PowerManagementRC6

Testing is Closed

The RC6 call for testing is now officially closed. We appreciate all of the testing feedback we received. It had a direct impact and influence on our decision to enable plain RC6 by default for Precise. Between the testing feedback we received and interaction with the upstream developers, we were also able to find and fix a handful of RC6 related issues. Kudos to all involved.

RC6

RC6 is a technology which allows the GPU to go into a very low power consumption state when the GPU is idle (down to 0V). It results in considerable power savings when this stage is activated. When comparing under idle loads with machine state where RC6 is disabled, improved power usage of around 40-60% has been witnessed.

As a bonus, when those states are reached, there are additional thermal and power space for both CPU and GPU to go into a more aggressive turbo mode. This can also improve the performance of intensive workloads by around 10%, and a bit more in some cases.

The above was taken from an indepth summary regarding RC6 posted by Eugeni Dodonov.

Ubuntu RC6 Status

Up until recently, RC6 was disabled by default for Sandy Bridge systems due to reports of hangs and graphics corruption issues when RC6 was enabled. Intel has now asserted that RC6p (deep RC6) is responsible for the RC6 related issues on Sandy Bridge. Patches were therefore submitted upstream to disable RC6p on Sandy Bridge:

In an effort provide more exposure and testing for these patches, the Ubuntu kernel team has applied them to 3.2.0-17.27 and newer Ubuntu 12.04 Precise Pangolin kernels. We have additionally enabled plain RC6 by default on Sandy Bridge systems so that users can benefit from the improved power savings by default.

How to Test

If you are running Ubuntu 12.04 (Precise Pangolin), we would like to capture any feedback regarding improved power savings for Sandy Bridge users witness while running a 3.2.0-17.27 or newer kernel. More importantly, we want to know of any issues which arise as a result of having RC6 enabled. To test, please follow the instructions below:

  1. Boot into 3.2.0-17.27 or newer Ubuntu 12.04 Precise Pangolin kernel
  2. Gather System Info (BIOS info, kernel version, VGA info)
    • sudo dmidecode -t 0 | grep "Version:"
    • cat /proc/version_signature
    • lspci -nn | grep "VGA"
  3. Install power measuring tools
    • sudo add-apt-repository ppa:colin-king/powermanagement
    • sudo apt-get update
    • sudo apt-get install powerstat
  4. Unplug the power from your laptop/netbook and run the following command:
    • sudo powerstat
  5. Note the average power consumption and standard deviation which are shown at the end of the powerstat run
  6. If you experience any issues due to RC6 being enabled, please file a bug:
    • ubuntu-bug linux
  7. Reboot into the same kernel but disable RC6. To disable RC6, do the following:
    • Hold down Left-Shift key to enter the Grub menu on boot
    • Hit 'e' to edit the kernel command line
    • Append i915.i915_enable_rc6=0 as a kernel boot parameter

      • For example: linux /boot/vmlinuz-3.2.0-17-generic <...> quiet splash vt.handoff=7 i915.i915_enable_rc6=0

    • Press Ctrl-x to boot
  8. Unplug the power from your laptop/netbook and run powerstat again to capture comparative test results with RC6 disabled
    • sudo powerstat
  9. Note the average power consumption and standard deviation (σ) which as shown at the end of the powerstat run
  10. Add test results to the Crowd Sourcing Test Results section below

Extended Testing

We would appreciate any extended testing with RC6 enabled. For example, exercising graphics doing normal day to day activities. We are keen to hear about any issues such as hangs or graphics corruptions which are triggered with RC6 enabled. If you experience such issues, please file a bug by running 'ubuntu-bug linux' from a command line.

Issues? File a Bug

If you do experience any issues such as hangs or graphics corruptions due to RC6 being enabled, please let us know by filing a bug. To file a bug, run the following command:

  • ubuntu-bug linux

Please be sure to file a bug and note it in the "Issues" column of the Crowd Sourcing Test Results. Otherwise, we will not know how to contact you in order to help resolve your issue.

Crowd Sourcing Test Results

Please enter the results in the table below sorted alphanumerically on the machine name. Thanks!

