Linux kernel 2.6.38 - major power usage increase

[craig10x]

There is only a little improvement in 2.6.39.

Some laptops are NOT affected by the 2.6.38 power bug. That's the only reason why z06gal can get normal battery life. And my HP Pavilion dv6730ed also has normal battery life after a BIOS update.

Interesting (about doing a BIOS update)....i have a 2 1/2 year old Toshiba Laptop and never tried to update the BIOS...going to their site, there appears to be at least 2 updates for my particular model...Are there any risks in doing these kind of updates? If there are two (posted on different dates) would you do both or the last one shown? Any tips about this? thanks

[MaddawgTL]

I installed the 2.6.35 kernel and have seen some difference. Fan not running constantly,CPU Frequency Scaling Monitor no longer going to the highest setting whenever it wants and seeing an increase of battery life. Only about 10% right now but better than what it was. Now if I can just figure out how to get grub to let me choose which kernel to load up on start up.

[z06gal]

I am now running the 2.6.39.3 kernel and I am not having any heat issues at all. My temps are usually in the 37 to 44 degree range depending upon what I am doing. Battery life is normal unless I turn my Nvidia Optimus graphics card on. Can't complain so far.


Robin

Ah...interesting, Robin, and thanks for posting that update...good to know...i wonder if any changes were made in that version that was the cause of the improvement...
Or perhaps you were just lucky on your particular hardware...

Hi Craig!

This is my temp readings right now:

temp1: +44.0°C (crit = +100.0°C)
temp2: +44.0°C (crit = +100.0°C)

coretemp-isa-0000
Adapter: ISA adapter
Core 0: +45.0°C (high = +86.0°C, crit = +100.0°C)

coretemp-isa-0002
Adapter: ISA adapter
Core 1: +41.0°C (high = +86.0°C, crit = +100.0°C)

coretemp-isa-0004
Adapter: ISA adapter
Core 2: +45.0°C (high = +86.0°C, crit = +100.0°C)

coretemp-isa-0006
Adapter: ISA adapter
Core 3: +38.0°C (high = +86.0°C, crit = +100.0°C)


This is about 5 degrees cooler than the 2.6.39.1 kernel so hopefully they are heading in the right direction.

[craig10x]

wow...nice Robin...you're lucky... i even ran a live disc of Kubuntu 11.10 alpha on my laptop and it was just as bad as ever (that has the 3.0 kernel on it)...i guess my laptop will have to wait for the full fix...
Meanwhile i will have to "hide out" in Mint 10 (with it's older kernel)....

[z06gal]

wow...nice Robin...you're lucky... i even ran a live disc of Kubuntu 11.10 alpha on my laptop and it was just as bad as ever (that has the 3.0 kernel on it)...i guess my laptop will have to wait for the full fix...
Meanwhile i will have to "hide out" in Mint 10 (with it's older kernel)....

I believe that Michael over at Phoronix said that the fix would be in the stable 3.1 kernel and 3.0 stable was released today. It really is mind boggling how some have no issues and others the extreme opposite. I wonder if the lmde folks have experienced this?

[craig10x]

That is soooo true,Robin....i think Michael said that the main problem seems to have to do with whether your power management setting are correctly set up in the bios of the computer...Seems prior to release of kernel .38, even if the kernel did not get an indication that there was a power management available, it would attempt to turn it on anyway and in most computers, it would come on as it is supposed to...

The change made is that now if the kernel doesn't seem to detect, it doesn't even ATTEMPT to turn in on...therefore, power management is not getting turned on in the computers affected and power consumption goes up, fans start running full blast and computers are overheating....

If he is right about that, makes one wonder why it is taking so long to get fixed...

And good question about LMDE...has debian done anything with this? (i know ubuntu is working on it from the bug report at launchpad)...

[craig10x]

Just read this over there...Michael's latest reports are kind of bleak...for those like me who have the problem, i think a true fix could be a very long time in coming (and not even in kernel 3.1 either)..he says several CYCLES that could mean a year or more away...I may just have to consider a new laptop that hopefully will have a properly set power management in the bios...otherwise, i will need to stay on Mint 10 (or older kernels) for a long time to come


For those hoping that the Linux 3.1 kernel will magically fix the power regression issues, it will not. So far I haven't seen any driver patches be pulled in that are setting the ASPM (PCI Express Active State Power Management) bits directly. Even if some drivers do, it will probably be a number of release cycles before there would be better driver coverage for knowing what hardware plays well with ASPM and which devices do not. Or the ASPM kernel code could also be improved to better detect when to actually enable ASPM (i.e. figuring out how Microsoft Windows is doing its ASPM detection), but so far the Linux kernel developers haven't yet figured that out. Nor do any of the changes happen to by chance address the power regressions I haven't yet documented in full.

