Liunx power consumption on laptops -- Enormous progress in the last few months
Hi,
with all the tickless and other goodies going into the kernel in the last few months, there is a lot of hope that this helps Linux reduce power consumption... and the good news is that it does... once you fix some bugs and fix a bunch of userspace applications.
While it's hard to show "one size fits all" number/percentage, we took a bog standard Lenovo T61 laptop (no vendor preference, they just were the first one to deliver a model with the latest Intel chipset to our cubes) and measured the effect. The baseline we used was a 32 bit Fedora 7 installation; note that this already has the tickless kernel, but is lacking several of the key bugfixes that came afterwards.
We've put our measurements in a graph at
http://ww.linuxpowertop.org/results.php
With kernel fixes and features, the power consumption of this laptop went from 21.06 Watts to 18.25 Watts; with 2 additional userspace fixes the power consumption ended up at 15.5 Watts.
(Don't worry that this is the end of it; there's more stuff in the various project pipelines, and we'll keep measuring the progress over time)
All in all, personally I'm very happy to see Linux making such a huge step forward with tickless and can't wait for this step to be available in all distros and for all architectures...
Greetings,
Arjan van de Ven
Quoted 7 lines · Arjan van de Ven
While it's hard to show "one size fits all" number/percentage, we took a bog standard Lenovo T61 laptop (no vendor preference, they just were the first one to deliver a model with the latest Intel chipset to our cubes) and measured the effect. The baseline we used was a 32 bit Fedora 7 installation; note that this already has the tickless kernel, but is lacking several of the key bugfixes that came afterwards.
Quoted 8 lines
We've put our measurements in a graph at
http://ww.linuxpowertop.org/results.php
With kernel fixes and features, the power consumption of this laptop went from 21.06 Watts to 18.25 Watts; with 2 additional userspace fixes the power consumption ended up at 15.5 Watts.
Where is the ondemand-fix.patch? I can't find any link to it.
Quoted 10 lines · Chuck Ebbert
Quoted 9 lines · Arjan van de Ven
While it's hard to show "one size fits all" number/percentage, we took a bog standard Lenovo T61 laptop (no vendor preference, they just were the first one to deliver a model with the latest Intel chipset to our cubes) and measured the effect. The baseline we used was a 32 bit Fedora 7 installation; note that this already has the tickless kernel, but is lacking several of the key bugfixes that came afterwards.
We've put our measurements in a graph at
Quoted 8 lines
Quoted 6 lines
http://ww.linuxpowertop.org/results.php
With kernel fixes and features, the power consumption of this laptop went from 21.06 Watts to 18.25 Watts; with 2 additional userspace fixes the power consumption ended up at 15.5 Watts.
Where is the ondemand-fix.patch? I can't find any link to it.
just click on it in the graph ;_
it's also http://www.linuxpowertop.org/patches/ondemand-fix.patch (which is submitted already to the maintainers)
Quoted 6 lines · Arjan van de Ven
Quoted 1 line
Where is the ondemand-fix.patch? I can't find any link to it.
just click on it in the graph ;_
it's also http://www.linuxpowertop.org/patches/ondemand-fix.patch
(which is submitted already to the maintainers)
Should be in cpufreq.git (and thus, -mm) too, waiting for .23 to open up.
Dave
Signature · 1 line
http://www.codemonkey.org.uk
Quoted 7 lines · Arjan van de Ven
Quoted 1 line
Where is the ondemand-fix.patch? I can't find any link to it.
just click on it in the graph ;_
it's also http://www.linuxpowertop.org/patches/ondemand-fix.patch (which is submitted already to the maintainers)
Ah, OK, that one just went into the latest Fedora 7 kernel.
Quoted 10 lines · Chuck Ebbert
Quoted 6 lines · Arjan van de Ven
Quoted 1 line
Where is the ondemand-fix.patch? I can't find any link to it.
just click on it in the graph ;_
it's also http://www.linuxpowertop.org/patches/ondemand-fix.patch (which is submitted already to the maintainers)
Ah, OK, that one just went into the latest Fedora 7 kernel.
All this information makes it obvious that even though upgrades are painful (given slow laptop disks, *really* painful), it looks as if there are major benefits to be gained.
Thanks for sharing the information.
As improved as USB drivers are, I assume that unloading the drivers is still desirable if no USB hardware is in use.
Signature · 3 lines
Bill Davidsen <[email protected]> "We have more to fear from the bungling of the incompetent than from the machinations of the wicked." - from Slashdot
Quoted 10 lines · Dave Jones
Quoted 6 lines · Arjan van de Ven
Quoted 1 line
Where is the ondemand-fix.patch? I can't find any link to it.
just click on it in the graph ;_
it's also http://www.linuxpowertop.org/patches/ondemand-fix.patch
(which is submitted already to the maintainers)Should be in cpufreq.git (and thus, -mm) too, waiting for .23 to open up.