Machine

BIOS Info

Kernel Version

VGA Info

powerstat output (RC6 Enabled)

powerstat output (RC6 Disabled)

Issues (Bug:#####)

Acer Aspire 4752

V2.09

3.2.0-17-generic

8086:0116 (rev 09)

9.52 W, σ=0.24

18.31 W σ=0.25

Kind of bug 935965 see #11 (Resolved now)

Apple MacBook Air 4,1

MBA41.88Z.0077.B0E.1110141154

3.2.0-17-generic

8086:0116 (rev 09)

7.92 W, σ=0.23

13.87 W, σ=0.29

None

Apple MacBook Air 4,2

MBA41.88Z.0077.B0E.1110141154

3.2.0-17-generic X86_64

8086:0116 (rev 09)

8.86 W, σ=0.13

11.27 W σ=0.62

None

Apple MacBook Pro 8,1

MBP81.88Z.0047.B26.1110311252

3.2.0-18.29-generic-pae 3.2.9

8086:0126 (rev 09)

8.03W, σ=0.08

13.05W, σ=0.16

none

ASUS N53SV

N53SV.214

3.2.0-17-generic

8086:0116 (rev 09) + 10de:0df4 (rev a1)

27.03 W, σ=2.29

36.54W, σ=0.99

None

ASUS N55SF

N55SF.204

3.2.0-17.27-generic 3.2.6

[8086:0116] (rev 09) + [10de:1247] (rev a1)

28.14 W, σ=0.12

35.42 W, σ=0.73

None

ASUS UX21E

UX21E.209

3.2.5-tunerc6v1-generic

8086:0116 (rev 09)

7.16 W, σ=0.86

12.22 W, σ=0.63

none

ASUS U30Sd

U30Sd.211

Ubuntu 3.2.0-17.27-generic 3.2.6

[8086:0116] (rev 09)

9.47 W σ=0.71

16.12 W σ=0.39

Bumblebee installed and running (means discret card was turned off during tests)

ASUS UX31E

UX31E.211

3.2.0-17-generic

8086:0116 (rev 09)

5.50 W,σ=0.40

6.21 W, σ=0.77

Bug #937378 hard shutdown -> Fix Released with .27

ASUS UX31E

UX31E.210

Ubuntu 3.2.0-17.27-generic

[8086:0116] (rev 09)

7.96 W σ=0.04

11.01 W σ=0.02

ASUS U36SD

U36SD.206

3.2.0-17-generic

8086:0116 (rev 09)

7.84 W σ=0.02

10.97 W σ=0.06

None

ASUS X53SV

K53SV.210

3.2.0-19.30-generic

8086:0116 (rev 09) + 10de:0df4 (rev ff)

17.88 W, σ=2.06

32.25W, σ=1.09

Without Ironhide installed

Dell Vostro 3450

A07

3.2.0-17-generic

8086:0116 (rev 09)

11.62 W, σ=0.18

17.64 W σ=0.38

None

Dell Inspiron N5100

A02

3.2.0-17-generic

8086:0116 (rev 09)

20.19 W, σ=0.44

24.98 W σ=0.47

HP Pavilion dm4-2120ss

F.12

3.2.0-18.28-generic 3.2.9

8086:0116 (rev 09)

9.04 W, σ=0.05

14.16 W σ=1.66

None

HP Pavilion Dv6-6030el

F.1A

3.2.0-19.30-generic 3.2.11

[8086:0116] (rev 09) + [1002:6760]

16.55W, σ=0.43

20.08W, σ=1.17

None

Lenovo Edge E420

8HET40WW(1.22)

3.2.0-17-generic-pae

8086:0126 (rev 09)

13.43 W, σ=0.36

19.92 W σ=0.27

None

Lenovo L520

8GET38WW (1.15)

3.2.0-17-generic

8086:0126 (rev 09)

12.81 W, σ=0.55

28.27 W σ=2.73

None

Lenovo T420s

8CET50WW (1.30)

3.2.0-17-generic x86_64

8086:0126 (rev 09)

8.91 W, σ=0.05

17.60 W, σ=0.07

rc6 enabled manually for a long time; occasionally got the whole screen black with 1/4 of it blue (on the right), forcing me to restart, but not sure I can blame that on rc6

Lenovo T420

83ET63WW (1.33 )

3.2.0-17-generic x86_64

[8086:0116] (rev 09)

8.91 W, σ=0.05

13.17W, σ=0.31

Lenovo U300s Ideapad

56CN38WW

3.2.0-17-generic

8086:0016 (rev 09)

