Q9650 Core Temps

[leonard619]

I just purchased a Q9650 (upgraded from a q6600). Here's the problem. I have one core that consistently runs 8-12 degrees higher than the others (at idle and load). For example, at idle my temps are 33-35-35-42 and at load 54-51-53-62.

Now here's the funny part; the above temps are from Real Temp 2.70. When I read the temperatures from Speedfan 4.35, they read like this 38-39-47-38 at idle and 59-57-67-59.

Again, I have one core consitently higher, however it is core#2 in Speedfan and core#3 in Real Temp.

PLEASE!!! PLEASE!!!PLEASE!!! Any thoughts on this would be greatly appreciated. I'm thinking I might have a defective thermal sensor in the cpu.

I want to know what you guys think. If I have a defective chip, I want to return it before the time runs out.

Here's some sytem info. if that may help

Evga 780i FTW
Q9650 4.1Ghz @1.36v
2-8800 Ultras SLI'ed and OC'd 730/1780/1208
Patriot 2x2Gb PC-8500(1066Mhz) Linked and Synced
Swiftech H20 system w/ GTZ CPU block
Swiftech Steath blocks for the ultras
and 320 triple 120mm radiator
Thermaltake 1200w PSU
Antec 900 case

 

[disports]

Real Temp 2.75 is more accurate or Core Temp 0.99.3. Don't use speedfan.

 

[leonard619]

I have core temp 0.99.3. I get the same skewed results.

Any other comments!!!!!!!

 

[BTRY B 529th FA BN]

I'll be back as soon as I get mine up and running... in another year or two

 

[Gillbot]

Originally posted by: leonard619
I have core temp 0.99.3. I get the same skewed results.

Any other comments!!!!!!!

Check your cooler mount. I've seen this happen with uneven pressure.

 

[Drsignguy]

As I too have seen pressure related temps go a bit crazy. But to let you know that my Q9450 has temps at full load 59-52-59-59 respectively. My Tuniq is lapped but my processor is not. So, if your HS is mounted correctly, don't be to alarmed.

 

[Idontcare]

Originally posted by: leonard619
Again, I have one core consitently higher, however it is core#2 in Speedfan and core#3 in Real Temp.

That part is normal, the core numbering sequence seems to be inconsistent for some of these programs. Really not a concern.

Originally posted by: leonard619
PLEASE!!! PLEASE!!!PLEASE!!! Any thoughts on this would be greatly appreciated. I'm thinking I might have a defective thermal sensor in the cpu.

I want to know what you guys think. If I have a defective chip, I want to return it before the time runs out.

I have a QX6700, one of the old original B3 stepping kenstfields, and it's temp readings are like yours. One core is always blazing hotter (figuratively) than the others.

At idle I get 38C-43C-36C-38C

At small FFT I get 66C-66C-62C-62C

Now this is for stock, 2.66GHz with stock Vcc and a Tuniq120. And I have lapped the IHS and the HSF to no improvement in the temperature delta between the hot core and the rest. (although they all improved and got cooler in doing the lapping).

I've worked with this chip six ways to sunday, repeated lappings and HSF reseats, multiple different TIMs, even different HSF's (I have a lot of Tuniqs lying around) and it just doesn't matter for my chip.

Whatever makes that one core reportedly hotter than the rest is happening under the IHS.

It is plausible that the one core really is hotter (process variation is a fact of life, even within a die). It is also plausible that the DTS non-linearity and lack of calibration for temperatures below TJmax is really what is throwing off the software reports on the core's temp.

In all cases it does not mean you have a defective themal sensor in the CPU. It just means you prolly aren't going to like seeing what it has to report for the remainder of your ownership of the CPU.

Is it really a concern for you that your one core is reportedly hotter than the others? I doesn't bother me, the performance is the same, nothing is "defective" in these situations.

edit: updated my fully loaded temps after things stabilized a bit more.

 

[nyker96]

you have to manually adjust the tjmax in realtemp to get a correct reading. realtemp and speedtemp sometimes differ by 5C or 10C because they use different tjmax number. from your description, they might have different core # as well. just use real temp and manually enter a TJMax. forgot the correct tjmax for Q9650 though, find out and input it, after which you will get the correct number.

 

[leonard619]

Originally posted by: Gillbot

Originally posted by: leonard619
I have core temp 0.99.3. I get the same skewed results.

Any other comments!!!!!!!

Check your cooler mount. I've seen this happen with uneven pressure.

I remounted my GTZ block at least half a dozen times with the same results.

