Hi Kapil, On Sun, 21 Jul 2013 23:23:00 -0400, Kapil Dev wrote: > I am using 2.6.39-020639-generic kernel and I am running Ubuntu 9.04. I couldn't find /etc/sensors.d/coretemp.config file. It's OK and expected, just create the file. Every file under /etc/sensors.d will be taken into account by libsensors. > I am using 3450S intel i5 processor, which has Tjmax of 103C (sorry for the typo in my original email), based on http://users.sosdg.org/~qiyong/lxr/source/Documentation/hwmon/coretemp?a=um . Note that this is our own driver documentation file. We did out best to make it accurate but it is not an official reference for TjMax values. > I found that for some processors, IA32_TEMPERATURE_TARGET MSR has the Tjmax value; I assume that the coretemp.c tries to read this register and if it doesn't succeed then, it uses the heuristic value. Yes, exactly. > Few other doubts I have are: > > 1) For a multi-core processor, are there separate Tjmax for each core or just 1 Tjmax value per processor? The Intel datasheet was not very clear on this, so the driver makes no assumption and reads IA32_TEMPERATURE_TARGET on each core. However in practice I have never seen a CPU with different TjMax values for the different cores. > 2) Inside coretemp.c file, the temperature is read in milli-degree C, and multiplied by 1000 to get C value. > Does it mean that DTS value stored in IA32_ThERMAL_STATUS MSR is in milli-degrees? I also notice that there are only 7 bits assigned to DTS value, which would not be sufficient to cover the wide range of temperate values if it is in mill-degree C. Am I missing something here? It's the other way around actually, the value is in (pseudo) degrees C in the MSR and milli-degrees C in sysfs. -- Jean Delvare http://khali.linux-fr.org/wishlist.html _______________________________________________ lm-sensors mailing list lm-sensors@xxxxxxxxxxxxxx http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
