be representing an usable
>> system)
>>
>
> I don't have an X series, but maybe you guys should write a script
> like the  suspend/resume script from Ubuntu
> (https://wiki.ubuntu.com/KernelTeam/SuspendResumeTesting). They keep
> querying the battery to check the consumption between tasks.
>
> With a script, you'd be sure that the same tasks have been ran, and
> you could also check the setting  before you run a new task, prompting
> the user to enable/disable the wireless lan etc...
>


-- 
Volker Krüger
CVMI-CIT
Aalborg University
Lautrupvang 15
3750 Ballerup
Denmark
phone:++45-9940-2495
web:www.cvmi.aau.dk/~vok