Mailinglist Archive: opensuse (2224 mails)

< Previous Next >
Re: [opensuse] Battery life
  • From: ken <gebser@xxxxxxxxxxxxx>
  • Date: Fri, 02 Feb 2007 06:22:07 -0500
  • Message-id: <45C31EDF.60202@xxxxxxxxxxxxx>


On 02/02/2007 12:38 AM somebody named M Harris wrote:
> On Thursday 01 February 2007 23:21, Mark Panen wrote:
>> I bought
>> a new one on Monday and it is supposed to get up to 4 hours but i only
>> get 1 hour 40 minutes. I have charged and discharged the battery a
>> couple of times.
> The time you are getting is 'bout normal...
>
> ... there are many things you can do to cut back on the current drain.
> 1) reduce the cpu operating frequency (bios) if your unit allows for this
> 2) reduce the up-time of the hd drive motor (timer)
> 3) reduce the brightness of the lcd, and up-time of the monitor (timer)
> 4) turn of peripherals when not used... including wan, wifi, modems, etc
> 5) set a suspend timer
> 6) suspend instead of booting
> 7) set lcd to power down (timer) instead of running a screen saver
>
> You may need to fully cycle (charge-discharge) your new battery several times
> before you will see optimum performance.
>
> I have never had more than 2.5 hours from my laptop (ThinkPad R30) under
> optimal loading. Obviously this is completely subjective.

There's two issues at play. One is what I call "sales speak",
specifically the wording, "you'll get *up to*...." Occasionally I'll
reply, "Well, I'll pay you *up to* [their stated price]." (This usually
gets a laugh.) The point is that "up to four hours" does include "two
hours", "five minutes", or any amount of time under "four hours". In
brief, "up to four hours" is a promising way to say almost nothing.

Secondly, and as M Harris wrote, there are several factors which
determine battery life. Years ago, when these batteries (this Dell
I600m has an optional second battery with approximately the same
capacity as the primary battery) were new, and when I was for testing
purposes implementing most of the measures cited above, I got more than
7.5 hours in a session, this without suspending during that time.

To get this amount of battery time, I implemented most, but not all, of
the measures cited above. This Inspiron came with (because I wanted it)
the so-called Mobile Technology. This is a concept, from my
understanding, pioneered by Linus Torwalds which induces the CPU to
adapt to the CPU load: when the demands on the CPU are lower, the CPU
runs more slowly, as little as 10% of its rated speed. This "stepping"
of the CPU to different speeds is, moreover, tunable in the BIOS and can
be performed also when logged in as root. More info can be found (in
9.3 anyway) at "man powersaved".

If you don't have a stepping CPU, your battery is going to drain a lot
faster.

Another factor (call it #8), then, is the demand placed on the hardware
by the number and kind of processes which are running. (a) The greater
the load placed on the system, the more CPU cycles will be required, and
the higher the CPU will have to step to accommodate it. (b) Similarly,
if you don't have enough RAM, your system will resort to swap, causing
the drive to spin more. Spinning the drive consumes much more
electricity than does RAM. Optimally, your system should hardly ever
touch swap; at the same time, having RAM that is never used consumes
energy without benefit to the system. So tune the amount of RAM in your
system to accommodate memory usage.

Another source of savings which can be had for free (only your time)
relates to (a) above: compiling your code. I would imagine that most
people using Intel-achitecture simply download and install the i686
binaries. Compiling the source code which has been configured to your
hardware can save a lot of CPU cycles, make the code run a lot faster,
and so too save some battery.

Also related to (a), and more obvious, is to shut down daemons which
aren't being used. This is a good idea from a security standpoint as well.


hth,
ken

--
To unsubscribe, e-mail: opensuse+unsubscribe@xxxxxxxxxxxx
For additional commands, e-mail: opensuse+help@xxxxxxxxxxxx

< Previous Next >
Follow Ups
References