On Tue, May 07, 2013 at 12:54:21PM +0000, Sebastian Werner wrote: > > > >>On Mon, May 06, 2013 at 12:31:11PM +0000, Sebastian Werner wrote: > >> Hello, > >> > >> I am trying to get the right temperature from two different pieces of hardware. > >> The first, newest piece (http://www.jetwaycomputer.com/ITX-JBC362F36.html) works right out of the box. Got lm-sensor on the debian system, sensors-detect tells me to modprobe lm78 and the temperature is roughly the same as the one shown in the bios. > >> > >> Acpitz-virtual-0 > >> Adapter: Virtual device > >> Temp1: +40.0°C (crit = +127.0°C) > >> Temp2: +26.8°C (crit = +127.0°C) > >> Temp3: +26.8°C (crit = +100.0°C) > >> > >This output is not from the lm78 driver. > > > >The "coretemp" driver works with this CPU (Atom N2600). Load it and see what it reports. > > Weirdly enough, trying to load "coretemp" on the new hardware gives me a > "No such device" error ... > Old drivers, most likely. What is the output of /proc/cpuinfo ? > Like Jean mentioned, I am running a rather old version, due to having old > packages and all their baggage. I took the newest version from http://dl.lm- > sensors.org/lm-sensors/files/sensors-detect and reran it on all the systems I > have access to. > sensors-detect output: > > Driver `coretemp': > * Chip `Intel digital thermal sensor' (confidence: 9) > > Driver `f71882fg': > * ISA bus, address 0x295 > Chip `Fintek F71808A Super IO Sensors' (confidence: 9) > > #----cut here---- > # Chip drivers > modprobe coretemp > modprobe f71882fg > /usr/bin/sensors -s > #----cut here---- > > > >> Not temp2 and temp3 though. Both temperatures never change and they have no connection to anything I can read in the bios. > >> I presumed that those are 'dead wires' and I am at the moment ignoring those, since temp1 is good enough for me. > >> > >The temperatures are reported by ACPI. No idea how that translates to hardware > >temperatures. If the temperature never changes, there may be a bug in the > >ACPI tables. Those are controlled by the board manufacturer, and would be > >updated with the BIOS. Do you have the latest BIOS versions installed ? > > > >> The second, the older piece (http://www.jetwaycomputer.com/ITX-JBC360F33-B.html) does not seem to work. Sensors-detect tells me to load coretemp, using the latest version of sensors-detect from the site wants to also me to load the f71882fg driver. That doesn't work out with a 'No such device' error. > >> > >What is your kernel version, and what exactly does sensors-detect report ? > > I am on debian sqeeze, running 2.6.32-5-686 > You will need a more recent kernel version, or install stand-alone drivers for both coretemp and f71882fg. Not sure what the lm78 detection is about, but that doesn't look correct as Jean pointed out. > sensors-detect gives me this: > > Driver `coretemp': > * Chip `Intel digital thermal sensor' (confidence: 9) > > Driver `lm78': > * ISA bus, address 0x290 > Chip `National Semiconductor LM78' (confidence: 6) > > Driver `f71882fg': > * ISA bus, address 0x295 > Chip `Fintek F71869F/E Super IO Sensors' (confidence: 9) > > #----cut here---- > # Chip drivers > modprobe coretemp > modprobe f71882fg > modprobe lm78 > /usr/bin/sensors -s > #----cut here---- > > coretemp is loadable, but f71882fg gives me this: > ~/detect# modprobe --first-time -v f71882fg > insmod /lib/modules/2.6.32-5-686/kernel/drivers/hwmon/f71882fg.ko > FATAL: Error inserting f71882fg (/lib/modules/2.6.32-5-686/kernel/drivers/hwmon/f71882fg.ko): No such device > > >If I interpret the information I find on the web correctly, the board might > >be MI812 or MI810 from iBT. If so, the Fintek chip on the board is only used > >for COM ports. Try loading the w83627ehf driver instead. > > Trying to load w83627ehf gives me the same "No such device" error. > > >> Acpitz-virtual-0 > >> Adapter: Virtual device > >> Temp1: +30.0°C (crit = +100.0°C) > >> Coretemp-iso-0000 > >> Core 0: +30.0°C (crit = +90.0°C) > >> > >The CPU is Atom N270, so the reported critical core temperature is correct. > >CPU temperature measurements are highly inaccurate that far below the maximum > >temperature, so the reported value is as good as it gets. Accuracy is better > >with higher temperatures. > > All in all I am not after exact values. I need to find out if the system had > been used in the wrong environments, which might have lead to the demise of the > hardware. How accurate are we talking here? I just want to set a threshold to > watch over, maybe setting it to ~50 degrees as a first stepping stone. > TjMax for the CPU is 90 degrees C. If you set your limit to 15-20 degrees C below that, you will be absolutely fine. To give you a reference, here is what my SandyBridge system is running at right now (full load): coretemp-isa-0000 Adapter: ISA adapter Physical id 0: +85.0°C (high = +80.0°C, crit = +98.0°C) Core 0: +82.0°C (high = +80.0°C, crit = +98.0°C) Core 1: +82.0°C (high = +80.0°C, crit = +98.0°C) Core 2: +85.0°C (high = +80.0°C, crit = +98.0°C) Core 3: +82.0°C (high = +80.0°C, crit = +98.0°C) Guenter _______________________________________________ lm-sensors mailing list lm-sensors@xxxxxxxxxxxxxx http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
