Re: ACPI throttling to death?

[Thomas Renninger <trenn-l3A5Bk7waGM@public.gmane.org>]

Jan Frey wrote:
> On Tuesday 05 April 2005 18:38, Stefan Dösinger wrote:
>>Am Montag, 4. April 2005 20:14 schrieb Jan Frey:
>>>Hi,
>>>
>>>I'm observing strange behaviour with a Gericom notebook running a
>>>2.6.9 vanilla kernel with custom configuration. The processor used is
>>>a P III (Tualatin) with 1.2GHz (which is quite rarely used). It is
>>>*not* a mobile processor (no speedstep etc.), instead it supports 16
>>>throttling levels.
>>>
>>>Sometimes (I guess when the machine is getting overheated - the
>>>notebook design is really bad and the fan is quite lousy) ACPI
>>>temperature monitor shows >70 degrees (Celsius) and the system starts
>>>throttling.
>>>
>>>Until here everything is fine IMO.
>>>But what happens next is that the systems slows down continuously (you
>>>can monitor the current throttling level going up and up) until it
>>>reaches the last level (12%). Soon after that the whole machine is
>>>locked up hard, display still showing the current graphics. I guess it
>>>throttled to a "deathly" 0%...
>>>
>>>Is this more a linux acpi problem or a fault in this particular
>>>machine? What could be done about it? how can I debug further?
>>My desktop machine locked up if it got to hot(70-75 degrees, amd duron
>>processor). I bought a better CPU fan and the problem is gone. Well,
>>that's impossible on notebooks.
>>
>>Is there a 0% throttling level? What happens if you stop processor
>>hungry tasks to allow the machine to cool down?
> 
> No, there is no 0% level. Unfortunately this also does not seem to be 
> related to any level of CPU utilization: sometimes the machines locks up 
> even if it is nearly idle, e.g. writing/sending email...
> 
> Really astonishing: CPU temperature seems to go up to roughly 70 degrees 
> quite quickly and then it stays there nearly forever: it does not matter 
> whether CPU load is 0 or 1.. And: throttling also does not lower CPU temp, 
> I thought this would be main reason...

The temperature is fixed to exactly 70 C?
I saw machines where a fixed temperature is returned depending on some bit/byte
in EC/SystemIO.
Sounds like this one does the same?
You'll find out by going back the _TMP code in the DSDT.
The value should be about 0xD66 ((70+273)*10, right?)

     Thomas


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_ide95&alloc_id\x14396&op=click