APU2 Temperature Problem

Installing and Using OpenWrt
1

Hello,
I've been running 18.06.2 for a bit now on my APU2 device but I have one issue thats bugging me, temperature. Sensors reports back on only a single core. It looks like the folks over at pfsense have reporting on 4 sensors for this processor. AMD GX-412TC

Any thoughts? Is this a kernel issue? Hwmon?

2

pfSense is, last I knew, running on FreeBSD.

For a piece of mind, Debian 9 ("Stretch") reports

jeff@apu3:~$ sensors
fam15h_power-pci-00c4
Adapter: PCI adapter
power1:        6.54 W  (interval =   0.01 s, crit =   6.00 W)

k10temp-pci-00c3
Adapter: PCI adapter
temp1:        +60.0°C  (high = +70.0°C)
                       (crit = +105.0°C, hyst = +104.0°C)

and on FreeBSD 12

[jeff@js-apu ~]$ sudo kldload amdtemp
[jeff@js-apu ~]$ sysctl -a | fgrep temperature
dev.cpu.3.temperature: 60.5C
dev.cpu.2.temperature: 60.5C
dev.cpu.1.temperature: 60.5C
dev.cpu.0.temperature: 60.5C
3

activating the hardware monitoring support in pfSense, you will find 4 cores indeed:

sysctl -a | grep "dev.cpu.*.temperature"
dev.cpu.3.temperature: 58.0C
dev.cpu.2.temperature: 58.0C
dev.cpu.1.temperature: 58.0C
dev.cpu.0.temperature: 58.0C

https://docs.netgate.com/pfsense/en/latest/monitoring/hardware-monitoring-support.html

But I actually came here for the thread title: some years back there was a bug in pfSense (I guess), that would lock the CPU at close to 100%. The issue wasn't fixed in weeks, so I stopped using pfSense for a while because the box was getting so hot, you could barely touch it. That's doesn't seem to be your problem though...

4

It's equally 'interesting' that all 4 cores report exactly the same temperature. I'd like to see proof from the manufacturer that each core has an individual temperature sensor. For this passively cooled device I've a suspicion that only one actual sensor exists.

5

Yes it looks like the BSD links a "temperature" property to each instance of "cpu," though they all read from the same single sensor.

6

I have wondered the same. Running stress on a single core may show the differences. I dont have my pfsense build anymore and my kid is here for several days so I cant really take everything down to test the theory.

7

I have this on the 19.07 tree (seems only to probe a single core as well):

# sensors 
fam15h_power-pci-00c4
Adapter: PCI adapter
power1:        6.53 W  (interval =   0.01 s, crit =   6.00 W)

k10temp-pci-00c3
Adapter: PCI adapter
temp1:        +54.6°C  (high = +70.0°C)
                       (crit = +105.0°C, hyst = +104.0°C)