14.87 W σ=0.91

19.84 W σ=1.12

None

Lenovo X201

6QET68WW (1.38)

3.2.0-17-generic

8086:0046 (rev 02)

6.40 W σ=0.17

6.54 W σ=0.70

RC6 manually enabled by kernel flag. May not have worked (pre-Sandy Bridge)

Lenovo X220

8DET56WW (1.26)

3.2.0-17-generic

8086:0126 (rev 09)

7.95 W, σ=0.10

15.95 W, σ=0.08

Lenovo X220

8DET54WW (1.24)

3.2.0-17.26

8086:0126 (rev 09)

6.66 W, σ=0.18

12.09 W, σ=0.16

Minor graphic issues, e.g. occasionally leaving the topmost window unrefreshed.

Lenovo X220

8DET54WW (1.24)

3.2.0-17.27

8086:0126 (rev 09)

6.67 W, σ=0.11

12.09 W, σ=0.16

All graphic issues have disappeared. Works perfectly.

Lenovo X220

8DET55WW (1.25 )

3.2.0-17-generic

8086:0126 (rev 09)

7.93 W, σ=0.77

14.80 W, σ=0.45

Lenovo X220

8DET56WW (1.26)

3.2.0-17-generic-pae

8086:0126 (rev 09)

9.16 W, σ=0.11

13.52 W, σ=0.08

Lenovo X220i

8DET54WW (1.24)

3.2.0-17-generic X86_64

8086:0116 (rev 09)

8.60 W, σ=0.04

9.74 W, σ=0.11

none

Lenovo X220i

8DET54WW (1.24)

3.2.0-17-generic #27-Ubuntu x86_64

8086:0116 (rev 09)

8.84 W, σ=0.03

17.06 W, σ=0.08

none (re-test)

Lenovo X220t

8DET56WW (1.26 )

3.2.0-17-generic-pae

8086:0126 (rev 09)

7.99 W, σ=0.12

17.66 W, σ=0.37

Lenovo X220T (with SSD)

8DET56WW (1.26 )

3.2.0-17.27-generic

8086:0126 (rev 09)

8.07 W, σ=0.24

13.45 W, σ=0.67

Samsung NP300V3A

03FI

3.2.0-17-generic X86_64 (in Oneiric)

8086:0116 (rev 09)

8.18 W, σ=0.72

11.85 W, σ=2.36

On an up to date Precise – none

Samsung NP350U2A

05PY.M020.20120105.WZW

3.2.0-17-generic

8086:0116 (rev 09)

7.50 W, σ=0.36

11.52 W σ=0.82

Bug 935965 (Resolved in 3.2.0-17.27)

Samsung NP700Z5B-W01UB

09/06/2011 (03FD)

3.2.0-17-generic X86_64

8086:0116 (rev 09)

not tested

not tested

Severe graphical glitches (tearing) Bug 935965 (Resolved in 3.2.0-17.27)

Samsung NP900X3BA01US

P03AAH

3.2.0-18-generic x86_64

8086:0116 (rev 09)

8.46, σ=0.20

7.69 W, σ=0.29

none

Sony VAIO Z (VPCZ21AJ)

R0170H5 06/16/2011

3.3.0-999-generic X86_64 (Daily-Build)

8086:0116 (rev 09)

9.09W, σ=0.14

15.45W, σ=0.44

none

System76 GazP6

4.6.4

3.2.0-17.27-generic

[10de:0dd1] (rev a1)

33.52W, σ=0.26

35.79W, σ=0.55

none

Toshiba Satellite R830 1DG

3.60

3.2.0-17-generic X86_64

8086:0126 (rev 09)

11.13W, σ=1.75

16.18W, σ=0.42

none

Dell Vostro 3450

A07

3.2.0-17.27-generic 3.2.6

[8086:0116] (rev 09)

26.54W, σ=2.74

36.34W, σ=3.58

none

Dell Latitude E6420

A06

3.2.0-18.29-generic 3.2.9

[8086:0126] (rev 09)

10.36W, σ=0.30

16.17W, σ=0.16

none

CategoryTestingReport

  • oddly the power was greater for me with it enabled on a new samsung series 9 -Ram

Kernel/PowerManagementRC6 (last edited 2012-04-11 20:19:10 by leannogasawara)