Record your Q6600 surface to core temperature difference

[orion456]

Is a Q6600 G0 just a better heat spreader?

I observe that my Q6600 B3 has a temperature difference between surface and core of 20c, but my G0 is just 10c. I wonder if Intel has found a better way to conduct heat to the surface and/or made the surface flatter and that is the main reason for the higher surface heat specification? I wonder if lapping a B3 will nearly produce a G0? What temperature improvement happened after your lapping?

Here is one rather pointed comment I found on the net: "Indeed, I had to lap my rather hot but golden E6700 to acheive 3.8ghz @ 1.46v load only. Now with the G0 Quad 6600, I don't need to bother lapping the suppose expensive CPU and retain the 3 yr warranty. I do also notice my Q6600 G0 IHS is preety flat when I put it on a piece of glass and used the razor blade test. The IHS doesn't look like the common poorly concave/convex IHS."


Record your full load difference in temperature on your Q6600 using CoreTemp 0.94 and SpeedFan 4.32 for consistency. PECI must be enabled in the bios.

eg:
Stepping: G0 [Q6600 only]
Load diff: 47c - 37c = 10c (core minus surface temp in celcius = difference) [CoreTemp 0.94 & SpeedFan 4.32]
Heatsink: water Swiftech h20-220
Lapped: no

Stepping: B3
Load diff: 58c - 38c = 20c
Heatsink: water Swiftech h20-200
Lapped: yes, 2c drop in CoreTemp, 3c surface temp (SpeedFan)

 

[bing]

These are the temp locations, also naming standards as Intel called it. Hopefully it will serves well for the discussions !


Courtesy of Intel


Orion,

Understand this thread is about finding the (Jc) right ?

Asuming you measure the core temp (Tj) using CoreTemp or SpeedFan, where to get that surface temps (Tc) ?

I also concern bout the IHS/surface (Tc) reading accuracy when its read by software.

Fyi, to have an accurate reading of Tc is to put a temp probe in the grooved IHS.

 

[orion456]

These are the temp locations, also naming standards as Intel called it. Hopefully it will serves well for the discussions !

.

Yes, good idea. That diagram makes it pretty clear how Intel defines its temperature locations.

Unfortunately as well all know, most temperatures reported are not very accurate and few of us have probes in the appropriate location. But I propose simply using CoreTemp or TAT to get the core temperature and use SpeedFan for the surface temperature. Alternatively if your mobo has some software to read its own settings then that would be preferred. As long as people state where their temperatures come from, we should get some good data (I have modified the suggested listings accordingly). I have a probe near the CPU surface and it gives readings within a few degrees of the SpeedFan number; that gives me some confidence we won't be that far off.

It would be interesting to see if B3 differences are consistently higher than G0. Even if we aren't very accurate the averages will work in our favor if we get enough readings and elliminate the spurious ones.

 

[bing]

Ok, actually I have a big interest with this thread's outcome as well !

My suggestions :

- Make it "uniform" on all CoreTemp version, either 0.94 or 0.95 pick your choice. This makes they're consistent, cause it seems Art (CoreTemp author) haven't update the program and also different interpreting the TJunction values especially for these new G0 series. Fyi, these supposed to be "undocumented" by Intel, so at least we have the same proggy, it will be consistent results as well. and hopefully the outcome will be meaningful.

- Again, same SpeedFan version, I assume using the latest one 4.32, not beta or older one.

- PECI must be enabled, or these "case value" from SpeedFan or BIOS will be useless.

Last, of course before data starts to pour in, update your 1st post about these requirement asap.

 

[orion456]

My suggestions : -Make it "uniform" on all CoreTemp version,
- Again, same SpeedFan version, I assume using the latest one 4.32, not beta or older one. - PECI must be enabled

Done. Can you explain what PECI is?

 

[bing]

Why I never trust software based cpu temp reading ?

Done. Can you explain what PECI is?

he..he. I was afraid you ask that !

Ok, "officially" I never know or have read Intel's PECI working mechanism document since they're classified, only their partners like mobo makers have it. If you google around bout PECI, you will not find anything usefull enough, or worst even Intel's search at their site wont give you anything meaningful.

Now, please don't take too seriously since I only "heard" from a guy who read that document.

PECI -> It is a communication mechanism or protocol made by Intel for all mobo components like CPU, NB, SB, IO chip and etc. If you notice, mobo bios can read cpu temp right ? Cause they're using PECI mechanism to communicate with the CPU and remember, it is not the same mechanism used by software cpu temp reading like Coretemp.

So far these PECI cpu temp reading is NOT about reading it's absolute value because they never accurate since the tdiode is not calibrated.

Instead they read the difference or delta since they are very consistent that is really crucial for cooling parameter to work properly like adjusting the fan speed or other extreme like instructing the PWM chip to throttle up or down and etc.
Hence, they called it PECI = Platform Environment Control Interface.

Now, the final blow, those CPU temp reported by the BIOS, actually it was through complex mechanism delivered by PECI protocol, and then they sort of "estimate" it from the power delivered to the CPU through the PWM they're controlling and some mixed analog values from the on die cpu sensors.

