Overclocking a Q9550

[SnoopKatt]

Hey guys,
About a week ago, I bought a Q9550 from another forum member. Before then, I have my E6420 overclocked from 2.13 GHz to 2.66 GHz (and up to three in the winter). Now, I just installed the new CPU, and my temperatures are already too high. Idle, it is at about 50°C, and under prime 95 after 30 seconds, it's between 65 and 70. It still running stock speeds and voltage. I thought it might be my case, so I moved everything from Antec P180 to a cCooler Master HAF912, and my temperatures are exactly the same.

I feel like I did everything right, but it's been a while, so I probably screwed up somewhere. I had some thermal paste lying around (Arctic Silver), and I applied about a really small pea, and spread it evenly with a credit card.

All of the system fans are going, it's a little warm in there, but definitely not hot. Any suggestions?
Thanks!

 

[FreedomEclipse]

are you using that N520 in push/pull config?

 

[SnoopKatt]

I believe they are both blowing air into the CPU heatsink. I should probably change that to a real push/pull.

I'm also a little suspicious that CoreTemp is reading 10° high. In the bios, after some use, it hangs around 40°C. Right after I load Windows, it is between 50 and 55°. 40° still seems too hot though.

Thanks for the quick response!

 

[Yukikaze]

I believe they are both blowing air into the CPU heatsink. I should probably change that to a real push/pull.

I'm also a little suspicious that CoreTemp is reading 10° high. In the bios, after some use, it hangs around 40°C. Right after I load Windows, it is between 50 and 55°. 40° still seems too hot though.

Thanks for the quick response!

If they are both blowing into the heatsink that might well be your problem. I used to own a Q9650 and a Scythe Katana 3, which is a small and not very high-performance cooler, could handle it at 4Ghz (With around 1.2v on the cores) without hitting the temperatures you are reporting now at stock.

 

[TC-man]

Hey guys,
About a week ago, I bought a Q9550 from another forum member. Before then, I have my E6420 overclocked from 2.13 GHz to 2.66 GHz (and up to three in the winter). Now, I just installed the new CPU, and my temperatures are already too high. Idle, it is at about 50°C, and under prime 95 after 30 seconds, it's between 65 and 70. It still running stock speeds and voltage. I thought it might be my case, so I moved everything from Antec P180 to a cCooler Master HAF912, and my temperatures are exactly the same.

I feel like I did everything right, but it's been a while, so I probably screwed up somewhere. I had some thermal paste lying around (Arctic Silver), and I applied about a really small pea, and spread it evenly with a credit card.

All of the system fans are going, it's a little warm in there, but definitely not hot. Any suggestions?
Thanks!

Hi,

It's not uncommon that certain Q9550 got a stuck temperature sensor, which means the idle temperature stays at a certain level, it does not matter what you do. And if this is the case with your Q9550 then you can't do anything about it. You can try using Realtemp instead and check the "Distance to tj Max" which is more important when you are stress testing the cpu. I am told for a healthy temperature keep the distance to tj Max no less than/close to 30.

On the other hand, try re-applying the thermal paste/compound, but this time without spreading the compound with a creditcard; perhaps air bubbles are the culprits for the high temperatures, the air bubbles may have been created with such compound spread method after you attached the cpu-cooler on the cpu/socket.
See the youtube video (see at 0:10, and especially at 0.36):

And one more thing, if you haven't cleared cmos/bios (including detaching the cmos battery) after the cpu upgrade, do that first. Sometimes weird things may happen after a cpu switch, e.g. weird temperature readings and so on. This is the easiest thing to do, it just may help.

 

[SnoopKatt]

Thanks for all of the advice!

Before I try that again, attached is a quick diagram of how my fans are set up. I'm guessing I should set up the CPU fans so they eject air into the exhaust to the left of it, but just to double check, is that the best way to do it?

I downloaded RealTemp, and the temperatures and the distance to Max all add up to 100. The distance to Max temperatures consistently move with the detected temperatures, does that mean the temperatures aren't stuck?

Thanks, and I will post back with the results later today!

