You are not logged in.
Pages: 1
Hello, recently moved to Arch from Windows and one thing I can't wrap my head around is getting sensors to work. I've installed Lm_sensors and carefully read the wiki page. Yet my "sensors" output is really weird. Here is output of sensors-detect and sensors: http://pastebin.com/0jykUY2z. Here is /etc/conf.d/lm_sensors content after running sensors-detect: http://pastebin.com/c41KvtMt.
Only weird things I have noticed are:
"libkmod: kmod_config_parse: /lib/modprobe.d/nvidia.conf line 2: ignoring bad line starting with 'Option'" in sensors-detect
and
[ 0.028054] [Firmware Bug]: AMD-Vi: IOAPIC[0] not in IVRS table
[ 0.028061] [Firmware Bug]: AMD-Vi: No southbridge IOAPIC found
[ 0.028065] AMD-Vi: Disabling interrupt remapping
in dmesg
For the latter I've tried searching for some workarounds and best I could find, was adding
amd_iommu_dump=1 ivrs_ioapic[0]=00:00.1 ivrs_ioapic[0]=00:14.0
to arch.conf, but that didn't fix sensor issues.
More system info:
OS: Arch Linux
Kernel: x86_64 Linux 4.0.3-1-ARCH
DE: Cinnamon 2.4.8
CPU: AMD Athlon X4 760K Quad Core @ 4.492GHz
Motherboard: MSI A78M-E35
GPU: GeForce GTX 750
RAM: 5850MB
Don't know what more information to give, any tips or ideas would be highly appreciated!
Offline
There seem to be two things going on.
The first is that one of your sensors, the "Fintek F71868A" does not have a Linux driver. See the status here.
The second is that there seems to be an error in the second line of the file /lib/modprobe.d/nvidia.conf. Did you modify that file? What does that line say?
In any event, it appears that error is preventing sensors-detect from probing your GPU (which may or may not have any sensors).
Offline
I did indeed add
Option "RegistryDwords" "PerfLevelSrc=0x2222"
to nvidia.conf in an effort to force maximum performance power mode. Don't know if it helped, but removing this line did make previous error go away. I run sensors-detect again and there was still no change in sensors output.
Could it be possible that the absence of Fintek F71868A driver messes up all of my sensors?
Offline
That may well be the case (although you still have CPU thermal)? What sensors do you expect to have and what do you propose doing with the data once you have it?
Regards,
Offline
Yea, i guess the one only one sensor, that shows up, is CPU although it's readings are riddiculous (0C). I expect to have CPU, GPU, SSD and Mobo sensors, atleast I had eadings from them back in Windows. Funny thing, I can still see my GPU temperature in Nvidia X Server Settings under GPU 0 -> Thermal Settings. I wanted to get sensors right so I could keep an eye for my CPU ( OC'd from 3.8 to 4.5) and throw CPU/GPU temperatures in conky.
Offline
The wiki contains sections on how to monitor nvidia GPU temperatures from the shell; I assume you could adapt those to conky (if you're willing to bear the overhead). I also have seen reference to an ${nvidia temp} setting for conky which could bear investigation.
In terms of monitoring drive health (including temperature), you may want to look at the smartmontools package. Again, you should be able to script it for conky (assuming there to be no native conky support).
Not sure what is going on with the CPU readings.
Regards,
Offline
The ${nvidia temp} should work, no need to mess with scraping commandlines, which has significant overhead. If it doesn't work, install libxnvctrl and recompile conky with --enable-nvidia. For disk temperature there's {hddtemp}, it requires that a dameon is running, the Arch hddtemp package has a systemd service to start the dameon.
Regarding the CPU, a quick search tells me that the sensor returns a relative value, not the absolute temperature. And in your case, it seems lm_sensors doesn't use the correct math to turn what k10temp provides into a proper temp value. No idea what to do about that, do a search yourself, maybe you'll get lucky.
For mobo sensors, you're plain out of luck, your particular Fintek chip isn't supported. You can try manually loading the f71882fg module, but chances it'll work are low.
Edit: The value k10temp returns is an offset from the maximum. lm_sensors assumes 70°C is the maximum, because it usually is for these processors, but your specific CPU it seems has a higher maximum. Conky allows modifying the temp value from the sensor:
hwmon (dev) type n (factor offset)
Hwmon sensor from sysfs (Linux 2.6). Parameter dev may be omitted if you have only one hwmon device. Parameter type is either 'in' or 'vol' meaning voltage; 'fan' meaning fan; 'temp' meaning temperature. Parameter n is number of the sensor. See /sys/class/hwmon/ on your local computer. The optional arguments 'factor' and 'offset' allow precalculation of the raw input, which is being modified as follows: 'input = input * factor + offset'. Note that they have to be given as decimal values (i.e. contain at least one decimal place).
So check the temp in the BIOS, then try random offsets in the conky config until it shows a value that's close to what you saw in the BIOS. An offset of 10.0 is a good place to start, then try 20.0, 30.0, etc.
Also, try this or something similar
cat /sys/class/hwmon/hwmon0/temp1_inputExplore the contents of /sys/class/hwmon/ to see what's there. This directory contains the raw values as returned by k10temp.
Last edited by Gusar (2015-05-19 11:46:06)
Offline
Pages: 1