Q6600 weird temps skyrocket

MURDEROUS G · 18 Jul 2008 at 07:58

Hi all,

I recently built a rig with a Q6600 and Gigabyte DS3P in a Coolermaster 330 case. The Q6600 proved to be unstable and I had to RMA it. Prior to sending it back, I took a screenshot of its idle temps, OC'd to 3GHz, which follows:

I just received the replacement processor and have put it in the rig, started everything up at stock, and all seems to be well, except for one thing:

Wtf? It's running .6GHz slower with less volts through it and it's ten degrees hotter than the old one? I don't get what can possibly be causing this. The new Q6600 has gone into the exact same rig as the old one. I've applied the same amount of the same thermal paste, and I'm using the same HSF (Arctic 7 Pro). The rig is in the same room with the same airflow at the same ambient temperature. I understand that VID and 'natural' idle temp can vary from processor to processor, but if I remember rightly, the previous Q6600, at stock, idled at ~33°C compared to this one's >50°C. This can't be within the expected spread of variance... right? Something has to be afoot?

I've reseated the HSF about four times and it hasn't made a difference, and just to be doubly sure, I've stood over the case pressing down hard on the four corner pins of the 7 Pro while watching the temperature in CoreTemp and it doesn't budge even a degree. I've also crosschecked the temps in SpeedFan to make sure CoreTemp wasn't misreporting them, and it isn't. Other circumstantial evidence: the metal cooling block of the 7 Pro is markedly hotter to the touch now than it was with the old Q6600 in there.

What gives? Am I being a dumbass (like I suspect) and have missed something really obvious?

Admiral Huddy · 18 Jul 2008 at 08:15

Stupid question but are you sure the fan is spinning, blowing air into the fins?

Check the speed of the fans using Everest or a simular monitor program.

Another point is to make sure the CPU fan isn't in contention with another nearby fan. These are just initial thoughts.

How much thermal paste do you apply?

You also need to make sure that the CPU and HSF are making good contact. Just apply some fresh paste as normal making sure it's smooth as a babies bum then fit the HSF as you would then remove immediately. Check the paste to see if it's been disturbed. Any remaining "Smooth" areas means the HSF isn't contacting correctly.

This is my Q6600 after 8 months or so - You can see one corner isn't making contact but is away from the die area. However, for maximum displacement the whole area needs to be making contact. I lagged this after the picture was taken.

MURDEROUS G · 18 Jul 2008 at 08:33

Hi Huddy,

Thanks for your reply.

Admiral Huddy said:
Stupid question but are you sure the fan is spinning, blowing air into the fins?

Yeah, both the cooler fan and the rear case fan are spinning and shifting the air out the back of the case.

Another point is to make sure the CPU fan isn't in contention with another nearby fan. These are just initial thoughts.

It isn't, AFAIK, and if it were I would have thought I'd have seen the same temps from the old processor? The rig's just been sitting there since I RMA'd the old Q6600 and hasn't been touched, so conditions are exactly the same for the new processor as they were for the old one.

How much thermal paste do you apply?

A friend told me an amount half the size of a grain of rice is about right. I tend to aim for roughly that much, applied such that there's just a very fine layer covering the processor.

You also need to make sure that the CPU and HSF are making good contact. Just apply some fresh paste as normal making sure it's smooth as a babies bum then fit the HSF as you would then remove immediately. Check the paste to see if it's been disturbed. Any remaining "Smooth" areas means the HSF isn't contacting correctly.

I will have another couple of goes at reapplying the thermal paste and reseating the HSF / checking the contact, and see what develops. Thanks for posting the picture, I see what you're getting at there.

Admiral Huddy · 18 Jul 2008 at 09:29

It sounds like you are doing everything correctly. It also might be worth investing in a thermal probe just to check it's not software/BIOS issues...Speaking of which, make sure the BIOS is up to date.

Pneumonic · 18 Jul 2008 at 09:45

Surface of the new CPU is different to the old... causing worse contact would be my guess...

MURDEROUS G · 18 Jul 2008 at 09:49

Pneumonic said:
Surface of the new CPU is different to the old... causing worse contact would be my guess...

That might be a contributory factor, but a 17°C jump in temps? I don't see how the surface of the new CPU could cause such a huge change unless it was, like, spherical or something.

SS-89 · 18 Jul 2008 at 10:07

what voltage have you got running through the chip in the bios? could possibly be that it reset to auto in the bios and thus is pumping a lot more than it actually needs.

IAmATeaf · 18 Jul 2008 at 10:11

MURDEROUS G said:
That might be a contributory factor, but a 17°C jump in temps? I don't see how the surface of the new CPU could cause such a huge change unless it was, like, spherical or something.

You'll be surprised, the surface of my Q6600 was completely concave but looking at your temps between the cores I'd hazard a guess that that isn't the case, for me I had something like a 10C diff between 2 of the cores.

Have you tried it under load to see what happens and what thermal gunk are you using? Did you clean off the HSF before remounting? If so what did you use to clean it?

MURDEROUS G · 18 Jul 2008 at 10:54

Thanks for all the responses.

what voltage have you got running through the chip in the bios?

Uhm, whatever the BIOS defaults to with everything on Auto. Haven't changed anything yet, just put it in, turned the machine on, thought "Jesus those idle temps are high" and ran to post on OcUK. Looking at CPUZ, the 'Core Voltage' readout hovers in the 1.10 to 1.15 range, which strikes me as very low. I'll set the volts manually in the BIOS anyhow just to be certain.