 

[SnoopKatt]

So I gave it a try, and no luck. It's now a push pull system, and I redid the thermal grease. Maybe I tweaked something on my cooler mount when I installed the new CPU?

 

[Jetster]

Try reapplying the paste and use a small grain of rice in the middle and don't spread it. What are you using to check the temps?

 

[SnoopKatt]

I tried the same method, except with a small pea. I did not spread afterward, but unfortunately the temperatures are the same. I'm using CoreTemp.

Thanks for the quick response!

 

[TC-man]

Thanks for all of the advice!

Before I try that again, attached is a quick diagram of how my fans are set up. I'm guessing I should set up the CPU fans so they eject air into the exhaust to the left of it, but just to double check, is that the best way to do it?

I downloaded RealTemp, and the temperatures and the distance to Max all add up to 100. The distance to Max temperatures consistently move with the detected temperatures, does that mean the temperatures aren't stuck?

Thanks, and I will post back with the results later today!

So I gave it a try, and no luck. It's now a push pull system, and I redid the thermal grease. Maybe I tweaked something on my cooler mount when I installed the new CPU?

The diagram for the fans and airflow is about right. But did you reverse the fan on the rear of the N520? You shouldn't, because it's designed like this and the airflow should be right according it's design. According Frostytech the fans should be aligned/mounted on the heatsink of the N520 like this. Are they like this at the moment?

As for the temperatures, what voltages did you use for the cpu (Vcore)? Use CPU-Z and check what the Vcore is when the cpu is idle and under load, especially you use the auto settings in the bios. Furthermore, I see you used Prime 95 to do a burn test. What type of test was it, small FFT, in-place Large FFT or blend? The first two tests will cause much higher temperatures of the cpu. And what bios version are you using for the Gigabyte GA-P35-DS3L (rev.1.0 or rev. 2.0?)? As for the stuck temperature, it's more like the temperature is stuck at certain °C, e.g. is always 50°C at idle, but the temperature will increase when doing a stress-test (e.g. LinX, Prime95) and the Distance to tj Max will decrease.

Also I found something interesting at Xbitlabs about the the N520:

The installation manual recommends installing the cooler into the system case so that the airflow from the fans could be directed towards the back of the case. However, our experiments revealed that the cooler performed best if turned by 90°:

In this case the CPU temperature under maximum workload was steadily 3-4°C lower than in the first case. The distance from the mainboard PCB to the lowest heatsink plate of the Cooler Master Hyper N520 cooler is 36.6mm, so the cooler will not interfere with any electronic components around the CPU socket.

 

[OneMoar]

cooler just is not up to the job the q9xx chips run very very hot replace the cooler with something better plain and simple
http://www.guru3d.com/article/cooler-master-hyper-n520-review/5 < review and temps

 

[InnocentCriminal]

I'm idling just above 40°C on 3 out of the 4 cores, and one just below. Maximum temps I get when gaming is 60-65°C and 70-75°C when stressing with Prime95.

You may find that once you OC, that the temps don't change that much. I'm running mine at 3.75GHz with 1.36v. My northbridge is the problem.

 

[FreedomEclipse]

cooler just is not up to the job the q9xx chips run very very hot replace the cooler with something better plain and simple
http://www.guru3d.com/article/cooler-master-hyper-n520-review/5 < review and temps

Did you even read the review you posted?

taken from the bottom of the page of the review that you linked to:

As such, as you can see, the Hyper N520 can hold ground pretty decent. It is competing with OCZ's Venedetta 2 and Zalman's CNPS 9900 LED cooler. So yeah, that's just fairly good.

temperatures only hit a serious high when Hilbert put 1.6v through it. Most E0 steppings didnt need even 1.45v to get a good OC.

On top of that - my previous rig was a Q9550 OC'd to 3.9Ghz with a OCZ Vendetta 2 and i was idling at 35'c and hitting temps of 65'c under load. the OCZ was in the same league as the Xigmatek HDT-S1283 which was given rave reviews buy almost every tech site.

So it could be ambient temps thats the problem

 

[OneMoar]

Did you even read the review you posted?

taken from the bottom of the page of the review that you linked to:


temperatures only hit a serious high when Hilbert put 1.6v through it. Most E0 steppings didnt need even 1.45v to get a good OC.

