Determining CPU Temp?

Reading the docs for SpeedFan, I get the impression that a
third-party app cannot tell for sure which temperatures are
which.

Is there some GigaByte-specific utility that will tell me the CPU
temp?

EP45-UD3L.
--
Pete Cresswell

never had any problems with speedfan.

always give cpu/gpu temps and fan speeds.

do have a giagbyte mobo.

"(PeteCresswell)" wrote in message ...
Reading the docs for SpeedFan, I get the impression that a
third-party app cannot tell for sure which temperatures are
which.

Is there some GigaByte-specific utility that will tell me the CPU
temp?

EP45-UD3L.
--
Pete Cresswell

(PeteCresswell) wrote:
Reading the docs for SpeedFan, I get the impression that a
third-party app cannot tell for sure which temperatures are
which.

Is there some GigaByte-specific utility that will tell me the CPU
temp?

EP45-UD3L.

See if there is a Easy Tune 5 or 6 available for your mb.
http://www.gigabyte.com/support-downloads/Utility.aspx

Per mikeyhsd:
always give cpu/gpu temps and fan speeds.

Yes, but the developer points out that there is no way for him to
determine which temp is which.
--
Pete Cresswell

(PeteCresswell) wrote:
Reading the docs for SpeedFan, I get the impression that a
third-party app cannot tell for sure which temperatures are
which.

Is there some GigaByte-specific utility that will tell me the CPU
temp?

EP45-UD3L.

There are a couple ways to read temperature. The SuperI/O chip
might have three thermal channels. There can be ambiguity, about
which channel is which. The speed of change in measured temperature,
sometimes helps resolve which is which.

Intel processors, past a certain generation, have PECI and on-die
digital temperature measurement. This utility reads that out.

http://www.techpowerup.com/realtemp/

The sensor in the CPU, measures a delta_T. The delta_T is relative to
TjMax, which is also used for other temperature control purposes
(throttling or THERMTRIP, that kind of thing). In the past, there
were disagreements about what starting value, the delta_T should be
measured against. For example, if TjMax is 100C and delta_T is 2C,
the actual junction temperature is 98C. If a person somehow got the
wrong TjMax, then the computed result of 98C could be wrong.

Junction temperature is different than case temperature. In the past,
the crudest way of checking the processor, was to put a kapton film
thermistor underneath the socket area, to measure the bottom of the
CPU. Whereas the PECI digital interface measuring the junction, will
get one value, the case temp will be 25-35C cooler. Even the thermal
specifications for the processor, can change from one generation to the
next. On an older processor, you might be warned to "stay below 72C".
Then, in a current generation one, you see "stay below 100C". The difference
is, the 72C one was Tcase referenced. While the 100C one could be
Tjunction based. So some people, scare the crap out of themselves, by
using the digital temperature measurement, and then remembering back
to the "72C days".

In terms of sensing methods, there is

1) Thermistor method - the thermistor resistance changes with temperature
- can be accurate to 1 degree C. Accuracy varies at
temperature extremes.
- always hard to get the thermistor, near the heat source
Motherboard manufacturers applied "fudge factor" to correct
the readout to give "reasonable" values.

2) Diode method - the diode equation relates current and voltage, and
a careful measurement can determine the temperature. The recommended
measurement method is two constant current sources, alternated, which
apparently removes some systematic error. Most implementations instead
use a single fixed current value, and to hell with errors.

The diode method can be:

1) Done with a diode, or with the junction of a transistor. On a transistor
with three legs, you could use two of the legs to build a sensor.
2) On previous generation processors, there was a diode right on the
silicon die, and the anode and cathode were pinned out on the CPU
socket. And that allowed the SuperI/O to read out the temp, using
the diode method. (The SuperI/O has a register, to set the thing for
diode-like voltages or thermistor readings. Some SuperI/O even have a
lookup table, so no math is required on readout.)
3) The Intel PECI/digital method uses the diode idea, and does the
measurement in an Intel chip. So no SuperI/O is needed for that as
far as I know.

It's always fun to collect as many utilities as possible, then test
them one at a time with controlled test conditions. Just to see, which
ones are full of crap :-)

Have fun,
Paul

The Gigabyte utility appears to read the socket sensor. CoreTemp reads
the cpu itself. No guarantee of accuracy for each; the socket should
read lower, but if they get close during a torture test than it's
encouraging.

You could compare their readings to those in SpeedFan to maybe determine
which sensor is which.
--
Ed Light

Better World News TV Channel:
http://realnews.com

Iraq Veterans Against the War and Related:
http://ivaw.org
http://couragetoresist.org
http://antiwar.com

Send spam to the FTC at

Thanks, robots.

you can reboot, go into the bios and see which is which, then compare them when booting and looking at speedfan.

"(PeteCresswell)" wrote in message ...
Per mikeyhsd:
always give cpu/gpu temps and fan speeds.

Yes, but the developer points out that there is no way for him to
determine which temp is which.
--
Pete Cresswell