1 |
On Montag 26 Oktober 2009, Grant wrote: |
2 |
> >> After upgrading from 2.6.28 to 2.6.31, I noticed my CPU temperatures |
3 |
> >> are reported a full 20C hotter. If I load the old kernel, the |
4 |
> >> reported temperatures drops back down to normal. Has anyone else seen |
5 |
> >> this? |
6 |
> > |
7 |
> > If you're using coretemp as sensor, the temps are always off (the |
8 |
> > coretemp sensor of Intel chips is not accurate, not by any stretch of the |
9 |
> > imagination.) It only reports the distance to the CPU's maximum thermal |
10 |
> > junction, which then the coretemp driver *tries* to translate into a |
11 |
> > temperature, but the result is wrong since the value reported by the CPU |
12 |
> > is not accurate to start with (it only gets accurate as you approach the |
13 |
> > max value). That maximum value is totally undocumented for desktop CPUs |
14 |
> > (the docs Intel provided recently are wrong.) |
15 |
> > |
16 |
> > You should use your mainboard's sensors instead for accurate values. |
17 |
> |
18 |
> I'm actually using k8temp. Do you think it is susceptible to the same |
19 |
> problems you're talking about? I also have an ACPI sensor available |
20 |
> named THRM. Should that one be more accurate? |
21 |
> |
22 |
> BTW, another system of mine (Dell laptop) only seems to have available |
23 |
> coretemp or an ACPI sensor which reports values like 46960 mWh. Am I |
24 |
> totally out of luck with that one? |
25 |
> |
26 |
> So, In the end, it's fairly impossible to monitor a CPU's actual |
27 |
> temperature in order to keep it below the published maximum? |
28 |
> |
29 |
> - Grant |
30 |
> |
31 |
|
32 |
there are no published maximums. Ruin a good evening going through CPU specs. |
33 |
|
34 |
And k8temp has the same problem. Or similar. Some CPUs report correct temps, |
35 |
other doesn't and some report some complete bogonium. |