On top of that - my previous rig was a Q9550 OC'd to 3.9Ghz with a OCZ Vendetta 2 and i was idling at 35'c and hitting temps of 65'c under load. the OCZ was in the same league as the Xigmatek HDT-S1283 which was given rave reviews buy almost every tech site.

So it could be ambient temps thats the problem

yes I read it thats why I posted it then benchmark they ran shows a avg of 66c at 3.2/1.2v with a QX9770
they run hot especially the older stepping temps in excess of 70c don't surprise me at all given that cooler paste application can give you 2 or 5c +/- either way they unless his room is like >75F it should not make that much of a difference either way some q9550's just ran screaming hot

 

[Mindweaver]

I still have a Q9550 @ 3.8 ghz crunching strong 24/7. If i where you I would get a COOLER MASTER Hyper 212 Plus and set up Push/pull. I had a problem with my old 92nm heatsink and temps.. It was fine with my duel core but when I put it on the Q9550 and over clocked it.. it chocked.. hehehe and the Q9550. Mine had a stuck sensor as well, but it only miss read one of the cores not all 4 cores. I have mine setup using v1.4. I also lapped mine, and dropped it a few degrees.

 

[Widjaja]

I shame you could not stress test the CPU in another motherboard.

I have had situations where motherboards have made the processor run hotter than in others.

 

[Bo$$]

What is your Voltage set to?

 

[AlienIsGOD]

im glad im learning it wasnt just me with stuck temps on a core for a Q9xxx series. My q9450 ALWAYS had a core 5-6 degrees hotter than the other three. I used a CM GeminII on mine and it was a modestly decent cooler for light o/c. I ran my Q9450 @ 3.2ghz for almost 3 years at 60-60 degrees under load.

 

[Sinzia]

...I had some thermal paste lying around (Arctic Silver)...!

AS5 has a cure time on it, and from what I remember its a rather long cure time, this can make an impact on temps.
If you can, try something else like mx-2 or IC-diamond 7, those have 0 cure time.

 

[SnoopKatt]

Thank you everyone for your input! This is all good information to know.

TC-man, the way I initially had the CPU cooler was exactly like what was written in the article. I would rotate it 90°, but unfortunately it hits my RAM. Even in the push pull set up that it's in right now, the temperatures are about the same.

As far as the motherboard goes, it is revision 2.0 with an F9 bios.

My voltage is set to 1.2 in the bios, but looking at CPU-Z, at idle, it says 1.04, and under load, it says 1.12.

None of the temperatures seem to be fixed at a specific value. Idle, it ranges between 48 and 52° on each core. However, if I start the computer cold, in the bios it says between 30 and 35°. Under about 5 min. of Prime 95 blend, each core is at about 70° (30 from Max).

I'm open to trying another CPU cooler. That Cooler Master 212 sounds good, and it's good to hear a success story from a fellow Q9550 user. What fan would be good to match it with for the pull?

I also ordered an external temperature sensor (I will need it later down the road for another project anyways) to double check the internal sensors.

 

[OneMoar]

Thank you everyone for your input! This is all good information to know.

TC-man, the way I initially had the CPU cooler was exactly like what was written in the article. I would rotate it 90°, but unfortunately it hits my RAM. Even in the push pull set up that it's in right now, the temperatures are about the same.

As far as the motherboard goes, it is revision 2.0 with an F9 bios.

My voltage is set to 1.2 in the bios, but looking at CPU-Z, at idle, it says 1.04, and under load, it says 1.12.

None of the temperatures seem to be fixed at a specific value. Idle, it ranges between 48 and 52° on each core. However, if I start the computer cold, in the bios it says between 30 and 35°. Under about 5 min. of Prime 95 blend, each core is at about 70° (30 from Max).

I'm open to trying another CPU cooler. That Cooler Master 212 sounds good, and it's good to hear a success story from a fellow Q9550 user. What fan would be good to match it with for the pull?

