QX9650 Temps and RealTemp

[Kreij]

Okay, RealTemp reports the following temps at idle ..

Core 0 : 19C
Core 1 : 22C
Core 2 : 27C
Core 3 : 38C

Test Sensors is as follows (Movement) ..

Core 0 : 4
Core 1 : 6
Core 2 : 7
Core 3 : 1

A stuck sensor? What exactly does that mean to me?

 

[erocker]

Thankfully only one is stuck! I would just use the other cores as a refference for the core that is unreadable. It's strange, many people with bad thermal sensors generally have a very good clocking CPU. My 5000+ BE for example, I'm not sure if there even is core sensors, but it OC's extremely well, and I just go by the CPU socket temp.

 

[bobseptic]

whats your tjmax set too ?

ive read loads and most people believe its 95 for the qx9650

lapped thermalright ultra 120 eXtreme idle

Core 0 : 30C
Core 1 : 30C
Core 2 : 20C
Core 3 : 19C

 

[Kreij]

With RealTemp I am assuming 95C, as CoreTemp is about 10C higher.

With CoreTemp I get (reporting 105 TjMax) ...
Core 0 : 29
Core 1 : 32
Core 2 : 38
Core 3 : 48

 

[Kreij]

I'm not sure what a "stuck" sensor means.
If I put it underload should that core be lower than the others because it is stuck?
Or is it stuck at a higher temp?

 

[CrAsHnBuRnXp]

Prime and find out.

 

[unclewebb]

Most of these sensors work fine but as the temperature drops, some Intel sensors stop decreasing. They get stuck. Your core3, even if you brought your computer to the North Pole, will probably never read less than about 37C or 38C depending on where it gets stuck at. The sensor is still fine for what Intel designed it to do and that's to trigger thermal throttling near TjMax. Stuck sensors can not be depended on to give accurate idle temperatures but other than that most of these processors still run fine and overclock just as much as another processor with sensors that don't get stuck so there's no need to worry about it.

As the core temperature increases, most stuck sensors will start moving upward along with the rest of them.

 

[Kreij]

Most of these sensors work fine but as the temperature drops, some Intel sensors stop decreasing. They get stuck. Your core3, even if you brought your computer to the North Pole, will probably never read less than about 37C or 38C depending on where it gets stuck at. The sensor is still fine for what Intel designed it to do and that's to trigger thermal throttling near TjMax. Stuck sensors can be depended on to give accurate idle temperatures but other than that most of these processors still run fine and overclock just as much as another processor with sensors that don't get stuck so there's no need to worry about it.

You da Man

I reseated my cooler three times without any change in idle temps. I didn't want to start lapping an $1000 CPU (makes me a little nervous). Never had this problem with my QX6700 and never knew that sensors could get "stuck".

Thank you very much.

... and your utility is excellent.

 

[unclewebb]

Thanks.

I left a word out of my original post:

"Stuck sensors can NOT be depended on to give accurate idle temperatures...."

I think the previous 65nm sensors could also get stuck but they only got stuck at such low temperatures that the majority of people never noticed. It was mostly only the hard core guys running extreme cooling that were running many degrees below zero. These sensors have a theoretical minimum of about -32C but very few of them are capable of going this low before sticking.

 

[Kreij]

I am still not sure whats up.
With my QX6700 all cores were within about 5C of each other with the same cooler.
I have not tried my old cooler, which is a great cooler, but not quite as good as the Nirvana.

It just seems odd that there is a definite temp climb from 0 to 3.
If it was a sensor I would think that 3 cores would be good and one would be off.

Or is that faulty logic?

 

[unclewebb]

The sensors on some 45nm Quads are more like random number generators. As delivered, these sensors are not accurate at reporting idle temperatures so don't put too much faith into what they're saying at idle. All your sensors are off.

If you want more accurate idle temperatures then read the RealTemp docs and try doing the calibration. Don't use a calibration factor for core3 since it is stuck. After calibration, these sensors can give you very accurate temps from idle to TjMax. Just reading about the problems these sensors can have will give you a much better understanding of what they can do and their limitations.