https://bugzilla.novell.com/show_bug.cgi?id=717454 https://bugzilla.novell.com/show_bug.cgi?id=717454#c1 --- Comment #1 from Richard Brown 2011-09-14 21:23:02 UTC --- Created an attachment (id=450846) --> (http://bugzilla.novell.com/attachment.cgi?id=450846) hwinfo of another Thinkpad X220 with identical symptoms Attaching a hwinfo output from another Thinkpad X220 with an i7 CPU that exhibits identical power consumption issues. eg 11W in kernel 2.6 and over 17W in kernel 3.1 -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=717454 https://bugzilla.novell.com/show_bug.cgi?id=717454#c3 Greg Kroah-Hartman changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |NEEDINFO InfoProvider| |jlee@suse.com --- Comment #3 from Greg Kroah-Hartman 2011-09-16 09:56:56 UTC --- Joey, can you dig around in your lab and see if you have a Sandybridge machine to see if you can duplicate this? If not, let me know and I'll push Intel to get me a sample box to track this down and fix it. Joey, also feel free to add anyone else to this bug that you think can help out. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=717454 https://bugzilla.novell.com/show_bug.cgi?id=717454#c5 --- Comment #5 from Greg Kroah-Hartman 2011-09-16 14:21:12 UTC --- (In reply to comment #4)

Is there a good simple way of monitoring power consumption in desktops/servers? Powertop does not show consumption unless a battery is discharging but I am interested in seeing if this bug occurs in non-laptop hardware

I use a "Save-a-watt" device that plugs in between the power code and the power plug and it shows the power being used by the system. But it's pretty "crude". In order to do real power testing, you need a fancy scope and a bunch of wires on the battery of a laptop to really determine exactly how much power is being drawn by the system as powertop lies a lot when you are trying to tune things down to the sub-watt level. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=717454 https://bugzilla.novell.com/show_bug.cgi?id=717454#c7 --- Comment #7 from Richard Brown 2011-09-16 17:57:02 UTC --- Created an attachment (id=451468) --> (http://bugzilla.novell.com/attachment.cgi?id=451468) hwinfo of Gulftown desktop I've bought a "Save-a-watt" like device and tested another one of my machines, a desktop with a Gulftown CPU (i7 980x) These tests are showing the problem also occurs on this hardware. After a straight boot, not logging in, kernel 2.6.38 draws 134W according to my meter Booting the identical install into kernel 3.0 from Tumbleweed draws 140W, even with the video card uninitialised (no nvidia prop. driver) hwinfo of this Gulftown machine attached. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=717454 https://bugzilla.novell.com/show_bug.cgi?id=717454#c9 Richard Brown changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |NEW InfoProvider|rdb@ccb.ac.uk | --- Comment #9 from Richard Brown 2011-09-17 09:51:18 UTC --- On my X220 (Sandybridge CPU) with 2.6 (Stock oS 11.4 kernel) running powertop in runlevel 1, I see power consumption as previously reported (11W) with the CPU spending almost all of its time in C4 state (99-101% according to powertop) Changing to kernel 3.0 (Tumbleweed version) running powertop in rl 1, power consumption is as previously reported (17W). powertop reports that the CPU is spending just as much of its time in C4 state (99-101%) Looks like the power drain is coming from something other than the CPU, or powertop is lying. Will be doing these same tests on my Gulftown hardware later today when I'm back at home -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=717454 https://bugzilla.novell.com/show_bug.cgi?id=717454#c11 --- Comment #11 from Richard Brown 2011-09-17 20:14:31 UTC --- As CPU doesn't appear to be related to the issue I've been investigating other potential causes. Phoronix have lots of articles on recent Linux kernel power issues (eg: http://www.phoronix.com/scan.php?page=article&item=linux_2638_aspm&num=2 http://www.phoronix.com/scan.php?page=news_item&px=OTYwNA ) Based on the claims of the article, I have attempted booting up kernel 3.0 with pcie_aspm=force This has seen a small (~1W) improvement in 3.0 power consumption on both of my machines, which might be a good start but clearly not the root cause. HTH -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=717454 https://bugzilla.novell.com/show_bug.cgi?id=717454#c13 --- Comment #13 from Joey Lee 2011-09-26 10:49:05 UTC --- I got a SandyBridge notebook in Taipei, now. Will also apply Richard and Andrew's suggestion then verify the power status by powertop. