> You could also consider just sticking to tuned and then having a look at the power management options as provided there. tuned-adm list will show you some predefined power management options which *can* be tweaked.
I have made many tests with tuned and written small scripts to switch
from one profile to the other (laptop-battery-powersave on battery,
default on AC).
Gains where in the 1W to 2W range vs. 9W gain with the kernel
arguments (which is nice now that I'm around 12W, but it was 25W at
the beginning!)
> Do you know what those options due to your machine in order to make the battery last longer? I mean really, do you know what they do?
They are related to Intel graphic drivers (follow links in OP):
http://www.williambrownstreet.net/blog/?p=387
http://askubuntu.com/questions/38117/battery-life-decreased-after-upgrade-to-11-04
I don't know much more, but what I know is that this single change
increased battery life on my laptop by a factor of two, that the fan
is not running at full speed all the time (it also was on AC), and
that nothing was broken for the last two days I have been working with
it.
> These could be bad options for a number of users and since it's set at kernel boot time how can you override it once the OS has booted? Can you disable this without altering boot parameters and rebooting? If the answer is yes than a tuned configuration should be created or altered to set them dynamically. Setting of these at boot time are likely just wrong. You likely only want these to be turned on when the laptop is not attached to power, which you can create hooks for.
Definitely, these could be bad options for some users (or, more
likely, irrelevant ones). I posted to the list, so that when somebody
will search for 'centos 6 thinkpad power consumption too high' he will
bump into the Ubuntu related post I linked to (which provides
additional links to the root cause) but also that this person will see
that it worked pretty well in my particular case.
> This is not a bug, it's a feature/workaround on specific hardware, that tweaks specific settings to get around a specific issue with the driver. Create a profile and submit it upstream.
The above links rather point to a regression.
I assume that CentOS users are experienced enough to do their own
risks/benefits analysis before applying such tweaks. We can probably
agree that we disagree on that point.
_______________________________________________
CentOS mailing list
CentOS@xxxxxxxxxx
http://lists.centos.org/mailman/listinfo/centos
