Incorrect CPU temps in Precise (Solved)

Message

Mr. Hughes · #1 Post by **Mr. Hughes** » Fri 03 Oct 2014, 02:42

I can't seem to get a proper CPU temperature reading in precise 5.7.1.

The k10temp reading in hardinfo is very erratic and randomly jumps anywhere from half a degree celsius to 20 degrees celcius.

Its the A6-6400K 3.9ghz APU in my signature.

#2 Post by **Flash** » Fri 03 Oct 2014, 04:19

Does the same thing happen when viewing the CPU temp in your BIOS? (Assuming your BIOS shows the CPU temp.)

bigpup · #3 Post by **bigpup** » Fri 03 Oct 2014, 05:56

Hardinfo may not be updated to provide good support for your hardware.

Try this:
Cputemp program
http://www.murga-linux.com/puppy/viewto ... 546#432546

Mr. Hughes · #4 Post by **Mr. Hughes** » Fri 03 Oct 2014, 07:22

@Flash My bios always shows the correct temp.

@bigpup I actually already tried that program and it gave 2 temperature readings, one of them was incorrect and the other said 51 degrees which sounds about right.

That program doesn't seem to show me the temp in real-time though. I have to keep clicking the flame icon each time I want an up-to-date reading it seems?

Moat · #5 Post by **Moat** » Fri 03 Oct 2014, 07:34

Have you tried this tray applet? http://www.murga-linux.com/puppy/viewtopic.php?t=74702

Bob

Mr. Hughes · #6 Post by **Mr. Hughes** » Fri 03 Oct 2014, 08:01

I also tried that one, it gave a temp reading between 10 and 13 degrees celsius. That would seem an incorrect value.

That program is really cool though if it worked.

EDIT: not sure if this will help but these are the results of running the sensors command in a terminal:

k10temp-pci-00c3
Adapter: PCI adapter
temp1: +1.4 C (high = +70.0 C)
(crit = +70.0 C, hyst = +69.0 C)

radeon-pci-0100
Adapter: PCI adapter
temp1: +51.0 C

#7 Post by **Flash** » Fri 03 Oct 2014, 12:59

70 C seems pretty warm. Is that also what your BIOS says?

Mr. Hughes · #8 Post by **Mr. Hughes** » Fri 03 Oct 2014, 17:04

If you read carefully 70 degrees isn't the actual temperature, but its telling you the Max temp, the critical temp, and the hysteresis temp.

The temp reading is where it says 1.4 degrees which is obviously wrong haha.

Proudog · #9 Post by **Proudog** » Fri 03 Oct 2014, 18:51

I also tried that one, it gave a temp reading between 10 and 13 degrees celsius. That would seem an incorrect value.

This very common and not is a issue, the BIOS of your PC is calibrating to always see a aprox. temp; You need re-calibrating this for lm-sensors. More info in:

https://wiki.archlinux.org/index.php/Lm ... and_Tricks

In windows 7, this simple don't show the temp CPU and fan speeds

Mr. Hughes · #10 Post by **Mr. Hughes** » Fri 03 Oct 2014, 18:56

In order to calibrate lm sensors, I need to find the config file in puppy. Any idea where it is in precise? Its not in the etc folder.

I have no problems in win 7, but I use puppy most of the time.

Mr. Hughes · #11 Post by **Mr. Hughes** » Fri 03 Oct 2014, 19:10

Create the following file overriding the default values:

/etc/sensors.d/Zotac-IONITX-A-U

chip "coretemp-isa-0000"
label temp2 "Core 0"
compute temp2 @-20,@-20

label temp3 "Core 1"
compute temp3 @-20,@-20

I think this is what I need to do but I think I need it in english

How would I do this in puppy?

EDIT: This motherboard has an advanced graphical bios with mouse support. The temp slowly climbs to 51 degrees and then stops while in the bios.

It doesn't randomly go back and forth like it does in puppy.