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=717454 https://bugzilla.novell.com/show_bug.cgi?id=717454#c15 --- Comment #15 from Joey Lee 2011-09-28 03:01:44 UTC --- I cann't reproduce issue by using a HP machine: HP openSUSE 11.4 2.6.37.1-1.2-desktop i586 kernel battery discharge brightness level one wifi off Desktop 3D effect enabled powertop 11.3 boot after 10 minues: 7.5 W long term: 10.8W HP openSUSE 11.4 3.0.4-43.3-desktop i586 kernel (openSUSE:Tumbleweed) battery discharge brightness lowest level wifi off Desktop 3D effect enabled powertop 11.3 boot after 10 minutes: 8.5 W long term: 10.3W AC Plug-in power monitor(removed battery): 12 W Didn't have big difference between 2.6.37.1-1 and 3.0.4 kernel. The HP machine's VGA is: 8086:0116 -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=717454 https://bugzilla.novell.com/show_bug.cgi?id=717454#c17 Jon Nelson changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |jnelson-suse@jamponi.net --- Comment #17 from Jon Nelson 2011-10-12 19:09:33 UTC --- This may help. I have a ThinkPad T520i. I placed the following in /etc/modprobe.d/99-local.conf and rebuilt the initrd (between '====' lines): ============= blacklist firewire_ohci blacklist mei options i915 powersave=1 i915_enable_rc6=1 ============= (I don't use firewire). I've also sometimes used i915_enable_fbc=1 but I'm not sure that helps or not. Furthermore, in /etc/modprobe.d/50-sound.conf I have: ============= options snd-hda-intel enable_msi=1 model=thinkpad power_save=30 power_save_controller=1 ============= and that helps quite a bit, too. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=717454 https://bugzilla.novell.com/show_bug.cgi?id=717454#c19 Bruno Friedmann changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |bruno@ioda-net.ch --- Comment #19 from Bruno Friedmann 2011-10-23 09:19:58 UTC --- Just a guess here Did someone check if the content of /etc/modprobe.d/50-thinkpad_acpi.conf is correct for the x220 model ? which come with acpid rpm package # # IBM/Lenovo ThinkPad ACPI driver options # # A detailed description of the parameters for the ThinkPad ACPI driver can be found # in /usr/src/linux/Documentation/laptops/thinkpad-acpi.txt which is part of the package # 'kernel-source'. # # If you encounter problems with the hotkey mask please file a bug on # http://bugzilla.novell.com/ # options thinkpad_acpi experimental=1 hotkey=0xffffff -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=717454 https://bugzilla.novell.com/show_bug.cgi?id=717454#c21 Greg Kroah-Hartman changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |NEEDINFO InfoProvider| |awafaa@opensuse.org --- Comment #21 from Greg Kroah-Hartman 2011-10-23 12:58:22 UTC --- Thanks, yes, that answers my question. So, with those options enabled, we are back to the old power consumption levels, right? So there's nothing else we can do here. We can't enable those options "by default", otherwise the kernel would have done so, as some hardware can't handle them, while others, like yours, needs them due to various integration issues (BIOS/motherboard/etc.) I still would like to have Andrew try this out on his platform, for verification that this does resolve the issue he sees. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=717454 https://bugzilla.novell.com/show_bug.cgi?id=717454#c23 Andrew Wafaa changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |NEW InfoProvider|awafaa@opensuse.org | --- Comment #23 from Andrew Wafaa 2011-10-24 08:51:56 UTC --- I've managed to run the same tests as Richard and my results are as follows based on my system which is running 11.4+GNOME3+Kernel 3.0.7: X220 powertop recorded idle power consumption with full screen brightness, wifi enabled and no i915 parameters ~21.8W General use (mouse movements, browser opening, etc) sees consumption spike upto ~40.1W X220 powertop recorded idle power consumption with full screen brightness, wifi enabled and i915.i915_enable_rc6=1 ~14.9W General use (mouse movements, browser opening, etc) sees consumption spike upto ~19.7W X220 powertop recorded idle power consumption with full screen brightness, wifi enabled and i915.i915_enable_rc6=1 i915.powersave=1 i915.i915_enable_fbc=1 i915.lvds_downclock=1 ~13.8W General use (mouse movements, browser opening, etc) sees consumption spike upto ~18.8W -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=717454 https://bugzilla.novell.com/show_bug.cgi?id=717454#c25 Andrew Wafaa changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |NEW InfoProvider|awafaa@opensuse.org | --- Comment #25 from Andrew Wafaa 2011-10-24 10:22:35 UTC --- Whoops - that was a rather key omission on my part, sorry. Booting into the 2.6.37 kernel I have the following: X220 powertop recorded idle power consumption with full screen brightness, wifi enabled and no i915 parameters ~13.2W General use (mouse movements, browser opening, etc) sees consumption spike upto ~18.0W These are with no kernel parameter passed at all and as you can see they are still better than with an optimised 3.0.7 kernel :-( Out of interest, would there be much difference in power consumption between -desktop and -default? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=717454 https://bugzilla.novell.com/show_bug.cgi?id=717454#c27 Greg Kroah-Hartman changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #27 from Greg Kroah-Hartman 2011-12-05 20:05:53 UTC --- There are plans in the future to hopefully integrate some automatic detection here, but it's not in the 12.1 timeframe as that's already shipping sorry. I'll close this for now, as the base problem is solved. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.