I'm really an obsessive compulsive type person, and although the chip is performing just fine (I'm OC'ed at 4.1Ghz@1.36v prime 95 stable for 8 hours), my feeling is if I'm paying almost $600 for a processor, I want it to be perfect. So more than likely I'm going to return it for another one.

Also, I was really expecting more out of this chip with water cooling. I'm kinda dissappointed with only 4.1 ghz. I was hoping to get 4.4-4.5Ghz out of this chip on water. I've seen oher people post threads about getting 4.0-4.1Ghz on air. Do you think 4.4-4.5Ghz is possible, or am I just insane.

 

[Concillian]

I think you're insane.

4.1 is respectable with a quad. Just realize that most of the people who make claims like this on the internet don't have accompanying 100 / 100 Linpack pass result screenshots or even 24 hour Prime screenshots.

You can't believe everything you read on the internet, especially in such an e-peen related subject.

It's possible someone has a golden boy, but it's more probable that their "stable" definition is different from other people's definition.

 

[JAG87]

Originally posted by: leonard619

Originally posted by: Gillbot

Originally posted by: leonard619
I have core temp 0.99.3. I get the same skewed results.

Any other comments!!!!!!!

Check your cooler mount. I've seen this happen with uneven pressure.

I remounted my GTZ block at least half a dozen times with the same results.

I'm really an obsessive compulsive type person, and although the chip is performing just fine (I'm OC'ed at 4.1Ghz@1.36v prime 95 stable for 8 hours), my feeling is if I'm paying almost $600 for a processor, I want it to be perfect. So more than likely I'm going to return it for another one.

Also, I was really expecting more out of this chip with water cooling. I'm kinda dissappointed with only 4.1 ghz. I was hoping to get 4.4-4.5Ghz out of this chip on water. I've seen oher people post threads about getting 4.0-4.1Ghz on air. Do you think 4.4-4.5Ghz is possible, or am I just insane.

Impossible. You will never see 500 fsb on a quad core with that board. Where you are now sounds about right, if you tweak to death you can probably get around 470-475 fsb and reach 4.2 ghz, but thats pretty much it.

 

[leonard619]

Originally posted by: JAG87

Originally posted by: leonard619

Originally posted by: Gillbot

Originally posted by: leonard619
I have core temp 0.99.3. I get the same skewed results.

Any other comments!!!!!!!

Check your cooler mount. I've seen this happen with uneven pressure.

I remounted my GTZ block at least half a dozen times with the same results.

I'm really an obsessive compulsive type person, and although the chip is performing just fine (I'm OC'ed at 4.1Ghz@1.36v prime 95 stable for 8 hours), my feeling is if I'm paying almost $600 for a processor, I want it to be perfect. So more than likely I'm going to return it for another one.

Also, I was really expecting more out of this chip with water cooling. I'm kinda dissappointed with only 4.1 ghz. I was hoping to get 4.4-4.5Ghz out of this chip on water. I've seen oher people post threads about getting 4.0-4.1Ghz on air. Do you think 4.4-4.5Ghz is possible, or am I just insane.

Impossible. You will never see 500 fsb on a quad core with that board. Where you are now sounds about right, if you tweak to death you can probably get around 470-475 fsb and reach 4.2 ghz, but thats pretty much it.

Yeah, I am insane. Thanks for making me see the light. I should be happy with 4.1Ghz.

Jag,
I know whatt your saying is true. I've read on several other forums (as well as personal experience) that nvidia boards are FSB limited to around 475.

 

[Dadofamunky]

Dang, Leonard, 4.1 GHz with a Quad? With those temps? Given the advice above on adjusting TJMax for better measurement, I'd say you're in VERY good shape! Particularly if your stress tests are passing.

I would make absolutely sure of your stability by also running OCCT and the IntelBurnTest as well. I try to do all three now. From what I hear, the Intel test is the best one for REALLY turning up any instability. YMMV, of course.

I've never used water and have no plans to, and there are some VERY knowledgeable people here who might disagree with what I'm about to say, but water doesn't necessarily outperform air in a lot of situations. I've seen reviews here and in other places coming to that conclusion (sorry, I wish I had links). But folks around here who should know swear by it. Every chip is slightly different and the one you get may just be capable of what it does now. I don't think it's inherently defective. If you DO return it or exchange it, don't tell them you OC'ed it to the eyeballs!

I thought I'd be able to get to 4.3-4.4 easy with my E8600, but I'm gonna have to live with a stable 4.1 GHz (at least if successful Intel Linpack 75 tests are anything to go by).