Real Temp - New temp program for Intel Core processors

[jaredpace]

Some feedback for unclewebb on RealTemp from an AT user:

"I do not have a E8xxx but have a E6400 L2. He also used this CPU as a comparison.
It seems Real Temp thinks my CPU has a Tjmax of 85C and not 100C.
Everything I have read says the E6400 has a Tjmax of 100C.
Now E6400's in general seem to run hotter than say a E6300 at the same clock speed and voltage. Both CPU's have a TJmax of 100C as far as I can tell.
This has always bugged me. Something is not being accounted for.

I have checked and calibrated speedfan's reported Tcase and Tjunction temps using Computronix Guide here. http://www.ocforums.com/showthread.php?t=543522
Now this guide could be all wrong but it seems to make logical sense and the results make sense. Tcase calibration would seem to be almost infallible IMO, at least then the CPU is at idle. Tjunction temps are a gray area do to Intel not releasing compete info.

Now I believe I can trust the Tcase temp reported by Speedfan at idle for sure because of the method of calibration in Computronix guide. Load temps are more questionable because there is no easy way to check accuracy. I'm not using a Nvidia based mother board which is known to report Tcase temp in a non linear way. So I'm going to just trust Tcase temps reported by Speedfan to be accurate. I also believe Tcase to Tjunction delta's on E6400's are 15C +-3C as the guide claims. Now Real Temp set to ++ as Unclewebb recommends on a E6400 reports Tjunction temps way to close to speedfans reported Tcase temp at idle IMO. Now Unclewebb claims E6400 CPU's under report Tjunction temps at idle but using Real Temp set as he specifies for a E6400 reports lower Tjunction temps than any other program. Speedfan (calibrated), Coretemp, HWmonitor etc. I just feel something is wrong here.

Looking at load temps Real Temp reports Tjunction temps lower than Speedfan reports Tcase temp. This is not possible so one or the other is wrong. I still wonder what the true Tjmax temp is on a E6400 really is? Load temps sure look nice using Real Temp compared to other programs but something just does not fell right.

Some screen shots, right click and choose view it they are hard to read.
Idle
http://i156.photobucket.com/al...erwrench/Misc/Idle.png
Load
http://i156.photobucket.com/al...wrench/Misc/load-1.png


In the end I say not enough is known about Intel's Core2 CPU's to be able to trust temps reported by any method. I think Unclewebb is on to something but too many pieces of the puzzle are still missing."

[Slay0r]

Here you go unclewebb

200*6 1.1v fan@full

calibration @ 0

load


idle


calibration @ ++

load


idle


"++" seems to give reasonable readings for this processor, "0" is deffinately off as ambient temperature in this room is ~20°C.. whaddya think?

[unclewebb]

It's always been my theory that the E4300 has a TjMax=85C and that you would need a ++ calibration adjustment to bring the idle temperatures in line with reality and helps confirm my theory that at this speed and voltage they should run at about 4C or 5C above ambient. Your data looks believable. I need to find a cheap L2 to put under the IR gun to help confirm.

jaredpace: CPU diode based temps reported by motherboards are rarely 100% right. You can't assume anything in this and can't just assume that what SpeedFan reports must be right for the CPU or any other piece of software. I know personally my CPU diode based temp on my P5B can change by 10C depending on whether I boot up or do a resume from Stand By. I definitely don't trust anything that sensor tells me because it is not accurate. Try dropping your CPU down to a low wattage setting like Slay0r did and report what it says.

[idiotekniQues]

exactly 5 degrees less than core temp both idle and under load.

[idiotekniQues]

hey slay0r that is a nice desktop wallpaper. where did u get it?

[unclewebb]

ken: For the previous generation B2 E6x00 chips, both CoreTemp and RealTemp use the same TjMax=85C so there'll be no difference.

[jaredpace]

@ unclewebb:

RealTemp looks like excellent software! good job and congrats!

Do you plan to do new revisions & updates?

[Infa]

yeah but you got an old Conroe, the problems we have are with the new penryns

[ixtapalapaquetl]

Below I have compiled some statistical information which strongly supports the use of Real Temp over the use of Core Temp. This information also corroborates unclewebb's statements regarding the inaccuracy of DTS sensors at lower temperatures (be they read with Core Temp or Real Temp). For more information on this analysis of Core Temp and the methodology involved, see http://www.xtremesystems.org/forums/...d.php?t=178232.

Let me state at the outset that it is my presumption is that CPU Temp as reported by my motherboard sensor is at least relatively accurate (that is to say, if CPU Temp is reading incorrectly, it does so consistently across its entire range of measurement - always 10C too high, for example). It may or not be absolutely accurate (that is to say, it always reads correctly without the need for a correction factor). However, the following does not require absolute accuracy to be meaningful, only relative accuracy, and thus I believe I am on solid ground.



In the above graph, you see the gross differences between Core Temp and CPU Temp. These differences are especially pronounced at lower temperatures. I have observed similar differences in the past at even higher temperatures.

Now consider the following results using Real Temp. (Note that there are far fewer data points because I had to throw out hundreds of values that were out of synch.)