Somebody should fix http://ww.linuxpowertop.org/known.php then, it says:
"The kernels ondemand CPU frequency management function currently has a high-frequency timer that samples to see if the CPU is idle. Intel fixed this and the patches to the kernel to effectively remove this timer are included in Linus' tree as of 2.6.22-rc1."
which confused me no end when I was looking at this stuff and wondering why it was in the -mm tree if Linus already picked it up. :)
Quoted 14 lines · On Mon, Jul 09, 2007 at 05:42:14PM -0400, [email protected] wrote:
Quoted 10 lines · Dave Jones
Quoted 6 lines · Arjan van de Ven
Quoted 1 line
Where is the ondemand-fix.patch? I can't find any link to it.
just click on it in the graph ;_
it's also http://www.linuxpowertop.org/patches/ondemand-fix.patch
(which is submitted already to the maintainers)Should be in cpufreq.git (and thus, -mm) too, waiting for .23 to open up.
Somebody should fix http://ww.linuxpowertop.org/known.php then, it says:
Quoted 7 lines
"The kernels ondemand CPU frequency management function currently has a high-frequency timer that samples to see if the CPU is idle. Intel fixed this and the patches to the kernel to effectively remove this timer are included in Linus' tree as of 2.6.22-rc1."
which confused me no end when I was looking at this stuff and wondering why it was in the -mm tree if Linus already picked it up. :)
There were two sets of fixes to ondemand. The first round did in fact go into .22rc1 The others I deemed too late in the cycle to go into .22 (Though they have been flawless, so if there was enough demand (ha!) we could
even push them to -stable once they're in mainline).
Dave
Signature · 1 line
http://www.codemonkey.org.uk
Quoted 11 lines · Bill Davidsen
Quoted 10 lines · Chuck Ebbert
Quoted 6 lines · Arjan van de Ven
Quoted 1 line
Where is the ondemand-fix.patch? I can't find any link to it.
just click on it in the graph ;_
it's also http://www.linuxpowertop.org/patches/ondemand-fix.patch (which is submitted already to the maintainers)
Ah, OK, that one just went into the latest Fedora 7 kernel.
Quoted 8 lines
All this information makes it obvious that even though upgrades are painful (given slow laptop disks, *really* painful), it looks as if there are major benefits to be gained.
Thanks for sharing the information.
As improved as USB drivers are, I assume that unloading the drivers is still desirable if no USB hardware is in use.
That should be no longer neccessary, IIRC.
Pavel
Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek (cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
PCI devices power management, w/o sysfs? [Was: Re: Liunx power consumption on laptops ...]
In reply to Arjan van de VenGreetings *!
I'm looking for a way to disable (PM low power mode) some devices I rarely/never use on my laptop to save some more power. With 2.6.17, IIRC, I was able to echo -n 2 > /sys/bus/pci/../power to put things into low power mode. (I don't know if it ever worked, but at least it looked like it did.)
..-2.6.22(-ck1) w/ CONFIG_PM_SYSFS_DEPRECATED doesn't let me change this state. evident :) -----------------^^^^^^^^^^
Is there a proper way to do this (setpci, if I'd know the registers)? Is it gone/going inside drivers (like USB_SUSPEND) and I don't have to worry?
I currently don't load the modules in question - but I'm not sure if it helps, as they might already be powered on by PCI initialization/BIOS (/sys/bus/pci/../state cats "0").
THX, ritch
Re: PCI devices power management, w/o sysfs? [Was: Re: Liunx power consumption on laptops ...]
Quoted 11 lines · Richard Mittendorfer
Greetings *!
I'm looking for a way to disable (PM low power mode) some devices I rarely/never use on my laptop to save some more power. With 2.6.17, IIRC, I was able to echo -n 2 >
/sys/bus/pci/../powerto put things into low power mode. (I don't know if it ever worked, but at least it looked like it did.)..-2.6.22(-ck1) w/
CONFIG_PM_SYSFS_DEPRECATEDdoesn't let me change this state. evident :) -----------------^^^^^^^^^^
Quoted 7 lines
Is there a proper way to do this (setpci, if I'd know the registers)? Is it gone/going inside drivers (like
USB_SUSPEND) and I don't have to worry?I currently don't load the modules in question - but I'm not sure if it helps, as they might already be powered on by PCI initialization/BIOS (
/sys/bus/pci/../statecats "0").
the real solution is to do it right in the driver.
Using a sysfs "whack behind the drivers back" method is just a disaster waiting to happen; it cannot be done reliable (for example, without knowledge of what you're doing, an application that just whacks all may whack devices that have hw quirks and cannot deal with this, etc etc).
Signature · 2 lines
if you want to mail me at work (you don't), use arjan (at) linux.intel.com Test the interaction between Linux and your BIOS via http://www.linuxfirmwarekit.org