New Q9550 - temp issues

[brakeinup]

I bought a new Q9550 to upgrade my E6600 (was running at 3.1Ghz). The Q9550 was installed with the current cooler (Arctic Freezer Pro). I was not happy with the idle temps and reading some OCUK reviews I bought a Zalman CNPS9500 AT cooler. The system is all on auto so no overclocking.

When installed with the included thermal paste I got the following temps:

Core 0 - 42, Core 1 - 39, Core 2 - 31, Core 3 - 39

Redid the HSF with MX-2 using a thin line application and got the following:

Core 0 - 42, Core 1 - 38, Core 2 - 29, Core 3 - 39

The point of me buying the Zalman was to reduce idle temps but there was little difference between the Zalman and Arctic Cooler (although I suspect the Zalman would be better under load).

Using Real Temp Core 3 never goes below 39 and says 0 movement for the sensor. Core 0 never goes below 42 but has movement.

Running the intel burn in test the hottest the CPU gets on any core is 64 (coldest is 56). Is a stuck sensor a fault on the CPU and therefore should be RMA'ed?

Are these normal temps (high + low)? Why such a spread of temps? Is this anything to worry about?

MB is an Asus P5K Pro with 4GB RAM running Vista 64.

Thanks

 

[brakeinup]

Thanks for the responses.

Could it be auto settings causing heat issues or do I have a duff CPU?

Nice temps on the setup Derek!

 

[brakeinup]

have you tried re-seating?

Also, you might be correct, the mobo might be giving the processor too much voltage.

download Coretemp, and check what the VID of your processor is, then go into you BIOS and check what the reported core voltage of you cpu is. Set you vcore to whatever the VID is.

Also! Make sure all power saving features are not active when checking VID. Best way to do this is open up programs when looking at coretemp, VID will be up and down. Whateve the highest value is, that's your VID.

I have reseated 5 times now using 2 HSF's and 3 different thermal pastes with the same results. The final way (Zalman with thin line of MX-2) dropped 1 degree on core 1 and 2).

The VID (from memory as I am now at work) I think is 1.25 but I will check the bios when I get home.

Cheers

 

[brakeinup]

Your temp look very good, you have to start worry about the temps when you pass 90C

How so when core 0 sits at 42 on idle and core 3 on 29?

On the intel burn in test the CPU never went above 64.

 

[brakeinup]

My Q9550 is a bit like that, although not quite as large a difference between the cores. I've got one that is about 6-7C cooler than the rest.

It also idles around 40ish, and that's watercooled, but it's the full load temps that really matter.

These Q9550 I think might be that way inclined.

I hope that is the case regarding these and that I do not have a duff CPU.

 

[brakeinup]

Just out of interesty does anyone know what the core layout of this CPU is?

If I am looking at the CPU and reading the text from left to right is core 0 top left and core 1 top right?

 

[brakeinup]

VID is 1.25 so I set it to 1.2 in the BIOS. No difference on temps.

 

[brakeinup]

Intel recommendations for 70C are quite conservative. For example i have a q6600 3.2 and gets 63C idle and 87C load(prime 95) for more than a year

So dangerous temperatures for a cpu is above 90-100C. End of thread.

I would expect a 45nm chip to be running cooler than 42 degrees on one core with decent cooling.

 

[brakeinup]

I have overclocked the cpu to 3.4 and it passes the intel burn in test. The max temp hits 67 (67, 67, 62, 61) degrees in a warm room but the system is stable.

My case is ancient though so airflow could be better. Antec 300 maybe?

 

[brakeinup]

I have left my q9550 stock for the time being as just upgraded from a 0 airflow case to the haf 932 so not began playing yet.

Under full load (prime95) cores highs have hit 58 53 57 56, while on idle i get 36 33 36 36 respectively - ambient was 20c. The heatsink+fan combo is thermalright hr-01 plus + noctua nf-p12 running at 12v. Update to this test, same ambient and using intel burn test with error detection and full test x 5 hit 68 63 63 68, so a full 10c hotter than prime95 could manage.

My vid is running below 1.25 and the processor drops down mhz when idle, ie multiplier goes from 8.5 to 6 (cannot remember which bios setting stop this )

I think its C1E you need to disable and any Intel Speed Step settings if you want your cpu to run at full speed all of the time.

With an ambient of 20 that seems hot (unsure if its OK) - I get less max temps in a hotter room on the burn in test and that is overclocked to 3.4 with 1.2 volts.

 

[brakeinup]

It does seem a bit hot however the heatsink has been in and out a few times so not redone the AS5 (ran out) and my coolermaster V8 should be here today or tom so will redo the temps with my new AS5 tube that I found last night

Could you post the temps when you have redone the HSF please?

 

[brakeinup]

No probs, will prob be at the weekend if our son goes to sleep

Cheers

 

[brakeinup]

I have the same OC and the Antec 300 case... it's a bit cramped in there really, and hence prob not the best for cooling. I'm looking to upgrade to something a bit bigger, maybe the Antec 902

i had an Antec 300 delivered today but am not happy with the quality.

 

[brakeinup]

Ambient of 20c

Idle - 36 33 32 34 (previous 36 33 36 36)

Load - 59 56 56 59 (previous 68 63 63 68)

So idle temps are about the same, if not a few c cooler on some cores, on full load with intel burn test made a 9c difference on cores 0 & 3 and a 7c difference on cores 1 & 2.

BTW all these test were carried out using the slowest fan speed of the v8, currently 1300rpm (ish)

Now this is where the comparison to the thermalright falls down, upon taking the old heatsink off found there to be no paste in the centre, eg you could read the chip markings, just paste around the edges - what was I thinking of when installing it .

Install of the v8 was easy took about 45 mins including removing and cleaning the old one. Did not take the mobo out for the install just held the v8 in place while reaching round with the other hand to screw it down on the back bracket. Also did not use the AS5 but instead used the sample of ThermalFusion 400 paste that came with it - if I come to remove it for any reason then will try the AS5 instead.

Not bad highs and lows. My lowest core gets to 26 on idle and max is on 66 on intel burn in test.