CompuTronix - E4300 Temp Discrepancy

[SuperFly03]

Ambient = 25c
Chipset = RD600
C2D = E4300
CPU Cooler= Thermalright Ultra 120 Extreme w/ 125CFM Fan
Frequency = 1.8GHz (just testing stock temps)
Load = TAT @ 100% 10 minutes = 39C Tcase
Motherboard = DFI ICFX3200
Vcore = 1.3

I run Intel TAT @ 100% load and I get a reading of 52c for both Tcases and Tjunction according to Intel TAT, but SpeedFan reads it as 39C Tcase (each core) and Tjunction as 52c.

I should mention that my proc has had the IHS removed and is using the X23 TP.

Personally I believe SpeedFan, but I wanted your opinion before I start setting WR's with this chip.

I need some reassurance from the Core 2 Temp master.

(ignore sig, all stuff is stock settings at this point)

 

[orangegator]

From what I understand, TAT reports temperatures 15C higher than actual. I assume since your post is titled "Computronix" you have read the Core 2 Dou temperature guide. You realize you could just send him a PM. But with my experience, TAT reports core temperature 13-15C higher that Speedfan and Everest.

 

[SuperFly03]

From what I understand, TAT reports temperatures 15C higher than actual. I assume since your post is titled "Computronix" you have read the Core 2 Dou temperature guide. You realize you could just send him a PM. But with my experience, TAT reports core temperature 13-15C higher that Speedfan and Everest.

Yeah I read it, the problem is that my E6600 reported temps correctly on the same board but the E4300 is throwing me off. I ruined the E6600 taking off the IHS, but the E4300 survived. I want some confirmation about temps so I judge the thermal limits of my OC properly. I still need to add a side inlet fan... dirty NB is freakin hot.

Also didn't know if CompuTronix liked pm's. Some people as active as him don't like PM's much, so I erred on the side of caution.

 

[kwalker]

have you tried ver .95?
http://www.thecoolest.zerobrains.com/CoreTemp/CoreTemp.zip

 

[SuperFly03]

This is exactly what I see.

Idle



Load



No moching my desktop.... I just got it up and running.

Darn dark side is giving me crap... I need to bake some cookies and maybe my temps will fix themselves. :lol:

 

[orangegator]

I think core temp 0.95 has the same problem as TAT. They report the temperature of the E4300 15C too high. Core temp 0.94 seems to report correctly.

 

[kwalker]

I see what you see 8)
The temps seem in line.
Core temp .95 has been adjusted to better suit the E4300.
Try the link I provided above.
the more I read the data sheets the more confusing it becomes but I almost have the new PECI understood for the newer processors without the thermal diode.(E4300 still has the diode on die I believe)
The problem with core temp is not the concept but the consistency.
remember the DTS gets its reference from the thermal diode and has an assigned value of -10 Deg C at Tcontrol which is 66C read from the diode The DTS reading at TCC activation (max temp) is zero and dependent on the package set temperature.
This makes the temps reported by core temp inconsistent from processor to processor being the T-Junction is set per package at manufacturing.
The PECI interface uses the DTS and must be enabled in bios under the CPU section to fully make use of this function but the bios revision has to support this also.
You can still use the motherboard supporting software like ASUS Probe to monitor temperature using 60C as max temp max load.
That should be 66 to 68C core temp aprox.

 

[kwalker]

I think core temp 0.95 has the same problem as TAT. They report the temperature of the E4300 15C too high. Core temp 0.94 seems to report correctly.

I'm not too thrilled about TAT being that it was designed for the mobile platform.
It burns the CPU beyond the thermal requirements in the desktop environment.

 

[SuperFly03]

Sure.... I kind of understood that. So are CoreTemp .94 and SpeedFan singing the correct tune in your opinion?

I understand your objection to Intel TAT, I use it as a worst case scenario to gauge my OC.

 

[orangegator]

I think core temp 0.95 has the same problem as TAT. They report the temperature of the E4300 15C too high. Core temp 0.94 seems to report correctly.

I'm not too thrilled about TAT being that it was designed for the mobile platform.
It burns the CPU beyond the thermal requirements in the desktop environment.

Are you saying that TAT puts more stress on desktop cpus than on mobile? I would think that it wouldn't matter on the cpu and that desktop cpus could handle the stress better because of better cooling.

Anyway, I see TAT as an absolute worst case scenario for stress on a cpu. It it can handle that, it should handle anything.

As for core temp 0.94, it reads the reference tjunction as 85C, which I understand is what it should be. Core temp 0.95 reads it as 100C, which is where the discrepancy comes from.

 

[SuperFly03]

As for core temp 0.94, it reads the reference tjunction as 85C, which I understand is what it should be. Core temp 0.95 reads it as 100C, which is where the discrepancy comes from.

Way to break a good thing....

Ok, if well agree that CoreTemp .94 is singin the right tune I will get to studying. Expect OC results on either Monday or Thursday (damn finals lol).

 

[kwalker]

Actually the readings are on the low side which I guess I didnt look close enough the first time.
Try the newer version of core temp and post a screen shot

 

[SuperFly03]

Actually the readings are on the low side which I guess I didnt look close enough the first time.
Try the newer version of core temp and post a screen shot

Alright I will, I am in the middle of OC'ing while I wait for my dinner to cook.

bear in mind that this E4300 is IHSless and the heatsink is lapped and the best made. You're right they are on the low side, but I am inclined to believe it... that 120mm is pushing 125CFM of ducted cool air.

I will post a screen shot after dinner.

Thanks for the help. crazy ass snoopy.

 

[SuperFly03]

Idle



Note:

vcore has gone up from 1.3v to 1.35
Core frequency has gone up to 3003 MHz
FSB is up to 333
Few voltage tweaks in the BIOS (4-5 increases of the smallest increment possible)

Dual Prime Stable through 1 round of small FFT's (just getting a feel for capability).

Load: 42c Tcase, 58c Tjunction

CoreTemp .95:

I don't like it... its off by 15c exactly as someone noted earlier, grr.

 

[kwalker]

I think you’re more consistent with .95 loaded.
All reporting software is within a few deg C of each other.
I can tell you’re an accountant.
The calculator is always handy :lol:

 

[SuperFly03]

SmartGuardian reads it from the BIOS which is known to be Tjunction which is why I believe that .94 is more correct than .95, but .94 may be off by 1c or so. Idk... it'd just be nice to not have this problem.


Yes, that calculators comes in handy quite often, always have to find the best value.

 

[dr_kuli]

http://www.alcpu.com/forums/viewforum.php?f=35

In case you're too busy to read all:

According to one of the more interesting threads I've read in Intel's open Dev forums, a very unpleasant picture arrises. Desktop/Server CPUs DON'T have a constant Tjunction temperature, unlike many Mobile CPUs which contains the Tjunction information in bit 30 of MSR 0xEE. This value is either 85/100. Possibly by a mistake, this value seems to exist in the desktop processors as well, but this does not make it valid at all.

TAT was developed for mobile CPU usage. The whole Tjunction minus DTSDelta was good for these CPUs(with small deviations of several degrees above or below), but it is basically useless for desktop CPUs. In desktop CPUs the calculation is PROCHOT# minus DTSDelta, and obviously we cannot read PROCHOT# temp.

And THESE CPUs were without an IHS...