sysctl question

[mfv]

Hello William,

I had the same problem using amdtemp but used the sysctl setting
"dev.amdtemp.%d.sensor_offset" to make it more accurate. The value was
determined by comparing the unadjusted value of admtemp with the value
shown in the BIOS.  In my case the setting in /etc/sysctl.conf is:

 dev.amdtemp.0.sensor_offset=24

A room thermometer was also helpful to measure the difference by
comparing the difference between the room temperature and the BIOS
after the CPU was placed under load.

To read the value I then bound the F2-key in csh.cshrc file as follows:

  bindkey -c "^[OQ" 'sysctl -e dev.amdtemp.0.core0'

The results of the reading are not as accurate as I would like but are
an improvement.

    Cheers ...                                 __o
                                             _ \<,_
    Marek                                   (+)/ (+)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

On Wed, 13 May 2015 22:08:57 -0453
"William A. Mahaffey III" <wam at hiwaay.net> wrote:

> 
> 
> My sysctl is reporting nonsense temps for my CPU:
> 
> [wam at kabini1, ~, 10:04:02pm] 357 % sysctl -A | egrep
> '(temperature|usage)' dev.cpu.0.temperature: 10.6C
> dev.cpu.0.cx_usage: 100.00% 0.00% last 445us
> dev.cpu.1.temperature: 10.6C
> dev.cpu.1.cx_usage: 100.00% last 11us
> dev.cpu.2.temperature: 10.6C
> dev.cpu.2.cx_usage: 100.00% last 21us
> dev.cpu.3.temperature: 10.6C
> dev.cpu.3.cx_usage: 100.00% last 9us
> [wam at kabini1, ~, 10:04:10pm] 358 % sysctl -a | grep -i model
> hw.model: AMD Sempron(tm) 3850 APU with Radeon(tm) R3
> dev.rgephy.0.%pnpinfo: oui=0xe04c model=0x0 rev=0x0
> [wam at kabini1, ~, 10:04:14pm] 359 % uname -a
> FreeBSD kabini1.local 9.3-RELEASE-p13 FreeBSD 9.3-RELEASE-p13 #0: Tue 
> Apr  7 03:01:12 UTC 2015 
> root at amd64-builder.daemonology.net:/usr/obj/usr/src/sys/GENERIC amd64
> [wam at kabini1, ~, 10:05:08pm] 360 %
> 
> 
> Room temp is about 75F, or ~23C. CPU is generating heat, 
> thermodynamically impossible for it to be cooler than room temp. 
> Obviously not critical, but I would like to know reasonably
> accurately how hot the CPU is in real time. Is this a sufficient bug
> report ? TIA ....
> 
>