The Truth about Temperatures and Voltages - Page 3 - Overclock.net

tbates1244 · 03-16-09

Thank you kpo6969, I was getting ready to tell people about the awesome testing done by unclewebb and rge over at the REal temp thread on XS forums.

Here is the other thing everybody needs to keep in mind; The temperature at which your chip will throttle will vary from CPU to CPU, even with CPU's of the same exact model and VID. It can even vary from core to core. This is because, as the OP probably stated already, each chip is calibrated individually, thus making all the sensors and calibrations vary slightly from chip to chip. One Q6600 may have a TJMax of 95 another might be 97.

ChickenInferno · 03-16-09

Quote:

Originally Posted by tbates1244

as the OP probably stated already, each chip is calibrated individually, thus making all the sensors and calibrations vary from chip to chip. One Q6600 may have a TJMax of 95 another might be 97.

The point of my post was to keep it very simply and easily understood. The point that you're alluding to is that what I have explained isn't an exact number but the numbers reported in post #1 are actually TJ Max Target values. These are the numbers that Intel was aiming for with each proccesor and most chips are close to the target.

The problem with calibrating to idle/stock temps/clocks is that you immediately introduce mass amounts of uncertainty into the readings. The article lists off possible values for temperature readings based on cooling type, but there are many reports of exceptionally "hot batches" of cpus, which would idle much higher than ambient temperature not due to the cooling but the thermal wattage dissipation. By saying that my cpu should idle around 8C higher at stock idle than ambient, I am already guessing at the capabilities of my cooler and at the heat it is dissipating. Depending on the cooler and the chip the uncertainty would just vary too much for my taste (Could be very accurate or could very inaccurate).

Intel documentation also alludes to the idea that distance to TJ Max is not reliable as it decreases. Thus, the documentation says that below 50C temperatures are saturated and considered inaccurate, but near TJ Max they are accurate. It is possible that you are setting an accurate temperature low, but that due to the non-linearity of the distance to TJ Max (non-linearity alluded to not specifically stated) you have actually set the load temperatures far from actual temps, but fairly accurate idle temps. This "calibration" may only be calibrating to lower temps and not higher temps. The only way to successfully test this would be to put a thermal diode on the cores themselves.

spacegoast · 03-26-09

great explanation

. i saw that chart intel released but i was unsure if it was accurate or not. you summed it up nicely though

MasterKromm · 03-26-09

This should be stickied regardless of preference or belief...

Personally I'd love to test this further myself, but I'm terrified of carving out a thin channel in the IHS to insert a thermal probe. Instead I might just buy an IR thermometer and play with that first.

Quote:

The problem with this argument is that Intel has implied that the Distance to TJ Max values are not linear, and therefore you would have accurate idle temps (not important) and very inaccurate load temps (very important).

It would be interesting if there were a way to have TJmax work in a variable manner, where realtemp applies the newly calibrated TJmax when temps are around a specific threshold, yet when that threshold is broken the original TJmax is used...

ChickenInferno · 03-26-09

Quote:

Originally Posted by MasterKromm

It would be interesting if there were a way to have TJmax work in a variable manner, where realtemp applies the newly calibrated TJmax when temps are around a specific threshold, yet when that threshold is broken the original TJmax is used...

I'm sure that a program could be written with a very simple modification like a while wend loop. For example, if the calibration TJ Max was found to be 99C, then as long as the distance is far (value is innaccurate) the calibrated temp would be correct. When the distance becomes less than 40, the values become more accurate and therefore you could use the Intel numbers.

Quote:

While 1=1
.......While Dist TJ Max>40
...............99=TJ Max
.......Wend
.......90=TJ Max
Wend

I don't pretend to be that good at programming, but this seems like it would work.

almighty15 · 03-26-09

Ive set the TJmax now in coretemp and HWmonitor and my temps look awesome.

I dunno weather to trust them though, i dont have a proper way of measuring my ambient tempreture but i just checked my central heating thermostat and it clicked 'on' at 18c so i assume that my ambient temp is around 18c. The stat is in another room so it could be colder or warmer in the computer room ( Feels colder IMO )

HWmonitor and coretemp report my idle temp at 18-19c

Are they right?

ChickenInferno · 03-26-09

Quote:

Originally Posted by almighty15

Ive set the TJmax now in coretemp and HWmonitor and my temps look awesome.

I dunno weather to trust them though, i dont have a proper way of measuring my ambient tempreture but i just checked my central heating thermostat and it clicked 'on' at 18c so i assume that my ambient temp is around 18c. The stat is in another room so it could be colder or warmer in the computer room ( Feels colder IMO )

HWmonitor and coretemp report my idle temp at 18-19c

Are they right?

See the following

Quote:

Originally Posted by ChickenInferno

TJ Max is an inherently unreliable and inaccurate method of measuring idle/low temperatures. Due to the way it was designed, the readings are so inaccurate under 50C that Intel says they can only be read as a number temperature below 50C. As the temperature approaches TJ Max, the precision increases and at TJ Max the temperature is considered to be 100% accurate. Because of this error at low temperatures, sub-ambient temperature/very high readings are sometimes given for idle temperatures. With the TJ Max method, your idle temperatures have no accuracy and therefore should be ignored. Under load, the temperatures become much more accurate and should be very carefully monitored.

That should answer your question. The inaccuracy at low temps is too high to say with any certainty what the temperature really is. If you are worried about your load temps, then you can buy a thermometer and measure the temp of the air being pushed away from the Heatsink. It won't be exactly accurate, but if core temp is saying 55C and the air temp is 90C, then you would know there is a problem.

almighty15 · 03-26-09

Quote:

Originally Posted by ChickenInferno

If you are worried about your load temps, then you can buy a thermometer and measure the temp of the air being pushed away from the Heatsink. It won't be exactly accurate, but if core temp is saying 55C and the air temp is 90C, then you would know there is a problem.

Im on water

ChickenInferno · 03-26-09

Measure the water temp with a probe after it has gone through the cpu block, or you can use something like this which will accurately measure the temps to the order of plus or minus 1 from -10C to 100C.

alex98uk · 03-26-09

I had always set my TJmax for my Q6600 G0 as 95c, but this seemed correct idle, but too high under load. Setting 90c is kind of what i wanted/expected from my watercooling setup.

From what you're saying, don't trust any idle temp readings as they could be hugely inaccurate, but load temps should be within a few degrees of what the programs readout?

		Overclock.net - Overclocking.net > Intel > Intel CPUs
The Truth about Temperatures and Voltages

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)