10C temperature difference between cores?

[dcaxax]

Hi folks,

I have an Intel Q9300 2.5GHz over-clocked to 3GHz on stock voltage, liquid cooling.

My CPU temps seem ok - 23C idle up to about 40C under full load, but my core temps are very strange: Core 1 is around 38-42C, Core 2 is 40-43, but core 3 and core 4 are around 50-55C (!). Especially Core 4 easily goes to 54 at idle and up to 73C under load.

Is this normal? I know core temperatures vary, but 10C+ seems excessive...

Could it be I messed up when I applied thermal paste to the CPU? I'm worried I may have been a bit too generous with my AS5 which I've heard can actually have a detrimental effect on cooling...

Any advise would be appreciated!

 

[NumericalMethods]

That does sound excessive. Definitely time for a remount if you haven't tried that yet.

In the world of TIM application less is more - so long as you use enough to cover the IHS when it spreads. It's a case of trial and error to find a good method for mounting. Some people prefer spreading the TIM evenly with a credit card. I myself just apply a rice size blob and let the mounting pressure do it's work.

 

[dcaxax]

Thanks for that! I was afraid this was the case


Do you think it's urgently dangerous at this level?

I've ordered some new thermal paste (IC diamond 7 carat) to replace my Arctic Silver 5 and it should be here day after tomorrow.

Am I running a scary risk waiting 48 hours? The CPU will be mostly idle till then and set the clock back to stock speed as well (I mostly use it to surf the web these days). I'd rather not go through the trouble of re-mounting the cooler twice in a week, if it's not gonna burn out my CPU...

 

[Insomniator]

huh interesting, my 9550 reads 33/33/41/37 idle but 50 across the board under load. Your cores pegging way higher under load probably suggests a bad mount.

I would think that at stock speeds you are fine... I bet its still better than the stock cooler.

 

[NumericalMethods]

I personally would be fine with it running at that for a couple of days. Low 70's is about the maximum temperature I take my 45nm chips to overclocking for 24/7 use. I have taken them higher for short periods with no noticeable adverse affects. You will be even safer if the chip is not under load producing maximum heat.

 

[dcaxax]

Thanks for all your advice. I'll wait until the new past gets here and I may upgrade the radiator fans if remounting the cooler doesn't do the trick...

 

[Anubis]

That does sound excessive. Definitely time for a remount if you haven't tried that yet.

In the world of TIM application less is more - so long as you use enough to cover the IHS when it spreads. It's a case of trial and error to find a good method for mounting. Some people prefer spreading the TIM evenly with a credit card. I myself just apply a rice size blob and let the mounting pressure do it's work.

Agree especially as you said you are running a WC setup

this has lots of good info on TMI application

http://benchmarkreviews.com/index.php?option=com_content&task=view&id=170&Itemid=38

 

[dcaxax]

Will chek it out - thanks!

 

[tweakboy]

That is perfectly normal as the CPU is concave and not even surface.

This is why people LAP their CPU's Lapping is where you take a grit sand paper and keep on sanding until its even.

If you dont lap your CPU it's possible you will have a difference 10c . There shouldn't be that much difference. Just make sure theres enough contact with HS and CPU and u used good amount of arctic silver 5 , or whatever you choice is. I did not lap my CPU and I the first two cores are 8c hotter. Also lapping is dangerous if your not tech savvy and also voids your warranty.. good luck, just make sure its tight contact on CPU,, good luck

 

[Tsavo]

My E3200 has 10c difference idle, but they are the same under load. No amount of fiddling with the crap Intel heat sink changed it.

I'm not concerned about it because the temp of both cores is the same under load.

 

[Edrick]

I have the same exact chip and I have the same exact problem.

I overclock to 3.0 also.

I found that running it on stock voltages would become unstable after about 6 hours and ran hot.

So I increased the voltage slightly and now it runs very well and slightly coolor. But I still have a 10C difference between 2 of my 4 cores. I do not worry about it since I have a good cooler and my overall CPU temp never goes above 40C under load.

 

[Kenmitch]

You could always just do the loosen up your water block mounts enough till you can slightly twist the block around a little and then tighten it up again. But like others have said the temps are safe for now anyways.

Have you watched task manager while monitoring the temps? Windows does do things in the background and may be just hitting that core more than others for some reason.

 

[dcaxax]

Well, I took everything apart, put it back together again - replacing the AS5 with IC Diamond 7 Carat paste, which is supposed to be better and ...nothing.
With the CPU overclocked again, Well temperature temp (from the MB sensor) dropped to 21C idle and 44C under Prime95 (in-place fft tests).

The core temps are about the same under idle - 36,38,49,50C and under load I get 60,59,65,68.

I'm pretty sure I set things up right this time, so I'm inclined to think just I got a weird CPU and leave it at that, unless if one of you has some different ideas to offer

 

[NumericalMethods]

If you got end temps of 60, 59, 65, 68 with a remount then don't worry about it. I definitely think that is within the realms of normal operation.

There is a second point of heat transfer that you can't alter - the one between the CPU die and the IHS. They are connected by solder applied at the time of manufacture. If the solder doesn't spread evenly then the chip will dissipate heat unevenly. I would go with it the way it is.

 

[dcaxax]

Thanks for putting my mind at ease!

 

[soccerballtux]

That is perfectly normal as the CPU is concave and not even surface.

This is why people LAP their CPU's Lapping is where you take a grit sand paper and keep on sanding until its even.

If you dont lap your CPU it's possible you will have a difference 10c . There shouldn't be that much difference. Just make sure theres enough contact with HS and CPU and u used good amount of arctic silver 5 , or whatever you choice is. I did not lap my CPU and I the first two cores are 8c hotter. Also lapping is dangerous if your not tech savvy and also voids your warranty.. good luck, just make sure its tight contact on CPU,, good luck

wrong
lapping doesn't make that big of a diff

 

[dcaxax]

I definitely wouldn't risk lapping since I draw the line at sandpaparing chips

I'm happy now, since if the core temps stay under 70C even after an hour of Prime95, it means the cooling is reliable.

As for re-mounting the cooler, I honestly doubt it would make a difference - After all these years, I can't still be so incompetent as to do it wrong twice!