Here we see a much different picture. We first notice that at load, there is virtually no difference between Real Temp and CPU Temp - this, I think, is a very revealing statistic! We also see that there is a greater difference at idle, which supports unclewebb's premise that the DTS sensor is unreliable at lower temperatures. Regardless, there is far less discrepancy in this graph compared to the one above it.

A brief aside: Some argue that the CPU Temps and DTS values are unavoidably different because the locations of their respective sensors (ie geometric center of die versus, well, versus anything else). This has always bothered me. If the material in question conducts heat well, as a CPU die surely must, it seems obvious that it would only be a matter of microseconds until that heat was transmitted to any other measuring point. Any observed differences in these temperatures strike me more as a way to determine the relative error in CPU Temps than as some condemnation of the sensor itself. In at least this board with this CPU, the relationship appears to be very close. In any event, I encourage you to draw your own conclusions regarding the graphs above.

The tables below tell a similar tale. Here are the average temperatures (I included all data points in these calculations since simultaneity was irrelevant; more on that in a moment):





Here we note the same trends as exhibited by the graphs above. Note the near identical figures for CPU Temp and Real Temp at load. Also note the wider variation in Core Temp readings (as shown by the larger standard deviation); this speaks further of Core Temp's unreliability (though not terribly loudly).

Finally, a look at the differences between CPU Temp and Core Temp/Real Temp, highlighting the observations above:



It should be noted that these differences between CPU Temp and Core Temp are very near to the differences I calculated independently in the other thread; any minor variations are handily explained away by differing ambient temperatures. It should also be noted the differences are not merely "10C lower." Recalibrating the DTS sensor to 95C together with "--" calibration setting has had a more profound impact on temperature readings than just a simplistic subtraction.

Unfortunately, I am unable to calculate the level of correlation between CPU Temp and Real Temp since I am relying on two different instruments to read these values. I could not presume that both readings were exactly simultaneous, and thus any conclusions drawn from such data would be meaningless.

Nonetheless, for the reasons detailed above, I am convinced that Real Temp reflects reality much more accurately than Core Temp. I am one satisfied customer! Thanks again to unclewebb!

[PhilDoc]

I would have to agree with you ixtap. Logically, I can't see any reason that cpu temp would be much different than what you would get for core temp. It certainly shouldn't be the ~ 10C difference that I've gotten using the Core Temp prog. I'm happy with what Real Temp reads and will be using it from now on. Thanks unclewebb, great job

[TMonte]

This is great...I'll add to the list of those requesting minimize to tray functionality.

Thank you!

[jaredpace]

ixtap: on your first coretemp graph between 440 and 540, what caused the slight curve in the graph representing a lower temperature during your load tests? It seems like someone opened the door and let in a cold draft that your case/system sucked right up, or that it was a fluctuation in voltage.

[Slay0r]

It's always been my theory that the E4300 has a TjMax=85C and that you would need a ++ calibration adjustment to bring the idle temperatures in line with reality and helps confirm my theory that at this speed and voltage they should run at about 4C or 5C above ambient. Your data looks believable. I need to find a cheap L2 to put under the IR gun to help confirm.

Yeah that's what i have been thinking, that is why i never really cared if i saw temps at 85/90°C heh . Unfortunately about every temp reading program (hell, even my motherboards own one) seems to flag it with a tj100. Realtemp calibrated seems to give a very likely result at load temps, whereas idle still seems a tad too cool, and as the wattage output (overclock and voltage) increases it looks like it gets more evident, but as i understand it's due to the dts itself .

The above is a conjecture anyhow, no way for me to prove or disprove it so i'll wait along for your test.

Thanks for the proggy unclewebb

[Corsa]

Ohh my.... many people are goin to add more volts now ..& the minimize to tray functionality would be icing on this cake
Well done unclewebb

[necron66]

Nice work mate.

So core temp has been reading 5C to much on the Q6600?

And yeah like some others have said, temps to show in system tray would be spot on.

[rge]

There is a way to settle the issue. Anyone have a mobile dual core cpu (Tjmax documented in intel specs) and an accurate IR gun? IR the back of that cpu, and compare to core temp reading at idle with no heatsink and see if any gradient exists between die temp and IR to cpu. If gradient is near 0, then realtemp is accurate.

The theory that Tcase/cpu temp may in fact be near die temp most of the time is interesting. Intel talks about a die gradient of up to 10C between cores primarily since one core may be inactive and the 2 cores are separated from each other much further than either from IHS, and also transient hot spots. But when both cores are feeding the same temp to tjunction or tcase, especially across ?couple hundred microns of highly conductive material at idle... well they just don't say, but certainly would seem to be low.

So anyone with a mobile duo core cpu (with known Tjmax) and IR gun want to confirm one way or another?

[bladec]

Hello:

I need a version of this program for vista64 or known how i can do to work in vista.

Thanks in advance.

