Linux power saving utility

[Jan Goyvaerts]

I didn't quite get the name of the power saving utility Dick talked about. Alas, it's not in the show notes either.

Can somebody point me the right web site ? Thanks !

Jan

[Dick Wall]

It's called Granola:

http://grano.la/

and despite what we said on the podcast, there are both Linux and
Windows versions available, not just Linux.

Dick

[Jan Goyvaerts]

Thank you so much for helping saving belgian trees ! :-)

--
You received this message because you are subscribed to the Google Groups "The Java Posse" group.
To post to this group, send email to java...@googlegroups.com.
To unsubscribe from this group, send email to javaposse+...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/javaposse?hl=en.

[Jan Goyvaerts]

I just saved a daisy. oohhhhhhhh :-)

On Tue, Aug 3, 2010 at 18:30, Dick Wall <dick...@gmail.com> wrote:

[Wildam Martin]

I am about to try this but I can't find any hint in the documentation
if this can be turned easily on and off - just for the case I get
problems after installation.

Can this be easily completely turned off (and back on) if installed?

--
Martin Wildam
http://www.google.com/profiles/mwildam

[Casper Bang]

Just a note since I have metered power outlets and have been
investigating power aspects for a long time. I have NOT been able to
measure any advantage in power consumption using the granola userspace
governor rather than Ubuntu's on-demand kernel governor. Furthermore,
Granola is closed source... we don't really know what goes on inside
the software.

What does save power though is to use a modern scaling CPU (>= Core
II) and under-clock it, use laptop drives and preferable SSD and
unless you really need GFX, go with something embedded. Also, many
PSU's waste about ~30% of the energy during the induction step but you
can get high-efficient ones which only waste ~10%.

Last but not least, unplug the PC + misc. when going to bed. My home-
office setup will happily draw about 40W while doing absolutely
nothing than producing heat.

/Casper

[Wildam Martin]

On Wed, Aug 4, 2010 at 12:53, Casper Bang <caspe...@gmail.com> wrote:
> Just a note since I have metered power outlets and have been
> investigating power aspects for a long time. I have NOT been able to
> measure any advantage in power consumption using the granola userspace
> governor rather than Ubuntu's on-demand kernel governor. Furthermore,
> Granola is closed source... we don't really know what goes on inside
> the software.

Thank you for sharing your experience.
When talking about Ubuntu's on-demand kernel governor - what are you
talking about - the "powernowd" package?

> Last but not least, unplug the PC + misc. when going to bed. My home-
> office setup will happily draw about 40W while doing absolutely
> nothing than producing heat.

Fortunately my Ubuntu incl login is up in about a minute - so no more
real need to have it running when not in use.
Sometimes (when many apps open) I also often put it to suspend - also
saves a lot of power (and I guess more than any other tool can save
when running).

[Casper Bang]

> When talking about Ubuntu's on-demand kernel governor - what are you
> talking about - the "powernowd" package?

No, powernowd is a just userspace client deamon much like cpufreqd
(which I happen to be using). Both interact with the sysfs interface.
Your Ubuntu system likely already runs with a modern kernel-level
scaling power governor, i.e. on my 10.04 I can observe it's presence
here:

/sys/devices/system/cpu/cpu*/cpufreq/scaling_governor

Note that possibly Granola removes this entry since it assumes
responsibility. The way it would do this is to issue commands much
like you could yourself. I.e. to see what your CPU's are running at
presently you issue:

cat /proc/cpuinfo | grep MHz

If your CPU support PowerNow, Cool n' Quiet and whatever the CPU
manufacturer call their speed-stepping tech, you would be able to
inspect this with:

cat /sys/devices/system/cpu/cpuX/cpufreq/scaling_available_freq

...(where x is your core no). My Core II 6550 reports capable of
throttling at 2670000 2336000 and 2003000 KHz (and is usually running
at 2003000 KHz during a development session using the kernel on-demand
governor).