Have you tried it under load to see what happens

Not really (like w/ Prime95 or anything), I'm concerned about melting it. I did have a tinker with some processor-intensive work applications but the temp rapidly approached 70°C and I shut them down worriedly.

what thermal gunk are you using? Did you clean off the HSF before remounting? If so what did you use to clean it?

Arctic MX-2; yes; a soft cloth.

tatts · 18 Jul 2008 at 11:04

I thought Mx-2 needed more than AS5? http://www.arctic-cooling.com/downloads/installation/install_mx2_retail.pdf

It looks to me like the hs/f isnt seated right, even the stock hs/f doesnt hit 70C

MURDEROUS G · 18 Jul 2008 at 12:24

sigh

OK, so I take the whole rig apart, remove the HSF, clean off the thermal paste, re-apply as per the instructions in the .pdf linked above, re-seat the HSF, put everything back together and start up again. CoreTemp reports idle temps of.... 35°C!!!! Great, fantastic, brilliant. Victory at last. I guess it had to be a badly seated HSF after all? I potter around for a while browsing, checking CPUZ and SpeedFan and such. 35°C the whole time.

I run a Prime95 test just for a few minutes to see what the load temp is, very roughly. It looks to be about 55°C. Wonderful. I stop Prime95 and wait for the temps to return to normal. They drop back to 48°C...

...

...

...and stay there.

Following is a graph of shortly before, during, and after the P95 test. Note the idle temp of 35°C before load, and then after load, the baseline idle temp has inexplicably risen 13°C to 48°C. It absolutely will not drop below this temp now.

What

the

****.

I've given it another 20 minutes and they haven't shifted. 48°C. 0% load. In fact, in the time I've given them to cool off, the temps have actually crept back up to the ubiquitous 50°C.

Somebody put me out of my misery here. What the hell is going on?

tatts · 18 Jul 2008 at 12:41

What hs/f is it? it looks like its struggling tbh or there isnt enough airflow getting through the case

The fan on the hs isnt temp controlled is it?

MURDEROUS G · 18 Jul 2008 at 12:47

tatts said:
What hs/f is it? it looks like its struggling tbh or there isnt enough airflow getting through the case

It's an Arctic Freezer 7 Pro. How can it be anything to do with airflow...? As I said above, my last Q6600 was installed in this exact same case with these exact same components, same placement of the wiring within the case etc., ergo same airflow, same ambient temps, same everything, and I didn't have any of these issues. If it was airflow or the HSF being unable to handle the heat a Q6600 produces then I would have had the same thing with the last processor, but I didn't.

The fan on the hs isnt temp controlled is it?

Right now it's on its default profile which I think is a flat 1.2k rpm. [edit] In other words no, I don't think it adapts reactively to temps atm.

ultrakipper · 18 Jul 2008 at 12:54

Your first Q6600 wasn't running hot ?

MURDEROUS G · 18 Jul 2008 at 12:56

ultrakipper said:
Your first Q6600 wasn't running hot ?

Nope. 33°C idle at stock.

IAmATeaf · 18 Jul 2008 at 13:01

Have had a feel around the base of the HSF, is it warm or hot?

MURDEROUS G · 18 Jul 2008 at 13:04

IAmATeaf said:
Have had a feel around the base of the HSF, is it warm or hot?

It's pretty hot, but not significantly more so than any other part of it. What am I looking for here?

tatts · 18 Jul 2008 at 13:06

MURDEROUS G said:
It's an Arctic Freezer 7 Pro. How can it be anything to do with airflow...? As I said above, my last Q6600 was installed in this exact same case with these exact same components, same placement of the wiring within the case etc., ergo same airflow, same ambient temps, same everything, and I didn't have any of these issues. If it was airflow or the HSF being unable to handle the heat a Q6600 produces then I would have had the same thing with the last processor, but I didn't.

No 2 components are ever the same though so dont expect the same temps.
Airflow can easily affect temps (not saying it is but a possibility) as without it the warm air just hangs around pushing the case ambient higher.

When you removed the hs/f, how well spread was the paste? my quad was quite badly concaved even though temps were near the same core to core

MURDEROUS G · 18 Jul 2008 at 13:16

tatts said:
No 2 components are ever the same though so dont expect the same temps.
Airflow can easily affect temps (not saying it is but a possibility) as without it the warm air just hangs around pushing the case ambient higher.

Yup, as I said before, I appreciate that temps vary by component, it's just that 17°C is a bit much! Wrt airflow again, I'm just saying that I doubt it is the cause, because all I have done is taken one processor out of the machine and put another one in. If there is an airflow problem now, then logically there would've been one before, because nothing about the physical configuration of the rig (or the stuff around it on the desk even) has changed.

When you removed the hs/f, how well spread was the paste? my quad was quite badly concaved even though temps were near the same core to core

It was spread evenly across the processor, and the whole surface of it had that kind of 'blistered' look - as in Huddy's pic further up - that indicates contact with the heatsink. To my (admittedly untrained) eye, it didn't seem like any part of it had been out of contact.

tatts · 18 Jul 2008 at 13:39

Very strange, only thing i can think of is the hs is either struggling (which it shouldnt) or its not sitting right. The only other possible cause is ambient temp (which you say is the same) or even a concave ihs but the core temps seem to be close enough to suggest it isn't.

There the only things that can cause this looking at the speedfan graph as it cant seem to shift the temps back down ounce its been under load.

Out of interest, is it a retail cpu? (may be worth trying the stock hs to rule some causes out?)