Greattings

[rosco]

A brief aside: Some argue that the CPU Temps and DTS values are unavoidably different because the locations of their respective sensors (ie geometric center of die versus, well, versus anything else). This has always bothered me. If the material in question conducts heat well, as a CPU die surely must, it seems obvious that it would only be a matter of microseconds until that heat was transmitted to any other measuring point. Any observed differences in these temperatures strike me more as a way to determine the relative error in CPU Temps than as some condemnation of the sensor itself.

Absolutely not. A silicon die is not a very good thermal conductor (bulk silicon ~148 W/m.K and far worse for others materials used to create transistors structures, etc.). Thermal gradient within the die can achieve 20~40+ °C between the hottest and the coldest point, depending of the load. That's why a die presents what we call 'hot spots' because there are hotter than the rest of the die (more power density such ALU or fetch units for instance) and they are the major problem to cool the chip and ensure reliability. Die temperature map is and won't be uniform at all, ever !

On a dualcore, it's clearly seen. Load only one core with TAT for instance and look values reported by the DTS sensors, there will be ~4-7 °C between them (depending of cooling system) and however the sensors are only ~5 mm far from each other. According to you, the sensors should be at the same temperature, which is nonphysical. There's always a thermal gradient when heat has to travel through materials. The worse the material is, the higher the gradient will be at a given distance, Fourier law. CPU TEMP provided by the third thermal diode don't have to be the same than DTS TEMP, it's located between the 2 DTS sensors but in a 'colder' place (results are affected by mobo circuitry unfortunately, but a external system could normally be used to calibrate and use it like on AMD CPU).

Nothing is linear here, you can't guess the difference between sensors, that's why temperature difference between them is evolving for each situation (idle/load but what load...). Intel already showed the large deviation between Tj provided by DTS and Tcpu provided by the third sensor depending of die loading map (power map more precisely), there's no relation at all between them ! We can see Tcpu like a 'mean' temperature if we want, that's all.

[g0dM@n]

i installed and on water at 1.3625v on my e8400 I'm idling at 38-39C and load at 49-51C.

I like the Minimum temp and Maximum temp display.

[rosco]

There is a way to settle the issue. Anyone have a mobile dual core cpu (Tjmax documented in intel specs) and an accurate IR gun? IR the back of that cpu, and compare to core temp reading at idle with no heatsink and see if any gradient exists between die temp and IR to cpu. If gradient is near 0, then realtemp is accurate.

The theory that Tcase/cpu temp may in fact be near die temp most of the time is interesting. Intel talks about a die gradient of up to 10C between cores primarily since one core may be inactive and the 2 cores are separated from each other much further than either from IHS, and also transient hot spots. But when both cores are feeding the same temp to tjunction or tcase, especially across ?couple hundred microns of highly conductive material at idle... well they just don't say, but certainly would seem to be low.

So anyone with a mobile duo core cpu (with known Tjmax) and IR gun want to confirm one way or another?

There's always a gradient when dissipated heat travels from one point to another, it's physical. And please stop using IR guns, there are the worse instruments often used by people like a toy. It requires some cautions (like IR cams) because of the emittance problem of the surface we want to scan (why the yellow adhesive tape near the IHS on the first page? What's for? TC bonded to IHS?). Without preparation, pointing an IR gun on a +/- shiny surface like IHS or die is not good, the error will be higher because of radiation dispersion (gun don't measure temp but radiation) and gun will tell you a lower temperature than reality. If not adjustable, the gun is generally tuned for a 0.95 emittance, it's useful for black/matte surfaces (near black body) but not reflective/shiny/clear ones. Using a thermocouple bead is more useful and confident. You can also paint the IHS with black paint to get a emittance near maximum during measurements, that is 1. Moreover, you can't conclude anything with measuring Tcase (at the IHS center) for Tj or Tcpu because of the thermal gradients (a silicon die is ~0.04 °C/W through its thickness and there's TIM1 and IHS in the heat travel to take in account too for desktop CPU).

What matters is the real Tjmax, nothing else. Unfortunately Intel keeps its secrets like many others and it's a bit strange to not reveal such 'non critical' information I think... It's near 100 °C each time, that all. A bad guess and you have your cores under ambiant temp, so you know there's a flaw in that value. The absolute value will be false, but not the relative one. That don't make any great difference if the CPU is stable and not throttling, even with an o/c. What is observed here is related to the fact that nothing is linear again. You are pointing an IHS when its surface could have a 10 °C temperature difference under load, IHS temp is not uniform at all. That's a lot of flaws and errors at the final in the process...

[zillaoc]

Reads sub-zero? Would be wicked sweet if it did

it does


Edit: Thanks for making a kick ass program m8

Finally something that reads my temps correctly. For some reason with my latest bios in this old ars board
core temp read totally random numbers lol...

[theonlybabyface]

Too bad it doesn't work on Vista x64.

[audiofreak]

@unclewebb:
Program is great and all, but could you switch to some driver which has x64 support? Many people use 64-bit OS here.

[emoners]

thanks again unclewebb!

[rge]

why the yellow adhesive tape near the IHS on the first page?

because

pointing an IR gun on a +/- shiny surface like IHS or die is not good, the error will be higher because of radiation dispersion