| 1 |
On Wed, 2010-08-04 at 18:22 +0300, Nikos Chantziaras wrote: |
| 2 |
> On 08/04/2010 02:14 PM, Xi Shen wrote: |
| 3 |
> > hi, |
| 4 |
> > |
| 5 |
> > after i setup lm_sensors on my gentoo amd64, i ran sensors, and got |
| 6 |
> > the below output |
| 7 |
> > |
| 8 |
> > coretemp-isa-0000 |
| 9 |
> > Adapter: ISA adapter |
| 10 |
> > Core 0: +61.0 C (high = +74.0 C, crit = +100.0 C) |
| 11 |
> > |
| 12 |
> > coretemp-isa-0001 |
| 13 |
> > Adapter: ISA adapter |
| 14 |
> > Core 1: +61.0 C (high = +74.0 C, crit = +100.0 C) |
| 15 |
> |
| 16 |
> This is the sensor inside the CPU. The kernel doesn't know how to |
| 17 |
> interpret this value on non-mobile CPUs, and it's usually off by 10C to |
| 18 |
> 15C on desktop CPUs. |
| 19 |
> |
| 20 |
> |
| 21 |
> > CPU Temperature: +49.0 C (high = +90.0 C, crit = +125.0 C) |
| 22 |
> |
| 23 |
> This is the sensor on your motherboard that resides directly under the |
| 24 |
> CPU. This an accurate temp and the kernel knows exactly how to |
| 25 |
> interpret the values. |
| 26 |
> |
| 27 |
> |
| 28 |
|
| 29 |
|
| 30 |
This is windows specific but has lots on how Intel/AMD work. |
| 31 |
http://www.alcpu.com/CoreTemp/ |
| 32 |
|
| 33 |
This page seems to imply that Linux's use of coretemp is not as |
| 34 |
"detailed" as available to windoze users and if an unknown cpu will use |
| 35 |
a default value which may be incorrect. |
| 36 |
http://www.mjmwired.net/kernel/Documentation/hwmon/coretemp |
| 37 |
|
| 38 |
The lm_sensor temperatures (and voltages) are always highly suspect, |
| 39 |
usually being based on users experiments rather than manufacturer |
| 40 |
information which is usually not available. Also, I suspect variation |
| 41 |
even between motherboards of the same type as sometimes the supposed |
| 42 |
lm_sensors values for one of my systems (often cpu voltage) are not even |
| 43 |
close. |
| 44 |
|
| 45 |
BillK |
Archives