If I want my second core to go bananas because I know it has work to
do, I could thus issue the command:

cpufreq-selector -c 1 -f 2670000

It would remain in this state until I tell the governor to return to
on-demand like this:

cpufreq-selector -c 1 --governor ondemand

So anyway, I would not use userspace deamons for this - I get reminded
of Windows bloatware.

/Casper

[Casper Bang]

> Fortunately my Ubuntu incl login is up in about a minute - so no more
> real need to have it running when not in use.

Oh you should see Ubuntu on an Intel SSD then, boots in just a few
secs after initial power-on and self-test. Also it uses just 0.14W
during load and 0.06W when idling. ;)

[Jan Goyvaerts]

So you can objectively measure whether granula make it consume less/more power ?

[Wildam Martin]

On Wed, Aug 4, 2010 at 14:09, Casper Bang <caspe...@gmail.com> wrote:
> No, powernowd is a just userspace client deamon much like cpufreqd
> (which I happen to be using). Both interact with the sysfs interface.
> Your Ubuntu system likely already runs with a modern kernel-level
> scaling power governor, i.e. on my 10.04 I can observe it's presence
> here:
> /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor

Yes, I am running on 10.04 and yes, I have that too.
And I have Intel-Processors.

> If your CPU support PowerNow, Cool n' Quiet and whatever the CPU
> manufacturer call their speed-stepping tech, you would be able to
> inspect this with:
> cat /sys/devices/system/cpu/cpuX/cpufreq/scaling_available_freq

discovered typo - should be:
sudo cat /sys/devices/system/cpu/cpuX/cpufreq/scaling_available_frequencies

> ...(where x is your core no). My Core II 6550 reports capable of
> throttling at 2670000 2336000 and 2003000 KHz (and is usually running
> at 2003000 KHz during a development session using the kernel on-demand
> governor).

I also have several values there. Thank you.

> So anyway, I would not use userspace deamons for this - I get reminded
> of Windows bloatware.

Agree. Thank you very much for your explanations again!

[Casper Bang]

Lately the topic objective vs. subjective and micro-benchmarks became
rather muddy territory and it seems like you are trying to push me out
there. What I can tell you is what I did to reach the conclusion based
on MY typical usage pattern.

I recorded the KWh over a period 3 hours from the outlet and made sure
I did the same kind of conservative work (NetBeans, Chrome + a few
other things). The meter is standards approved and rated at 1%
accuracy, or +/- 0.75W considering the system draws around 75W. There
was no real difference to observe, the two numbers came very close to
one another.

That setup is good enough for me to assert that it does nothing for
me. If you have another demand/usage that differs from mine you might
come to another conclusion, however considering my CPU already runs at
the lowest possible stepping, in hindsight I am not sure which other
conclusion I might actually get to. It's not like applications can
just magically underclock or undervoltate the hardware.

Can you objectively theorize how Granula would work to achieve it's
goal better than a kernel level ondemand driver?

/Casper

On Aug 4, 2:34 pm, Jan Goyvaerts <java.arti...@gmail.com> wrote:
> So you can objectively measure whether granula make it consume less/more
> power ?
>
>
>
> On Wed, Aug 4, 2010 at 14:22, Casper Bang <casper.b...@gmail.com> wrote:
> > > Fortunately my Ubuntu incl login is up in about a minute - so no more
> > > real need to have it running when not in use.
>
> > Oh you should see Ubuntu on an Intel SSD then, boots in just a few
> > secs after initial power-on and self-test. Also it uses just 0.14W
> > during load and 0.06W when idling. ;)
>
> > --
> > You received this message because you are subscribed to the Google Groups
> > "The Java Posse" group.
> > To post to this group, send email to java...@googlegroups.com.
> > To unsubscribe from this group, send email to

> > javaposse+...@googlegroups.com<javaposse%2Bunsubscribe@googlegroups .com>
> > .

[Jan Goyvaerts]

Ow ! It was meant as an honest question ! No sarcasm intended.

I'm certainly not pretending granula does a better job. I'm just interested how one measures power consumption. Hence my question.

To unsubscribe from this group, send email to javaposse+...@googlegroups.com.

