Hi,
Seife or Holger: Can one (or both) you take look at this problem? Please read
my answer before. I assume the CPU freq problem is the known and annoying
DELL D600 problem, but there are some questions/problems with powersave.
Thanks,
Danny
---------- Forwarded Message ----------
Subject: Using powersave in everyday life (was: Proposal for resolving
powernowd/apmd...)
Date: Wednesday 19 April 2006 22:53
From: "Michel D'HOOGE" <[EMAIL PROTECTED]>
To: "Kubuntu Developer Discussion" <[EMAIL PROTECTED]>
I had time to experiment in the train this afternoon and here are my news
comments...
First of all I must say that the Kernel documentation about CPUFreq is quite
clear and I played a lot with /sys/devices/system/cpu/cpu0/cpufreq/. I also
found the cpufreq_stats kernel module that collects statistics on the
frequencies used. It is a must if you want to have fine-grained feedbacks
without the hassle to watch constantly what say tools like kpowersave for
instance ;-)
2006/4/19, Danny Kukawka <[EMAIL PROTECTED]>:
> Was the kernel ondemand governor loaded?
I can't remember the initial state of the system but I switched for a while
to the ondemand governor and nothing happened: the frequency remained at its
highest while I was doing nothing but reading on the screen. I can do other
experiments if you think it could be of interest. And that time I'll write
everything down.
What say
> 'cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_available_governors' ?
"ondemand userspace powersave performance"
I played with all of them! But neither ondemand nor userspace changed the
frequency dynamically - at least not within a minute.
And then I changed /etc/powersave/cpufreq to use the userspace governor.
When I restarted powersaved the userspace governor was selected with the
highest frequency and cpufreq_stats showed no sign of reduced frequency. I
wondered then whether I had to activate another program. But after a while,
it started to work. Maybe it is due to my tests on the LID events or maybe
it started to work after my last resume from RAM. However, right now, after
a cold boot it still works!
If your machine support cpufreq, don't touch throttling. Btw. throttling
> save
> not really power or reduce the power consumtion of your machine, but slow
> down all processes.
Good news. One thing less to try to understand :-P
> I meant that after resuming from STR, Kpowersave always says the frequency
>
> > is 800MHz even if I ask it to switch to the performance scheme. I still
> > need to experiment to determine the best configuration.
>
> What say: 'grep . /sys/devices/system/cpu/cpu0/cpufreq/ -r' (as root)
> before
> suspend and what after resume?
Being pleased with what I learned so far I decided to experiment with
powersave scripts and events. So I added beeps to the LID open/close events.
And I thus noticed that 3 scripts (beep_hilo, beep_lohi and notice) use the
usleep command. Unfortunately it is not available in my box and it seems to
be missing in ubuntu and debian either. I still have another debug warning
in the log ([powersave-hotkey_handler][5884]: WARNING: EV_ID: 55, unclean
exit ...) but I can't find the name of the faulty script.
I have the following configuration:
EVENT_BUTTON_LID_OPEN="beep_lohi switch_vt"
EVENT_BUTTON_LID_CLOSED="beep_hilo screen_saver"
BTW, when I close and open the lid, I have 3 ACPI events: button/lid
(close), video VID and button/lid (open). Is the video event used by
powersave? I can't find any trace of it.
Finally, I noticed some minor bugs: in Kpowersave, notifications for "resume
from disk" is written twice while "resume from RAM" is absent. I also
noticed a 'bold' ('b>')error in a french message.
I do start thinking that my laptop will soon be as good with Linux than with
the closed drivers provided by Dell on windows :-D
--
Michel
-------------------------------------------------------
_______________________________________________
powersave-devel mailing list
[email protected]
http://forge.novell.com/mailman/listinfo/powersave-devel