[ibm450]

folks, stay with 2.6.34 or better still 2.6.32 but nothing greater then .35 as power regressions start on .35 onwards

atm linux kernels have gone pear shaped and will not improve for a very long time

.32 for example is on my motorola atrix mobile phone and on many other latest smart phones. i realise mob phone kernels are different to desktops in certain ways, but, dosnt that short of tell you in advance that older kernels are far more stable then the APLHA - BETA new kernels?.

[sjonesy]

hmm. did anyone actually compile the kernel or just yank it from the repos? Ive 2.6.38 running on various machines for at least 2 months with no issues like that.

[AlbertP]

.35 as power regressions start on .35 onwards

The .35 regression is very little - people that don't measure the battery life with a stopwatch often don't notice anything. No need to worry about this - only some minutes of lost battery life isn't any problem for many people.

[aes2011]

.35 as power regressions start on .35 onwards

The .35 regression is very little - people that don't measure the battery life with a stopwatch often don't notice anything. No need to worry about this - only some minutes of lost battery life isn't any problem for many people.

I guess I'm not having this problem?

Code: Select all

aes@Dell1545 ~ $ inxi -F
System:    Host Dell1545 Kernel 2.6.38-8-generic i686 (32 bit) Distro Linux Mint 11 Katya
CPU:       Dual core Intel Core2 Duo T6400 (-MCP-) cache 2048 KB flags (lm nx sse sse2 sse3 sse4_1 ssse3) bmips 7979.48
           Clock Speeds: (1) 1200.00 MHz (2) 1200.00 MHz
Graphics:  Card Intel Mobile 4 Series Chipset Integrated Graphics Controller X.Org 1.10.1 Res: 1366x768@60.0hz
           GLX Renderer Mesa DRI Mobile Intel® GM45 Express Chipset GEM 20100330 DEVELOPMENT x86/MMX/SSE2 GLX Version 2.1 Mesa 7.10.2 Direct Rendering Yes
Audio:     Card Intel 82801I (ICH9 Family) HD Audio Controller driver HDA Intel BusID: 00:1b.0
           Sound: Advanced Linux Sound Architecture Version 1.0.23
Network:   Card-1 Broadcom BCM4312 802.11b/g LP-PHY driver b43-pci-bridge BusID: 0c:00.0
           Card-2 Marvell 88E8040 PCI-E Fast Ethernet Controller driver sky2 v: 1.28 at port de00 BusID: 09:00.0
Disks:     HDD Total Size: 250.1GB (2.3% used) 1: /dev/sda WDC_WD2500BEVT 250.1GB
Partition: ID:/ size: 84G used: 5.4G (7%) fs: ext4 ID:swap-1 size: 2.11GB used: 0.00GB (0%) fs: swap
Sensors:   System Temperatures: cpu: 49.5C mobo: N/A
           Fan Speeds (in rpm): cpu: N/A
Info:      Processes 149 Uptime 33 min Memory 264.3/1975.4MB Runlevel 2 Client Shell inxi 1.4.23


I know when the fan comes on on my laptop. It's not frequent. In case it matters, room temperature is ~30Celsius.

[AlbertP]

I cannot see from the inxi output if you're affected - at least your graphics is Intel so it's not as much as other people (using (for example) ATI graphics). Just measure your battery life with a stopwatch on both Windows (or an older Linux kernel) and Linux 2.6.38+. Usually your CPU temperature is not that higher - it are the PCI-E devices that consume more power.

[ddaann]

Hi Guys,

I'm having similar problems on the 3.0.0.12 kernel. I ran the inxi -F test and got this output. (Clock speeds are on power saving mode in Jupiter). I don't have the sensors installed, but I can tell you that my laptop runs from around 41c-60c, quite a range for doing everyday tasks. Streaming video makes it the warmest.

