Thinkpad X60s power usage on battery - Windows/Linux comparison

Thinkpad X60s power usage on battery - Windows/Linux comparison

[Vedran Rodic]

I've decided to write up something that I've been aware of for some time,
but that didn't really bother me that much, since I've been waiting for the
situation to improve quietly.

The basic observation is: My trusty X60s uses less power when on battery
on Windows than on Linux.

This situation hasn't changed significantly despite the appearance of powertop
to trim the number of processor wakeups per second thus enabling the
processor to stay in deep sleep states for longer time and improvements in
the Xorg intel video driver.

I get arround 3.3 wakeups per second minimally when
my Xorg/KDE is not active, with laptop using 9.1 W (as reported by
powertop). When Xorg/KDE is running, with one konsole to monitor usage with
powertop, I get arround 13 wakeups per second and power usage of 9.4 W. All
the measurements are done with LCD brightness on the lowest level, wireless
interfaces disabled, and all power saving features that powertop can detect
enabled. Processor is in C3 state 99.9% of the time. I tried removing
various driver modules that I didn't use at the time (irda, pcmcia,
ethernet, sound), but no change.

Windows power monitoring tool from Lenovo reports power usage of 6.5 W under
the same LCD brightness level and with wireless disabled.

This was tested under Linux 2.6.26-rc8 (some git level after the rc8
release). Older
versions of the kernel didn't show any statistically significant
differences. powertop
is version 1.10, and the software used under Windows XP
is Lenovo ThinkVantage Power Manager.

Since I doubt that Windows XP has less wakeups per second than my
Linux setup, I guess
there are still a couple of IBM specific power saving
registers/controls that can be
tweaked so that the laptop uses less power. Does anybody have any idea
on what would that
be? Any friendly contacts at lenovo?

Thanks
Vedran Rodic

Re: Thinkpad X60s power usage on battery - Windows/Linux comparison

[Andi Kleen]

"Vedran Rodic" <vrodic@gmail.com> writes:
>
> Windows power monitoring tool from Lenovo reports power usage of 6.5 W under
> the same LCD brightness level and with wireless disabled.

It would be better if you compared using the same "neutral" method.

Most cheap power measurements tend to quite inaccurate and who
knows in which direction this Windows power monitoring tool is wrong.
It might be wrong in a different direction than what you used 
under Linux.

A reasonable accurate method might be to just measure the time how
long the battery lasts in both cases using a similar workload
(e.g. keeping it completely idle)

-Andi

Re: Thinkpad X60s power usage on battery - Windows/Linux comparison

[Vedran Rodic]

On Tue, Aug 19, 2008 at 3:42 AM, Andi Kleen <andi@firstfloor.org> wrote:
> "Vedran Rodic" <vrodic@gmail.com> writes:
>>
>> Windows power monitoring tool from Lenovo reports power usage of 6.5 W under
>> the same LCD brightness level and with wireless disabled.
>
> It would be better if you compared using the same "neutral" method.
>
> Most cheap power measurements tend to quite inaccurate and who
> knows in which direction this Windows power monitoring tool is wrong.
> It might be wrong in a different direction than what you used
> under Linux.

Yeah, or they might not be wrong and my eyes could be :( It turns out
that Windows can get the LCD brightness down more than thinkpad-acpi,
or whatever is turning down the brightnes when I press the Fn-End
combination (both in X and in the console - I guess it's the BIOS).

>
> A reasonable accurate method might be to just measure the time how
> long the battery lasts in both cases using a similar workload
> (e.g. keeping it completely idle)

Difference between lowest Windows LCD brightness levels and Linux
brightness levels:

3:40h on Windows (it hibernated the laptop when the battery was
arround 5% percent, did arround 10 more minutes sitting on the grub
screen).
2:36 on Linux.

Now the question is how do I get the LCD brightness down even more on
X, but this may already be solved in the latest Xorg intel driver,
just not with the Fn keys.

If I see any more surprising battery life differences when I manage to
get the same low brightness level on both OS-es, I'll let you know.

Vedran

Re: Thinkpad X60s power usage on battery - Windows/Linux comparison

[Henrique de Moraes Holschuh]

On Thu, 21 Aug 2008, Vedran Rodic wrote:
> Yeah, or they might not be wrong and my eyes could be :( It turns out
> that Windows can get the LCD brightness down more than thinkpad-acpi,
> or whatever is turning down the brightnes when I press the Fn-End
> combination (both in X and in the console - I guess it's the BIOS).

Intel embedded graphics?  You will need something that is not in the kernel
yet, but it is in the works (IGD ioregion support).

ThinkPad-acpi really can't to much more than what it is doing already,
AFAIK.

-- 
  "One disk to rule them all, One disk to find them. One disk to bring
  them all and in the darkness grind them. In the Land of Redmond
  where the shadows lie." -- The Silicon Valley Tarot
  Henrique Holschuh

Re: Thinkpad X60s power usage on battery - Windows/Linux comparison

[Karthik Gopalakrishnan]

What sort of a Graphics Accelerator does the system have. I faced a
similar problem on my Laptop & I narrowed down the issue to a bad
driver for my NVidia Graphics card. Once I fixed that, the battery
life on Linux matched up to that on Windows XP.

Regards,
Karthik

On Mon, Aug 18, 2008 at 6:54 PM, Vedran Rodic <vrodic@gmail.com> wrote:
> I've decided to write up something that I've been aware of for some time,
> but that didn't really bother me that much, since I've been waiting for the
> situation to improve quietly.
>
> The basic observation is: My trusty X60s uses less power when on battery
> on Windows than on Linux.
>
> This situation hasn't changed significantly despite the appearance of powertop
> to trim the number of processor wakeups per second thus enabling the
> processor to stay in deep sleep states for longer time and improvements in
> the Xorg intel video driver.
>
> I get arround 3.3 wakeups per second minimally when
> my Xorg/KDE is not active, with laptop using 9.1 W (as reported by
> powertop). When Xorg/KDE is running, with one konsole to monitor usage with
> powertop, I get arround 13 wakeups per second and power usage of 9.4 W. All
> the measurements are done with LCD brightness on the lowest level, wireless
> interfaces disabled, and all power saving features that powertop can detect
> enabled. Processor is in C3 state 99.9% of the time. I tried removing
> various driver modules that I didn't use at the time (irda, pcmcia,
> ethernet, sound), but no change.
>
> Windows power monitoring tool from Lenovo reports power usage of 6.5 W under
> the same LCD brightness level and with wireless disabled.
>
> This was tested under Linux 2.6.26-rc8 (some git level after the rc8
> release). Older
> versions of the kernel didn't show any statistically significant
> differences. powertop
> is version 1.10, and the software used under Windows XP
> is Lenovo ThinkVantage Power Manager.
>
> Since I doubt that Windows XP has less wakeups per second than my
> Linux setup, I guess
> there are still a couple of IBM specific power saving
> registers/controls that can be
> tweaked so that the laptop uses less power. Does anybody have any idea
> on what would that
> be? Any friendly contacts at lenovo?
>
> Thanks
> Vedran Rodic
> --
> To unsubscribe from this list: send the line "unsubscribe linux-acpi" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>