[Bug 911325] Re: pm-utils: additonal power.d scripts to save power

[Bug Watch Updater]

Launchpad has imported 2 comments from the remote bug at
https://bugs.freedesktop.org/show_bug.cgi?id=44672.

If you reply to an imported comment from within Launchpad, your comment
will be sent to the remote bug automatically. Read more about
Launchpad's inter-bugtracker facilities at
https://help.launchpad.net/InterBugTracking.

------------------------------------------------------------------------
On 2012-01-11T12:55:59+00:00 Martin Pitt wrote:

Created attachment 55424
power.d script for autosuspending USB bluetooth devices

>From https://launchpad.net/bugs/911325:

Colin King spent some days doing extensive and accurate power benchmarking [1],
amongst others the effect of the low-power modes of various hardware.

Here are two scripts which handle USB bluetooth adapters and a subset of
PCI devices which are safe to turn into low-power mode.

Typical savings: 0.38 W per machine with some machines seeing > 1-2W
savings.

[1] http://zinc.canonical.com/~cking/power-benchmarking/

Reply at: https://bugs.launchpad.net/pm-utils/+bug/911325/comments/3

------------------------------------------------------------------------
On 2012-01-11T12:57:01+00:00 Martin Pitt wrote:

Created attachment 55425
power.d script for autosuspending safe subset of PCI devices

Reply at: https://bugs.launchpad.net/pm-utils/+bug/911325/comments/4


** Changed in: pm-utils
       Status: Unknown => Confirmed

** Changed in: pm-utils
   Importance: Unknown => Wishlist

-- 
You received this bug notification because you are a member of Ubuntu
Sponsors Team, which is subscribed to the bug report.
https://bugs.launchpad.net/bugs/911325

Title:
  pm-utils: additonal power.d scripts to save power

Status in pm-utils:
  Confirmed
Status in “pm-utils” package in Ubuntu:
  New

Bug description:
  The kernel allows us to save power on a range of devices - currently
  pm-utils power.d does not make full advantage of these power savings.

  Tools like PowerTop provide advice on which devices can be set into a
  "Good" power management state.  Analysis on a selection of
  representative modern mobile x86 based platforms (e.g. netbook,
  laptop) has shown that a subset of devices can be safely put into a
  power saving state.

  Each of the "Good/Bad" PowerTop device recommendations were measured
  accurately using a 6 digit digital multimeter. Each measurement was
  based on measuring current drain over 5 x 60 second periods on the
  Good and Bad settings and seeing which device shows the most promising
  power saving.  The resulting data can be found here:

  http://zinc.canonical.com/~cking/power-benchmarking/powertop-good-bad-
  recommendations/powertop-good-bad-recommendations.ods

  And the write-up of the methodology and conclusions can be found here:

  http://zinc.canonical.com/~cking/power-benchmarking/powertop-good-bad-
  recommendations/results.txt

  From this analysis we can draw a conclusion that the following devices
  behave well to power savings as follows:

  	* Webcam
  	* Audio
  	* DRAM
  	* Ethernet
  	* Wifi
  	* Bluetooth
  	* SATA link
  	* MMC/SD 

  From this, two pm-utils power.d scripts were written to control the
  power state of the above devices.  The script pci_devices controls the
  PCI specific devices and usb_bluetooth controls USB webcams into and
  out of power saving modes.

  These were then tested by community - see:
  https://wiki.ubuntu.com/Kernel/PowerManagementPMUtils

  Power savings were calculated from the ACPI battery information.  The
  characteristics and reliability of ACPI battery data to measure power
  consumption has been studied here: http://zinc.canonical.com/~cking
  /power-benchmarking/acpi-battery-results/results.txt - the test
  scripts used by the community testing use a battery power measuring
  tool that has been calibrated and checked using a 6 digit multimeter
  and calculates the average power consumption and standard deviation
  (to allow us to discriminate against unreliable samples).

  From these results, I discarded data where the standard deviation was
  rather high and also results where the savings or losses were smaller
  than the standard deviation.  This left me with a set of results where
  we have conclusive data showing us power savings or losses with these
  new scripts.   The conclusion is that the vast majority of machines
  save power.   Also, users tested their machines to see if the devices
  behaved correctly.

  On the samples of machines I have (Dell Inspirion 6400, Lenovo X220i,
  HP Mini 210 + HP Mini 10) I see no regressions. I therefore am
  requesting that the two new pm-utils/power.d scripts are added to pm-
  utils.

  Typical savings: 0.38 W per machine with some machines seeing > 1-2W
  savings.

  Attached to this bug report are the two new scripts for pm-
  utils/power.d

To manage notifications about this bug go to:
https://bugs.launchpad.net/pm-utils/+bug/911325/+subscriptions