[arch-general] powertop vs archlinux vs ubuntu
Stefan Erik Wilkens
stefanwilkens at gmail.com
Mon Feb 22 11:09:15 EST 2010
2010/2/22 Stefano Z. <mie.iscrizioni at gmail.com>:
> seems this is my same problem:
>
> http://bugs.archlinux.org/task/17771
>
> On Mon, Feb 22, 2010 at 10:42 AM, Stefano Z. <mie.iscrizioni at gmail.com> wrote:
>> nothing, the problem still exist, also with i915.modeset=0.
>>
>> On Mon, Feb 22, 2010 at 11:30 AM, Andrea Fagiani <andfagiani at gmail.com> wrote:
>>> On 02/22/2010 09:04 AM, Stefano Z. wrote:
>>>>
>>>> no, my dual celeron su2300 dosen't support speedstep, it stay fixed to
>>>> 1.2ghz
>>>> but i have found the problem, i think this problem have to do with KMS...
>>>> I have istalled kernel26 2.6.31.6-1 and the problem disappered but
>>>> obviously
>>>> this is not the solution i want ;-)
>>>> If i install 2.6.32+ kernel the problem reappear.
>>>> Another thing that happens is that the cpu(s) Temperature with kernels
>>>>
>>>>>
>>>>> 2.6.31 stay
>>>>>
>>>>
>>>> about on 53+C while with kernel<2.6.32 stays on 42C...
>>>>
>>>>
>>>>
>>>> On Mon, Feb 22, 2010 at 9:43 AM, Andrea Fagiani<andfagiani at gmail.com>
>>>> wrote:
>>>>
>>>>>
>>>>> On 02/22/2010 04:53 AM, Brendan Long wrote:
>>>>>
>>>>>>
>>>>>> On 02/21/2010 04:55 PM, Xavier Chantry wrote:
>>>>>>
>>>>>>
>>>>>>>
>>>>>>> On Mon, Feb 22, 2010 at 12:42 AM, Stefano Z.<mie.iscrizioni at gmail.com>
>>>>>>> wrote:
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>>
>>>>>>>> hi
>>>>>>>>
>>>>>>>> i've bought a new notebook (hp pavilion dm1-1150sl) and installed
>>>>>>>> archlinux.
>>>>>>>> i have see a strange thing with powertop, i'm running the vanilla arch
>>>>>>>> kernel26,
>>>>>>>> and i have see this behaviour:
>>>>>>>> Cn Avg residency P-states (frequencies)
>>>>>>>> C0 (cpu occupata) (31,6%)
>>>>>>>> C0 0,0ms ( 0,0%)
>>>>>>>> C1 mwait 0,1ms ( 0,7%)
>>>>>>>> C4 mwait 0,0ms (67,8%)
>>>>>>>> Wakeups-from-idle per second : 31483,5 interval: 10,0s
>>>>>>>> ---
>>>>>>>> as you can see, i have a LOT of wakeups per seconds very low c1 states
>>>>>>>> lot of c0 and c4 states,
>>>>>>>> a wattmeter tell me that archlinux consume about 25/26watt
>>>>>>>> Then i have boot a live ubuntu distro and see this:
>>>>>>>> Cn permanenza media P-state (frequenze)
>>>>>>>> C0 (cpu occupata) ( 0,6%)
>>>>>>>> polling 0,0 ms ( 0,0%)
>>>>>>>> C1 mwait 26,7 ms (68,6%)
>>>>>>>> C4 mwait 1,2 ms (30,8%)
>>>>>>>> Wakeup-da-idle al secondo: 281,3 intervallo: 15,0s
>>>>>>>> ---
>>>>>>>> as you can see the wakeups are a LOT lower than on arch and we have
>>>>>>>> lot of c1 and c4 state,
>>>>>>>> power consumption is about 20W, the same that i have with win7 (about
>>>>>>>> 18/20w).
>>>>>>>> For meaning about cX state see here:
>>>>>>>> http://www.lesswatts.org/projects/powertop/powertop.php
>>>>>>>>
>>>>>>>> thanks!
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>
>>>>>>> You didn't give enough information so we will need to check the basis :
>>>>>>> - which cpufreq driver and governor are you using in both cases (check
>>>>>>> cpufreq-info)
>>>>>>> - what processes does powertop show as causes for wakeups ?
>>>>>>> - what processes does top show in term of cpu usage ?
>>>>>>>
>>>>>>> Here is what I got in the last few minutes when i was writing this :
>>>>>>> C4 mwait 3.4ms (92.5%) 800 Mhz 98.0%
>>>>>>> Wakeups-from-idle per second : 279.4 interval: 10.0s
>>>>>>>
>>>>>>> I have a core 2 duo with acpi-cpufreq loaded and conservative governor.
>>>>>>> $ grep cpufreq /etc/rc.conf
>>>>>>> MODULES=(acpi-cpufreq)
>>>>>>> DAEMONS=(syslog-ng net-profiles crond dbus hal alsa cpufreq
>>>>>>> storage-fixup)
>>>>>>> $ grep governor /etc/conf.d/cpufreq
>>>>>>> # valid governors:
>>>>>>> governor="conservative"
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>
>>>>>> They may just not have cpu-freq-utils installed maybe? I'm using
>>>>>> laptop-mode-tools with compiz and GNOME running (but not doing anything)
>>>>>> and it's saying 99.2% C4, 0.8% C0. This is with another Core2 and
>>>>>> laptop-mode is set to use the powersave governor on battery (which is
>>>>>> how I tested). When I plug it in, it jumps up to 25%, but I don't really
>>>>>> care how active the processor is when it's on battery.
>>>>>>
>>>>>>
>>>>>>
>>>>>
>>>>> Be sure to have cpu-freq-utils installed, as well as loading the right
>>>>> modules (acpi-cpufreq, cpufreq_ondemand, if you plan on using the same
>>>>> governor as ubuntu does), and add `cpufreq` to your DAEMONS array in
>>>>> rc.conf.
>>>>> On my system, with laptop-mode-tools the powertop output is very similar
>>>>> to
>>>>> what Brendan said. Also, the 31k+ wakeups definitely mean there's some
>>>>> issue
>>>>> on your config.
>>>>>
>>>>>
>>>>
>>>>
>>>
>>> If you think it's KMS causing problems, try adding `nomodeset` to your
>>> kernel boot line and see what happens.
>>>
>>
>
http://bugzilla.kernel.org/show_bug.cgi?id=14742
and I quote:
http://bugzilla.kernel.org/show_bug.cgi?id=14742#c4
"ok so there is something very interesting here:
your system, for some reason, seems to exit the C2 state immediately
when it gets entered. (C2 is only available because the bios announces
its presence). This is a hardware/BIOS bug, and a bad one at that.
In the old code, for some reason, C2 is not used in
practice. With the new code, the governor will try to use C2,
repeatedly.
I think the real solution is not to change the governor, but to make a
quirk for your system so that Linux will just not use C2 on your
system...
Can you attach the output of "dmidecode"; we'll need that to make a
quirk."
and, a little further down:
http://bugzilla.kernel.org/show_bug.cgi?id=14742#c8
"[ A request for people who see a similar issue: Please file a seperate bug for
each different machine. So unless you have a HP PAvilion zv5000, you need a
different bug]"
They assumed that this issue would not rise on many different systems
but apparently more are showing. I would definately suggest you guys
to take this upstream.
--
msn: stefan_wilkens at hotmail.com
e-mail: stefanwilkens at gmail.com
blog: http://www.stefanwilkens.eu/
adres: Lipperkerkstraat 14 7511 DA Enschede
More information about the arch-general
mailing list