[Vince O'Sullivan]

On Aug 4, 5:32 pm, Jan Goyvaerts <java.arti...@gmail.com> wrote:
> I'm certainly not pretending granula does a better job. I'm just interested
> how one measures power consumption. Hence my question.

At home we have one of these installed. Basically it's consists of a
small battery powered transmitter that loops around the main power
inlet and detects the current passing along the wire, and a small
battery powered wireless receiver that sits anywhere in the house.
Standard batteries last for several months.

http://www.amazon.co.uk/Owl-CM-119-Wireless-Electricity/dp/B002C76WTW/ref=sr_1_2?ie=UTF8&s=electronics&qid=1281008274&sr=8-2

V.

[Casper Bang]

Cool device. However, the problem with these cheap meters(which rely
on induction and the hall effect) is that they are not accurate enough
to be useful for for anything but fridges, ranges etc.

If you notice the datasheet [http://www.farnell.com/datasheets/
409580.pdf] and apply this to a PC (75W@230V), that's 326mA which is
way below the specified 1A/10% threshold - so you're probably looking
at about 20-30% accuracy.

The ones I use are similar to this one [http://www.savingtrust.dk/
consumer/products/energy-saving-equipment/electricity-meters/facts-and-
figures] and are accurate to within 1% even for low watt devices. The
only problem with these direct/galvanic connected ones is that they
can only be used on one-phased appliances, i.e. I can not use it for
my stove/range.

/Casper

> http://www.amazon.co.uk/Owl-CM-119-Wireless-Electricity/dp/B002C76WTW...
>
> V.

[Paul Gearon]

On Thu, Aug 5, 2010 at 9:17 AM, Casper Bang <caspe...@gmail.com> wrote:
> Cool device. However, the problem with these cheap meters(which rely
> on induction and the hall effect) is that they are not accurate enough
> to be useful for for anything but fridges, ranges etc.

Errr..... Induction: yes. Hall effect: no.

You'd have to strip down the wires on your appliance to use the hall
effect, since this requires measuring a current across the wire.
Instead, they use a toroid around the wire to get the best inductive
coupling they can and measure the current there.

Paul

[Casper Bang]

> Errr..... Induction: yes. Hall effect: no.
>
> You'd have to strip down the wires on your appliance to use the hall
> effect, since this requires measuring a current across the wire.
> Instead, they use a toroid around the wire to get the best inductive
> coupling they can and measure the current there.

Thanks for the correction, it's obviously a long time ago I took a
physics course. :)

[Fabrizio Giudici]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Actually you made me feel younger by mentioning the Hall effect (that
I only vaguely recalled :-)

- --
Fabrizio Giudici - Java Architect, Project Manager
Tidalwave s.a.s. - "We make Java work. Everywhere."
java.net/blog/fabriziogiudici - www.tidalwave.it/people
Fabrizio...@tidalwave.it
-----BEGIN PGP SIGNATURE-----
Version: GnuPG/MacGPG2 v2.0.14 (Darwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAkxayYsACgkQeDweFqgUGxdXlACgipR6SfDFbO4F8a3+H5H4R9DE
WHoAnj6GKe0zcIAcjx1flG5Hhoc7hWF8
=w013
-----END PGP SIGNATURE-----

[Paul Gearon]

On Thu, Aug 5, 2010 at 10:24 AM, Fabrizio Giudici
<fabrizio...@tidalwave.it> wrote:
> Actually you made me feel younger by mentioning the Hall effect (that
> I only vaguely recalled :-)

Given the conversations in here, I'm guessing that quite a few people
in here got their degree in Elec. Eng. :-)

Paul

[Casper Bang]

Hehe interesting observation. Personally I jumped over to computers
and programming when I discovered how fast/cheap mistakes could be
corrected as oppose to having to etch a new PCB, erase the EPROM, wait
for new parts etc. :)

On Aug 5, 4:28 pm, Paul Gearon <gea...@ieee.org> wrote:
> On Thu, Aug 5, 2010 at 10:24 AM, Fabrizio Giudici
>