Mr. Hughes · #12 Post by **Mr. Hughes** » Fri 03 Oct 2014, 19:27

If I type
#rmmod k10temp
# modprobe k10temp force=1

it starts working for a little bit then stops

Proudog · #13 Post by **Proudog** » Fri 03 Oct 2014, 20:54

How would I do this in puppy?

Is not easy for average user, you need first calculate who is the offset of error (Maybe you need a external thermometer)

The commands:

#sensors
sensors is used to show the current readings of all sensor chips, and to set all limits as specified in the configuration file.

#sensors -u
Treats all chips as unknown ones. Output will be of much lower quality; this option is only added for testing purposes. (please show the output)

#sensors -f
Prints the temperatures in degrees Fahrenheit instead of Celsius.(please show the output)

The configuration file is in:
/etc/sensors.conf

And you add lines below for -20 offset example:

chip "[This is variable]"
label temp2 "Core 0"
compute temp2 @-20,@-20

label temp3 "Core 1"
compute temp3 @-20,@-20

Mr. Hughes · #14 Post by **Mr. Hughes** » Fri 03 Oct 2014, 21:11

Well this is odd, the sensors command is no longer working in the terminal. I must've screwed something up.

Like I said though forcing the k10temp module to load makes it display the correct temp of 49 to 51 degrees, but only for a few seconds.

Not sure how to get it to stay loaded.

I set it to force the module on startup and that didn't seem to work either.

I think I might just give up on this one...

Proudog · #15 Post by **Proudog** » Fri 03 Oct 2014, 21:26

In Google .

k10temp

This driver permits reading of the internal temperature sensor of AMD
Family 10h/11h/12h/14h/15h/16h processors.

All these processors have a sensor, but on those for Socket F or AM2+,
the sensor may return inconsistent values (erratum 319). The driver
will refuse to load on these revisions unless you specify the "force=1"
module parameter.

Due to technical reasons, the driver can detect only the mainboard's
socket type, not the processor's actual capabilities. Therefore, if you
are using an AM3 processor on an AM2+ mainboard, you can safely use the
"force=1" parameter.

There is one temperature measurement value, available as temp1_input in
sysfs. It is measured in degrees Celsius with a resolution of 1/8th degree.
Please note that it is defined as a relative value; to quote the AMD manual:

Tctl is the processor temperature control value, used by the platform to
control cooling systems. Tctl is a non-physical temperature on an
arbitrary scale measured in degrees. It does _not_ represent an actual
physical temperature like die or case temperature. Instead, it specifies
the processor temperature relative to the point at which the system must
supply the maximum cooling for the processor's specified maximum case
temperature and maximum thermal power dissipation.

The maximum value for Tctl is available in the file temp1_max.

If the BIOS has enabled hardware temperature control, the threshold at
which the processor will throttle itself to avoid damage is available in
temp1_crit and temp1_crit_hyst.

Mr. Hughes · #16 Post by **Mr. Hughes** » Fri 03 Oct 2014, 21:32

Thanks, i'll read more into later. Gotta leave for work for now.

#17 Post by **Flash** » Sat 04 Oct 2014, 03:19

Just for general reference, the CPU temperature is probably measured using PN diode junctions included in the layout of the IC for just that purpose. See Silicon bandgap temperature sensor for more detail.

Mr. Hughes · #18 Post by **Mr. Hughes** » Sat 04 Oct 2014, 07:27

Interesting read flash, I get the gist of it but I can't say i'm very good at math equations, or math in general for that matter

Anyway I think I found an answer that explains this all.

Core Temp Displays Unrealistic Values
http://www.lm-sensors.org/wiki/FAQ/Chap ... sticvalues

Apparently there is no real solution, its a hardware limitation of k8 and k10 procs, which is why I was pulling my hair out and nothing was working hehe

Hope I didn't waste to much of your time. I shoulda just did some more research.

(old)Puppy Linux Discussion Forum

(old)Puppy Linux Discussion Forum

Incorrect CPU temps in Precise (Solved)

Incorrect CPU temps in Precise (Solved)