C2Q - Tcase and Tjunction calibration procedural questions

[cwell1337]

Alright. I have read graysky's guide to overclocking C2D/C2Q and have also read CompuTronix guide on temperatures for C2D/C2Q as well.

Now based on my understanding after reading those two guides, Tcase should always be lower than Tjunction, correct?

That is obviously not the scenario here. Temps here are being reported via Everest. I checked them against Core Temp 0.95.4, and PC Probe II as well and they give identical readings.

I have made sure PECI is Enabled in my BIOS settings. I even tried rebooting with it Disabled and still no change in the inversion I am experiencing. I want to overclock this Q6700. I'm running it on a P5K Deluxe with a Zalman 9700 heatsink and AS5. I've reseated the heatsink twice now with no change in temperatures idle or load.

It's going to be hard to overclock though when I don't know if my Tcase is being reported correctly, or if the problem is Tjunction. If max thermal spec for me is ~71C how will I know when my Tcase is approaching that level?

Has anyone else experienced this problem? Does anyone have any advice on how to correct it? I'm running Vista x64 and I have the most up to date BIOS out now (0603).

Thanks in advance for the help guys,

C

 

[CompuTronix]

CWell1337, thanks for reading my Guide. You're temperature problem is very common. You may have noted that I'm running the same motherboard. I'm also running SpeedFan Beta 4.34, which required no offset correction on BIOS 0404, however, when I recently flashed up to 0603, Tcase required a -15c offset correction for Temp 2. While this is certainly an easy problem to correct, I was quite diappionted in Asus with their lack of thoroughness in debugging this new BIOS release.

As we're running the same BIOS, this is most likely the same problem, however, don't base any conclusions on my personal rig, but instead, I recommend that you simply follow through with the Calibrations as shown in Section 9 of the Guide. Please note that to manually set 1.6Ghz, use 201 x 8. Additionally, after you perform Parts 1 thru 5, use the following format to complete your Results and Variables at stock settings, including Ambient and Vcore:

Results

Tcase = Idle ??c / Load ??c
Tjunction = Idle ??c / Load ??c, Hottest Core

Variables

Ambient = Room Temp ??c
Chipset = P35
C2Q = Q6700
CPU Cooler = Zalman 9700 (AS5)
Frequency = 2.66Ghz (266 x 10)
Load = Prime95 - Small FFT's - 10 Minutes
Motherboard = P5K Deluxe
Stepping = G0
Vcore = ???

I'm sure that your findings will conform closely to norms. I'll be standing by...

Comp

 

[cwell1337]

Thanks for the reply Computronix! I appreciate the information you've given me. I'm going to do the calibrations this coming weekend as I've no time during the week.

Thanks again,

C

 

[cwell1337]

Well since sometimes life steps in and gives us other things to do I have yet to run the calibrations. I see that a lot of changes have been made however so I am going to do them today and see what I come up with.

Recently I replaced my Zalman 9700 with the Thermaltake DuOrb CPU cooler. It's comparable to the 9700. I also updated the BIOS to 0812 and my Tcase dropped a lot, below my Tjunction temps where it should be. I'll post again with my findings!

 

[CompuTronix]

Good to hear it. I'll be standing by.

Comp

 

[cwell1337]

Ok, I've been going through the calibrations and have a few questions. I also want to confirm my process thus far.

Ambient temperature as measured by a digital thermometer was found to be 21C near where my side window intake fan would be had I not removed it along with the side cover. Speedfan was reporting a CPU (Tcase) temp of 21C as well. Based on the guide I assume that since I use a high end cooler (Y=1) and a high idle power Q6700 (X=6), my ambient to Tcase Delta should be = 4C (rounding up). So I set the offset in Speedfan to +4 accordingly.

I then start Prime95, make sure CPU-Z shows Speedstep increasing what it should and walk away for 10 minutes. When I come back I get these readings:

CPU (Tcase) = 49C
Core 1 = 58C
Core 2 = 54C
Core 3 = 50C
Core 4 = 52C

Yielding a Tjunction average of about 54C. +5C what Tcase is, nice right where it should be right? Now when I go to correct for the offsets at this point is where I'm a little confused. It would make sense to me to apply a +/- 2C to each core individually to get them closer to their ideal value (54C). ie -2 to Core 0, no offset to Core 1, +2 to Core 2, and +2 to Core 3. That would make their new values 56C, 54C, 52C, and 54C respectively, leaving the Tjunction average right where it should be, 54C.

However, I'm not sure if that is what I am supposed to be doing since it says in Section 9: Part 2: (C) "Enter identical correction values to each Core." Or do I even need to set an offset at all because my Tjunction average is right where it should be to begin with? If that is the case then the rest of this post is moot.

Assuming I did that right I recorded all of my data, then closed Prime95 and allowed the computer to idle for 15 minutes. Speedfan now reports:
CPU = 24C
Core 0 = 29C
Core 1 = 24C
Core 2 = 26C
Core 3 = 27C

Tjunction average now being about 27C, +3C to Tcase which is fine I'm assuming. I am instructed again to readjust individual cores so what I want to do is:
-2 again from Core 0 (total offset -4)
+3 to Core 1 (total offset +3)
+1 to Core 2 (total offset +3)
0 to Core 3 (total offset +2)

That changes the temps to:
Core 0 = 27C
Core 1 = 27C
Core 2 = 27C
Core 3 = 27C

Now if I'm doing this right I need to repeat steps (A) thu (D) and do the exact same process again? Or am I doing this right at all?

Please bear with me as it is now noon and I work nights so I have not been asleep yet. Weariness is setting in lol.

Thanks again for your help,

C

 

[CompuTronix]

CWell1337, sorry it's taken so long to reply. I think you've overshot the objective. It is not intended that all the Cores show Tjunction average. From the Guide:


Part 2: Calibration - Tjunction Load (Uses maximum cooling at stock Vcore, Frequency and Multiplier)

(B) A tolerance of +/- 2 allows for deviations between highest and lowest Cores.

(C) Enter identical correction values to each Core. For Quad Cores, if an individual Core is still out of tolerance, then enter only enough additional Offset correction to reach tolerance. Do not set individual Cores to Tjunction average.


Comp