* Re: [lm-sensors] coretemp readings are static/constant
2013-11-19 6:28 [lm-sensors] coretemp readings are static/constant Tom Metro
@ 2013-11-19 7:57 ` Jean Delvare
2013-11-29 21:41 ` Tom Metro
2013-11-29 21:56 ` Jean Delvare
2 siblings, 0 replies; 4+ messages in thread
From: Jean Delvare @ 2013-11-19 7:57 UTC (permalink / raw)
To: lm-sensors
Hi Tom,
On Tue, 19 Nov 2013 06:28:53 +0000 (UTC), Tom Metro wrote:
> Running:
>
> % sensors -v
> sensors version 3.3.1 with libsensors version 3.3.1
>
> on Ubuntu 12.04 on an Acer Aspire 8930G laptop with:
>
> % cat /proc/cpuinfo | fgrep name
> model name : Intel(R) Core(TM)2 Duo CPU P8400 @ 2.26GHz
> model name : Intel(R) Core(TM)2 Duo CPU P8400 @ 2.26GHz
>
> and I see:
>
> % sensors
> acpitz-virtual-0
> Adapter: Virtual device
> temp1: +38.0°C (crit = +106.0°C)
> temp2: +48.0°C (crit = +106.0°C)
> temp3: +48.0°C (crit = +106.0°C)
> temp4: +43.0°C (crit = +106.0°C)
>
> nouveau-pci-0100
> Adapter: PCI adapter
> temp1: +49.0°C (high = +100.0°C, crit = +110.0°C)
>
> coretemp-isa-0000
> Adapter: ISA adapter
> Core 0: +33.0°C (high = +105.0°C, crit = +105.0°C)
> Core 1: +33.0°C (high = +105.0°C, crit = +105.0°C)
>
> The values reported by coretemp are always 33000 millidegrees:
>
> % sudo cat /sys/devices/platform/coretemp.0/temp2_input
> 33000
>
> as observed over the course of a week. I have not tried to intentionally
> stress the CPU, but had assumed that if it was reading real values, I would
> have seen some variation.
>
> However, I read in the FAQ:
> http://lm-sensors.org/wiki/FAQ/Chapter3#coretempreturnsunrealisticvalues
>
> It isn't really returning degrees Celsius. At high temperatures, the
> (small) thermal margin is almost expressed in degrees Celsius, but at low
> temperature, the (high) thermal margin is no longer expressed in actual
> degrees Celsius. So, if the temperature value reported by coretemp is
> unrealistically low, all it means is that you are far away from the
> critical limit so your systems are running totally fine and cool...
>
> So does that mean it is expected to see a constant (to 5 decimal places)
> value, until the cores heat up beyond some threshold? (If so, I'll add this
> to the FAQ.)
Don't get fooled by the 5 decimal places, all drivers return values in
millidegrees Celsius regardless of their accuracy and resolution. The
coretemp resolution is 1 degree so you'll always see 000 as the last 3
digits.
You should really stress the CPU and see what happens. With really good
cooling and no CPU load at all, it is possible for temperatures to stay
stable within one degree Celsius.
Another possibility is that this particular incarnation of the sensors
do not report values below 33°C. We know these sensors have poor
accuracy at low temperatures, maybe Intel decided to not expose too low
values to make them look plausible. In that case you'd see 33°C most of
the time, as a floor measurement, with temperature peaks on high CPU
load.
--
Jean Delvare
_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [lm-sensors] coretemp readings are static/constant
2013-11-19 6:28 [lm-sensors] coretemp readings are static/constant Tom Metro
2013-11-19 7:57 ` Jean Delvare
@ 2013-11-29 21:41 ` Tom Metro
2013-11-29 21:56 ` Jean Delvare
2 siblings, 0 replies; 4+ messages in thread
From: Tom Metro @ 2013-11-29 21:41 UTC (permalink / raw)
To: lm-sensors
Hi Jean,
Jean Delvare wrote:
> Tom Metro wrote:
>> % sensors
>> [...]
>> coretemp-isa-0000
>> Adapter: ISA adapter
>> Core 0: +33.0°C (high = +105.0°C, crit = +105.0°C)
>> Core 1: +33.0°C (high = +105.0°C, crit = +105.0°C)
>>
>> The values reported by coretemp are always 33000 millidegrees:
>
> You should really stress the CPU and see what happens.
I ran:
% stress --cpu 2 --timeout 7200
stress: info: [14967] dispatching hogs: 2 cpu, 0 io, 0 vm, 0 hdd
stress: info: [14967] successful run completed in 7200s
And just to be sure, I also thew in a couple of these:
% dd bs=10 if=/dev/urandom | bzip2 -z -s -9 -c > /dev/null &
and let the latter run for 8 hours.
I saw some new peaks on the 'acpitz-virtual-0' sensors, but the
'coretemp-isa-0000' sensors didn't budge.
> Another possibility is that this particular incarnation of the sensors
> do not report values below 33°C.
Right, that was the theory I was going on based on the FAQ entry I
quoted. Apparently not the case here.
So this would appear to be a problem with the coretemp kernel module.
This CPU (Intel Core 2 Duo P8400) is from the right era to be supported
by coretemp. Where is the best place to go to file a bug against that
module?
Thanks for the suggestions.
-Tom
_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
^ permalink raw reply [flat|nested] 4+ messages in thread