GoogleCodeExporter
commented
9 years ago http://www.spinics.net/lists/lm-sensors/msg37308.html
"No. Problem is that PECI does not report an absolute temperature, but the
difference to Tjmax, which is the maximum CPU temperature (reported as critical
temperature by coretemp). The NCT6776F has a register which needs to be
programmed
to that value. Usually that is done in the BIOS. Looks like the BIOS did not
program the correct value on your board."
http://www.motherboardpoint.com/exactly-pecl-agent-temp-opposed-cpu-temp-t166984
.html
http://askubuntu.com/questions/373154/sensors-command-from-lm-sensors-keeps-repo
rting-the-same-temperature
http://en.wikipedia.org/wiki/Platform_Environment_Control_Interface
"From a control standpoint, the main difference between PECI and the previously
used thermal monitoring methods is that PECI reports a negative value
expressing the difference between the current temperature and the thermal
throttle point (at which the CPU reduces speed or shuts down to prevent damage
due to overheating) instead of the absolute temperature. For example, for a CPU
with maximal temperature of 85 °C and a current temperature reading of 35 °C,
the value reported by PECI would be −50 °C."
I don't believe there's a single solution to your issue. Some users report PECI
being wrong, too cold, then there's that above statement about what it actually
reports. I assume lm-sensors takes that and processes it back to something
closer to reality in terms of raw temp, not diff of temps.
I can't fix one set of data for this without breaking another.
More data is required, sorry. For your specific case, it would fix it, and be
more accurate, but for another case, it would break it, and make it wrong to do
that fix.
sensors are a total mess, always have been, and I assume always will be.
Note sure how to handle this one to be honest.
I can see one case where if temp is not set by the time it reads down to PECI
then use PECI, that would solve one problem of no CPU temp unless PECI.
This needs more data, not just specific failures by individual motherboard
models to do their bios right. ie, a solution that works for almost everything,
or at least more than most.
Note that of the samples I saw online, most PECI and CPUTIN values were so
close as to be essentially the same, which argues against switching to a
default PECI, particularly if the reports of PECI showing 10 degrees too cold
are also correct.
Original comment by inxi-...@techpatterns.com on 28 Apr 2014 at 1:47
- Changed state: Accepted
Original issue reported on code.google.com by
inkasso....@gmail.comon 27 Apr 2014 at 6:18