[elrepo] kmod-coretemp
Simon Wilson
simon at simonandkate.net
Mon Oct 19 06:59:49 EDT 2009
Quoting Phil Perry <phil at elrepo.org>:
> Simon Wilson wrote:
>> I notice from various reading that the TJunction max varies
>> between
>> various processors in the Core2 Duo range.
>>
>> I have just put an e6850 into my CentOS box, installed lm_sensors
>> 2.10.8 and kmod-coretemp. The coretemp output at idle is as
>> follows:
>>
<snip>
>>
>> The tJunction-max for my G0 stepping E6850 is 90°C. Where does
>> the
>> coretemp driver get the 100°C high value from? There is no
coretemp
>> section in sensors.conf... Is it programmed into the driver?
>>
>> The system's (it8718 driven) CPU temp (tCase I assume) is 24°C at
the
>> same time as the above reading.
>>
>> Just want to make sure that I'm monitoring the right values...
>>
>
>
> Hi Simon,
>
> Yes, TjMax is hard coded in the code for each processor family.
It's
> either 100C or 85C depending on the processor model. I do note that
some
> recent commits to the code were made at the end of last month,
including
> support for Atom processors that appear to have a TjMac of 90C, so
it is
> quite possible that your processor also has this value. Some mobile
> Penryn CPUs also appear to now use a TjMac of 90C.
>
> Let me backport these updates to our driver in elrepo and then lets
see
> how that affects support for your chip. I'll do my best to get that
done
> this week. If it's still definitely not right, then we can look to
> develop (and test) a patch to submit upstream at kernel.org.
>
> Regards,
>
> Phil
> _______________________________________________
> elrepo mailing list
> elrepo at lists.elrepo.org
> http://lists.elrepo.org/mailman/listinfo/elrepo
>
>
Doing some reading - e.g. http://www.overclock.net/5639179-post5.html
It would appear that the TjMax on some of these is a little more hit
and miss. A change in the hard coding from 100 to 90 would drop the
computed temp to be in the 15 to 20 degree range (even though ambient
is probably about 20 this evening), so it would appear that the
computed result, albeit based on a not "spec" 100 is probably closer
to actuality in this instance.
At the end of the day, it's the distance from TjMax that is I guess
critical, and as long as I am sitting down at around 60 degrees from
that at idle it's unlikely to be too much of an issue... even at flat
out, which the server rarely is...
Thoughts?
Simon
--
Simon Wilson
www.simonandkate.net
More information about the elrepo
mailing list