Power Usage Regressions in 2.6.38/.39

admiralspark · 2011-04-23 18:43:35

See the Phoronix writeup here: http://www.phoronix.com/scan.php?page=a … ffda&num=4
It looks like for some reason 2.6.38 and up are having major power usage issues. Is anybody else experiencing this? This alone is apparently a 13% increase in power consumption at idle, which would make me lose almost an hour in battery life on my netbook. Thoughts/comments?

Last edited by admiralspark (2011-04-23 18:44:01)

demian · 2011-04-23 21:50:14

Yeah, there seems to be an increase in power consumption since .38. For me it's 0.5 - 1W (in most situations 7 to 20 % less battery life).

A fellow archer also blogged about it (itgen.blogspot.com) and made this graph:

This one is me comparing .37 to .38 when idling:

mrmylanman · 2011-04-24 15:29:16

I read the article on Phoronix as well, and it's interesting because my laptop is consistently getting as good (or better) battery life as what it got in Windows. I honestly haven't spent much time using 2.6.37, since I just got my laptop not long ago, but with getting 6-7 hours now, I can't wait to see what I'll get when/if they fix this power draw regression.

hrobeers · 2011-04-25 08:38:26

mrmylanman wrote:

I read the article on Phoronix as well, and it's interesting because my laptop is consistently getting as good (or better) battery life as what it got in Windows. I honestly haven't spent much time using 2.6.37, since I just got my laptop not long ago, but with getting 6-7 hours now, I can't wait to see what I'll get when/if they fix this power draw regression.

Install 2.6.37 and you can see it today! (or install the LTS kernel, and you can switch between 2.6.32 and 2.6.38)

muffinss · 2011-04-26 00:41:01

I've noticed the power usage as well with my laptop. Before I was getting 4 hours in a-half of battery power, but now, its about 3 in a-half - 4 hours =/

mrmylanman · 2011-04-26 01:47:38

hrobeers wrote:

mrmylanman wrote:
I read the article on Phoronix as well, and it's interesting because my laptop is consistently getting as good (or better) battery life as what it got in Windows. I honestly haven't spent much time using 2.6.37, since I just got my laptop not long ago, but with getting 6-7 hours now, I can't wait to see what I'll get when/if they fix this power draw regression.
Install 2.6.37 and you can see it today! (or install the LTS kernel, and you can switch between 2.6.32 and 2.6.38)

Well my laptop's GPU is not supported well until 2.6.38 and the closed-source drivers don't work with xorg-server 1.10 last time I checked so I figured it wasn't really worth the hassle (I don't have 1.9 in my pacman cache).

Thanks for the tip though! New hardware is sometimes an issue with open-source software, lol.

snack · 2011-04-26 10:39:59

A new, even worse power regression has been found in 2.6.35:

http://www.phoronix.com/scan.php?page=a … ess2&num=1

pataphysician · 2011-04-26 12:50:39

snack wrote:

A new, even worse power regression has been found in 2.6.35:

I was going to post my memory that 2.6.37 seemed like it brought a huge power performance increase, and that I did notice a power performance drop with 2.6.38, but the drop was much less than the increase I saw with 37.

Since I started with 2.6.35 on this laptop, it all makes sense now, though even the horrible 2.6.35 had the same battery life for me as windows 7(clean install, so no manufacturer crapware eating power) in multiple tests.

heatblazer · 2011-10-20 04:20:05

Hello, I am having the same issue. I am using asus eeepc 904HD, and arch linux. Tested laptop_mode and now using power manager under xfce4, the battery lives under hour or even worse, around 40-35 minutes. Month ago it lasted for 1 h and 20-30 min.

ethail · 2011-10-20 07:20:28

heatblazer: That's for 2.6.39, we live in 3.0.6.

try searching the bbs for more info and solutions about that power regressions.

heatblazer · 2011-10-20 07:23:30

ethail wrote:

heatblazer: That's for 2.6.39, we live in 3.0.6.
try searching the bbs for more info and solutions about that power regressions.

3.0-ARCH #1 SMP PREEMPT Fri Oct 7 10:11:05 UTC 2011 i686 Intel(R) Celeron(R) M processor 900MHz GenuineIntel GNU/Linux

It`s still a power eater... Any info?

pogeymanz · 2011-10-20 11:49:30

If you have intel hardware, there are a few articles at phoronix.com about the power regressions and possible workarounds for some of it.

Even with those fixes, my computer uses more power than it did in the 2.6.34/35 days.

kokoko3k · 2011-10-20 12:38:03

My experience with a 2 years old (happy birthday!) eeepc 1005ha monitored with ibam says that the estimated battery life is just 5% under the the limit claimed by asus in idling mode with wifi on and backlight to 25%.
This with linux 3.0.4-pf, (adding pcie_aspm=force, i915.i915_enable_rc6=1)and all the suggestions by powertop.
http://phoronix.com/forums/showthread.p … post227549

Last edited by kokoko3k (2011-10-20 12:46:15)

Arch Linux

#1 2011-04-23 18:43:35

Power Usage Regressions in 2.6.38/.39

#2 2011-04-23 21:50:14

Re: Power Usage Regressions in 2.6.38/.39

#3 2011-04-24 15:29:16

Re: Power Usage Regressions in 2.6.38/.39

#4 2011-04-25 08:38:26

Re: Power Usage Regressions in 2.6.38/.39

#5 2011-04-26 00:41:01

Re: Power Usage Regressions in 2.6.38/.39

#6 2011-04-26 01:47:38

Re: Power Usage Regressions in 2.6.38/.39

#7 2011-04-26 10:39:59

Re: Power Usage Regressions in 2.6.38/.39

#8 2011-04-26 12:50:39

Re: Power Usage Regressions in 2.6.38/.39

#9 2011-10-20 04:20:05

Re: Power Usage Regressions in 2.6.38/.39

#10 2011-10-20 07:20:28

Re: Power Usage Regressions in 2.6.38/.39

#11 2011-10-20 07:23:30

Re: Power Usage Regressions in 2.6.38/.39

#12 2011-10-20 11:49:30

Re: Power Usage Regressions in 2.6.38/.39

#13 2011-10-20 12:38:03

Re: Power Usage Regressions in 2.6.38/.39

Board footer