So the CPU temp is NOT solely detected from single source, but sort of calculated and gathered from different readings with multiple sources.

Again, don't trust everything I say here since I never read that document my self with my eyes, its just sort of blindly trusting what my pal said there. And yes, I've begged him already, but still no candy !

Now about the "absolute" IHS/suface temp reading, simple, even Intel's hardcore cpu engineer themself never use software based CPU temp reading.

Proof ?

Here -> Intel® Core™2 Extreme Quad-Core Processor and Intel® Core™2 Quad Processor Thermal and Mechanical Design Guidelines Published at July 2007 (PDF file)

For those who lazy to read and just love pictures , just look starting on page 81 and so on .. ! Me ? I do read em all !

Snippets, Intel's way to measure cpu temp ! See that temp probe burried in the IHS ?
Same HERE at our own OcF front page.


Courtesy of Intel Corp.


Now you know why I never trust "software based" cpu temp reading !!

Hope this usefull !
-

 

[Hipcrostino]

Software shouldn't be trusted as a rule, because as you said, the extra resistances of the other equipment on the board change thre reading. Wouldn't be such a problem if the probs were so sensitive. It dose however let us see the temperature Delta. Mine is about 20c atm.

I have heard of PECI aswell. It was in one of thier slides when presenting info on the core 2's a while back. Its just mentioned, nothing else. All i ever found it that it was a hardware based monitoing system for temps and voltages, but nothing more. Thanks for the info Bing. Helpful as always

The other thing to note is that speedfan will not measure the tcase. Infact its probably going the measure the temp from beneith in the socket, but i could be wrong. The only way to get the true tcase is by doing somthing like the pic above. Personally i don't think it matters anyway. We are only trying to prove if the IHS on a G0 is better than a B3. So with the same mobo, and same heatsink/TIM dose the core temp change between them? If one IHS is worse, than its core will be hotter.

 

[Brolloks]

Bing, do you know of anyone except Intel who has embedded a temp probe into their IHS to measure accurate temps, I have a E6400 that I would be prepared to "sacrifice" for the sake of accuracy.

 

[Maverick0984]

For sake of argument, the E6700 I had was flat as can be, no lapping was needed.

 

[orion456]

Lapped my Q6600 B3 in the name of science; 3-4c maximum change though it will take the as5 a while before reaching the best result. The core/surface difference remained unchanged by lapping. My chip clearly was higher around the edges and it took a while before I could get down to the surface on the inside using 290/600/1200 grits. The final result is pretty shiny and as you can see nice and flat.

 

[bing]

Bing, do you know of anyone except Intel who has embedded a temp probe into their IHS to measure accurate temps, I have a E6400 that I would be prepared to "sacrifice" for the sake of accuracy.

LOL busted, you did not read my post thoroughly, I did post a link there !

Its our own OcF's red stars administrator -> Joe Citarella (nick : JoeC) with his OcF front page column, again -> HERE !
He rarely posts here in OcF and handles the front page !

Just scroll down at the last pic for the poor grooved D805 ihs !

Imo, you should PM or mail him for tips or asking him to share his experiences when making that groove ! Or the best is to invite him and start a discussion thread for it.

Just read that Intel's document above on how they did it, by looking at those series of pics, not an easy task IMO.

Well, if you're really serious to do it and share with us on the process and the results here, for sure that thread of yours definately worth stickied, or at least I swear I will vote for it !

Please PM me if you really want to do it, I just love to participate in planning those series of testings and it's preparations if you don't mind !

Sorry for the OT post Orion !

Now folks, post your result please !!!

 

[Brolloks]

LOL busted, you did not read my post thoroughly, I did post a link there !

Its our own OcF's red stars administrator -> Joe Citarella (nick : JoeC) with his OcF front page column, again -> HERE !
He rarely posts here in OcF and handles the front page !

Just scroll down at the last pic for the poor grooved D805 ihs !

Imo, you should PM or mail him for tips or asking him to share his experiences when making that groove ! Or the best is to invite him and start a discussion thread for it.

Just read that Intel's document above on how they did it, by looking at those series of pics, not an easy task IMO.

Well, if you're really serious to do it and share with us on the process and the results here, for sure that thread of yours definately worth stickied, or at least I swear I will vote for it !

Please PM me if you really want to do it, I just love to participate in planning those series of testings and it's preparations if you don't mind !

Sorry for the OT post Orion !

Now folks, post your result please !!!

Yep, I hate reading stuff, so you got me...get too excited sometimes...

I'm tied up with work stuff but I'd definately would like to tackle the CPU temp probe mod, so once I get some free time, I'll PM you for some advice and tips before I tackle the delicate surgery process...

 

[bing]

Yep, I hate reading stuff, so you got me...get too excited sometimes...

I'm tied up with work stuff but I'd definately would like to tackle the CPU temp probe mod, so once I get some free time, I'll PM you for some advice and tips before I tackle the delicate surgery process...

Great !!! Can't wait !!

Start a thread where we all can see the progress from the beginning till finished, that will be awesome !


Uh..oh ... another bump up for this thread !