[Gllug] Still interpreting output from "sensors"

[John Winters]

> On Sun, Apr 24, 2005 at 04:14:08PM +0100, John Winters wrote:
[snip description of CPU temp rising from 31 to 39 in 1 second]
>>
>> Can I expect the CPU temperature reading to increase that quickly?
>
> I am surprised by both the value of the temperatures and the rate of
> increase.
>
> On my dual Opteron, the 'system' and one of the cpus sit at about 45 deg
> ( the other cpu has a 1u fan to fit in the case and runs hotter). The
> machine is well cooled with a total of 7 fans including 2 120 mm ones.
>
> When the machine is compiling a kernel the cpu temperatures rise by
> about 5 deg. over several minutes. The 'system' temp rise is much
> smaller.
>
> I presume that the thermal characteristics of the AMD64 and Opterons are
> very similar. The AMD data sheet specifies a maximum of 70 deg for my
> 246s (OSA 246CEP5AL)

The BIOS screen shows a CPU temp of 39 degrees C.  I'm assuming this is
higher than the Linux idle temperature of 31 because the BIOS doesn't have
an intelligent idle loop.  The lm-sensors documentation describes exactly
this phenomenon.

Is anyone else using lm-sensors on an Athlon system?  Do you get the same
rapid temperature climb?  I recall seeing descriptions of how Athlons
would self-destruct in a fraction of a second if not properly cooled. 
Perhaps the thermal characteristics *are* different.

OTOH, if it's not the CPU, what might genuinely increase in temperature
that quickly?

John

-- 
Gllug mailing list  -  Gllug at gllug.org.uk
http://lists.gllug.org.uk/mailman/listinfo/gllug