I have talked to others over on the blog about downgrading the kernel. As soon as that word kernel comes up I am at a loss- I am unknowledgeable in that territory. What is the news on this power efficiency issue? How would I go about doing so? Is there anyway of discovering the optimal kernel to use with my particular machine or is it a case of trial and error?

I have sandybridge which phoronix has mentioned is at issue....when I was running mint 11 I thought I was running nvidia...maybe a cause is there? (I'm not even sure if they perform the same function, of course...).

Can anyone shed any light on this?

System: Host daniel-ThinkPad-X121e Kernel 3.0.0-12-generic x86_64 (64 bit) Desktop Gnome Distro Linux Mint 12 Lisa
Machine: System LENOVO (portable) product 3045CTO version ThinkPad X121e
Mobo LENOVO model 3045CTO Bios LENOVO version 8QET23WW (1.07 ) date 06/16/2011
CPU: Dual core Intel Core i3-2357M CPU (-HT-MCP-) cache 3072 KB flags (lm nx sse sse2 sse3 sse4_1 sse4_2 ssse3 vmx)
Clock Speeds: (1) 800.00 MHz (2) 800.00 MHz (3) 800.00 MHz (4) 800.00 MHz
Graphics: Card: Intel 2nd Generation Core Processor Family Integrated Graphics Controller
X.Org 1.10.4 drivers intel unloaded: vesa,fbdev Resolution 1366x768@60.0hz
GLX Renderer Mesa DRI Intel Sandybridge Mobile GLX Version 2.1 Mesa 7.11
Audio: Card Intel 6 Series/C200 Series Chipset Family High Definition Audio Controller driver HDA Intel
Sound: Advanced Linux Sound Architecture v: 1.0.24
Network: Card-1 Atheros AR8151 v2.0 Gigabit Ethernet driver atl1c
IF: eth0 state: down speed: 4294967295 Mbps duplex: full mac: e8:9a:8f:97:5a:04
Card-2 Realtek RTL8188CE 802.11b/g/n WiFi Adapter driver rtl8192ce
IF: wlan0 state: up mac: d0:df:9a:4a:98:12
Drives: HDD Total Size: 320.1GB (6.8% used) 1: /dev/sda HITACHI_HTS72323 320.1GB
Partition: ID:/ size: 291G used: 21G (8%) fs: ext4 ID:swap-1 size: 3.13GB used: 0.00GB (0%) fs: swap
Sensors: Error: You do not have the sensors app installed.
Info: Processes 165 Uptime 39 min Memory 1061.2/2934.1MB Client Shell inxi

After the initial scare of power consumption and potential increased heat (which I'm not even so sure of as I wasn't that aware of recorded temperatures before installing 12rc) I am becoming more comfortable with running a machine like this. The temperatures are mostly average by most accounts...but of course, Linux is a learning process and I can always make my machine run better so I remain interested and slightly concerned...

[ibm450]

My laptop stays at 36 degrees and fan is whisper quiet with w7

My laptop goes up to 55+ with m11 and even more on m12 and the fan goes ballistic while on linux.

I get about 6 hrs (battery) on w7 and under 2.5hrs with m11 & even less with m12.

so really i cant see this being, normal, as i get far superior battery life and also far, far, far less heat issues in win7

running lower end / older linux kernels did not improve the issues im having including the post GRUB hacks / tweaks by adding acpi=force (sorry cant remember what it was called exactly)

hmmm .... unfortunately, i may have to dust off and say "helllo there Mr Win 7.." .... its a shame really, at one stage it was looking very promising for linux but it went all pear shaped after 11.04

[wayne128]

My laptop stays at 36 degrees and fan is whisper quiet with w7

My laptop goes up to 55+ with m11 and even more on m12 and the fan goes ballistic while on linux.

Similar experience some time ago on Asus laptop testing

running Mint 10 for hour with 47 deg c.
Mint 11, about 60 degc, constant hot air blowing from start up
win 7, run too cool, little hot air blowing out...

[Anakinholland]

Touched on this subject before, thought I'd make a post here as well.

viewtopic.php?f=49&t=84877

Basically, there's a bug in some kernel-versions, 2.6.37 and up, and 3.x

My last post in that topic links to this article, which states that the problem is identified and should be solved, but maybe not soon. Kernel 3.2 has been feature-frozen, so likely will not be implemented till 3.3.

There's also mention of a kernel-patch which should remove the issue as well.

Good luck!

Anakin