x230 battery life --> powertop --> rabbit trail

[tarverator]

The problem I am trying to solve is that my Thinkpad x230 i5-3360M running Linux Mint Mate Maya 32-bit has only been getting about 3 hours of battery life, whereas it was getting about twice that a few months ago.

I was recently checking my power consumption with powertop, and it was about 9 or 10 watts. My impression is that just sitting around with the screen not too bright it should be around 5 watts, but I wasn't monitoring this very closely until recently.

I have tried (and currently am booting with):

Code: Select all

GRUB_CMDLINE_LINUX_DEFAULT="quiet splash i915.i915_enable_rc6=1 i915.semaphores=1 pcie_aspm=force"

...as per http://www.thinkwiki.org/wiki/Intel_HD_Graphics even though the pcie_aspm force flag may have no effect on v3 kernels (not sure, but my impression is that it can't hurt).

I upgraded to a recent kernel as per http://www.upubuntu.com/2013/02/install ... 76-in.html.

At some point very recently (after the kernel update?) powertop (v1.97) stopped showing wattage. Odd, but maybe powertop is out of date. So I tried version 2.2 from here: https://01.org/powertop/downloads but it won't configure because it complains that it can't find ncurses. Ok, so I found a patch here https://lists.01.org/pipermail/powertop ... 00200.html that is supposed to eliminate the ncurses dependency (which is also odd because it looks like ncurses is available in Mint, but not in the flavour that powertop is looking for). Anyway, the patch didn't apply properly.

I think I may be going down a rabbit trail here.

How can I fix or replace powertop so I can track my power consumption? Better yet, how do I restore the solid six (or more!) hours of battery life of which I know this otherwise lovely little laptop is capable, so I don't even need to check powertop? Suggestions? Thanks!

[AlbertP]

If you are compiling a program from source then you need the -dev versions of the packages that it depends on, in this case libncurses5-dev from the repository for example.

[tarverator]

Thanks for the tip for satisfying the dependencies.

I got powertop compiled and installed to /usr/local/sbin/powertop, and when it runs it says it is version 2.1 even though I downloaded 2.2... also a little strange, but whatever.

I can also run the version that apt-get pulls down and puts in /usr/sbin/powertop which is v1.97.

Neither version shows wattage. What next?

[AlbertP]

I don't really know what to do further, I've looked up the X230 at ThinkWiki to see what components are in it (http://www.thinkwiki.org/wiki/Category:X230) and nothing suspicious like power consuming graphics cards is found there.

[tarverator]

Thanks AlbertP for checking.

Well, I tried running powertop --calibrate and it got part way through it's thing, but then the machine locked up when the calibration got to the part where it checks the radios. (My machine suffers from the occasional hard lock-up problem, which I thought had to do with an HD 4000 Graphics bug, but may have something to do with the wifi; in any case it looks like kernel 3.7.6 didn't fix that.)

After the reboot, powertop 2.1 now shows wattage. (powertop 1.97 still does not show wattage.)

So, with the screen turned right down, my machine is burning between 9 and 10 W and powertop projects 2.5 hours of life on a fully-charged battery.

Something is wrong here, but what? It would be nice if the fix for the power leak and the hard lockups were the same.

[Matt267]

tarverator,

While you're troubleshooting powertop, you can run

Code: Select all

upower -d

to give you a "snap shot" view of power usage among other battery/power related information.

Matt

[tarverator]

Thanks for the upower tip. Nice!

The line that stands out is:
capacity: 45.5301%

I noticed that a month or two ago, at boot time, I started getting the "Battery may be old or broken" message. I turned it off, as clearly this is bogus (a) because lots of other people are getting it, and (b) the machine is a high-quality unit not yet a year old. And yet, here is upower reporting sharply diminished capacity. What's with this?

[Matt267]

I'm surprised the capacity in so low after less then a year. My dell mini 1018 is just over two years old and it's capacity is about 72%. Try a web search and see if premature battery failure is common with you laptop/battery combo.

Matt

[tarverator]

No, that's just it: this model is famous for long battery life. I think it's a Linux Mint thing, because (my impression is that) lots of people started getting the battery capacity warning a few months ago, across many makes and models.

[tarverator]

I have (cross-)posted this issue to the Lenovo forum, in case anyone there has any ideas from a strictly hardware point of view. http://forums.lenovo.com/t5/X-Series-Th ... -p/1027401 I just find it hard to believe it's a purely hardware problem, because I feel happier when I imagine that ThinkPads (especially mine) are bullet-proof exquisite little masterpieces of flawless engineering.

[AlbertP]

Well, after seeing the insides of an X230 on thinkwiki.org I found it hard to believe that this is a software problem.

I just find it hard to believe it's a purely hardware problem, because I feel happier when I imagine that ThinkPads (especially mine) are bullet-proof exquisite little masterpieces of flawless engineering.

That was also my impression of ThinkPads; I once borrowed a T41 when my own laptop was being repaired and this laptop still worked wonderful (except that it couldn't run Cinnamon and did not support PAE) despite being 7 years old.

[tarverator]

I am starting to think it might be hardware because I logged into Windows (which I only keep around for these kinds of eventualities and otherwise haven't even booted since September) and Lenovo's own "Power Manager" app also reports the battery condition as "poor" with 45% capacity.

One difference is that Linux reports 2.5 to 3 hours of life remaining on a fresh charge at this point, depending on screen brightness, but Windows only projects 2 hours. Small comfort.

Power Manager also reports that the battery was manufactured just one month before I started using it last summer (good), and that there have been 66 charge cycles to date (reasonable, right?).

I guess I am going to have to look into warranty service.

In the meantime, is there any way Linux/Mint could have caused such battery deterioration? The only parts of Windows that I miss (that I know of) are some of the highly specialized hardware tweaks like having the touchpad detect the difference between left and right mouse-clicks. Please don't tell me that Lenovo ships some kind of proprietary driver for the battery, without which it dies prematurely. Or rather, even if I don't want to hear it, please do tell me if it's the case.

[AlbertP]

In the meantime, is there any way Linux/Mint could have caused such battery deterioration?

Software & drivers can of course influence the power consumption, but apart from that I don't see a reason why the battery would deteriorate more. But, you posted a forum thread about it already, so it seems it's a known issue. And such nasty driver things should not be found in Lenovo laptops. Lenovo is not against using Linux, even though they don't support it as much as Windows.