I also ordered an external temperature sensor (I will need it later down the road for another project anyways) to double check the internal sensors.

it comes with a cooler master blade
COOLER MASTER R4-BMBS-20PK-R0 Blade Master 120mm C...

 

[arnoo1]

it's pretty wierd, i had a q9550 and teh idle temps where a fake 23-25c idle on a 20c room and load 50c with a scythe ninja 2 @ 3.6ghz 1.25v
maybe use other temp programms core does wierd things with my 2600k, realtemp gt works just fine

 

[m1dg3t]

I just wanted to know how can people reccomend the 212 when the 520 is a beter cooler, supposed to be anyways? Push/pull is best and orient the cooler so that the exhaust is directed to case exhaust, if possible.

I have a VenomousX cooler but looking to swap it for a d14 or SilverArrow sb-e if I can find one, my case flow is pretty horrid (HTPC on/in a rack). My current max temp is 70c after 6Hrs ++ Prime stress. I'm currently @ 444 x 8.5 with 1.30vcore, 4 x 2GB RAM @ 1066 I've spent the last week or 2 trying to figure out the best GTLs and think I've found the sweetspot for my rig so gonna try to drop the vcore and see if she stays stable

I have a q9550 as well, c1 not e0. I have a stuck sensor, it read's ~10c higher on idle. Should be pretty easy to get 3.4 at least out of it. I'm away from home ATM but I'll chime in later with some ideas you can try. Best to start gathering stress test/monitoring software now and do some reading before you start clocking it, I see to many people just hitting the chip with volts to get it stable instead of making the small stability tweak's. There's a good link to GTL info on the 2nd last page of the Asus p45 owners club on our forum. From what I've been able to understand is the stock GTL voltage for these quad's is ~.8v so when Calculating your multi remember to leep it in that range, VTT or FSBtermination voltage it's called example mine is set to 1.26 Fsb term v and my multi is .635 & .645 IIRC, this has been the best so far for my combo. When I get back home I'll post some more info, hopefully, to help get you going. Don't take this the wrong way but hit up Google as well, lot's of info out there amd prolly lot's that will get left out of thos discussion lol Gotta remember these are now vintage chip's lol

 

[Mindweaver]

I'm using 2x Scythe SY1225SL12SH 120mm "Slipstream". They are 110.31 CFM fans and I use a cheap Potentiometer(around $2) to control both fans. It's a cheap fan controller.. Or you can buy a good fan controller. Here is a nice little cheap one that will do the trick and look good doing it.

 

[SnoopKatt]

I just wanted to know how can people reccomend the 212 when the 520 is a beter cooler, supposed to be anyways? Push/pull is best and orient the cooler so that the exhaust is directed to case exhaust, if possible.

I have a VenomousX cooler but looking to swap it for a d14 or SilverArrow sb-e if I can find one, my case flow is pretty horrid (HTPC on/in a rack). My current max temp is 70c after 6Hrs ++ Prime stress. I'm currently @ 444 x 8.5 with 1.30vcore, 4 x 2GB RAM @ 1066 I've spent the last week or 2 trying to figure out the best GTLs and think I've found the sweetspot for my rig so gonna try to drop the vcore and see if she stays stable

I have a q9550 as well, c1 not e0. I have a stuck sensor, it read's ~10c higher on idle. Should be pretty easy to get 3.4 at least out of it. I'm away from home ATM but I'll chime in later with some ideas you can try. Best to start gathering stress test/monitoring software now and do some reading before you start clocking it, I see to many people just hitting the chip with volts to get it stable instead of making the small stability tweak's. There's a good link to GTL info on the 2nd last page of the Asus p45 owners club on our forum. From what I've been able to understand is the stock GTL voltage for these quad's is ~.8v so when Calculating your multi remember to leep it in that range, VTT or FSBtermination voltage it's called example mine is set to 1.26 Fsb term v and my multi is .635 & .645 IIRC, this has been the best so far for my combo. When I get back home I'll post some more info, hopefully, to help get you going. Don't take this the wrong way but hit up Google as well, lot's of info out there amd prolly lot's that will get left out of thos discussion lol Gotta remember these are now vintage chip's lol

Thank you; I would greatly appreciate that!

I still have my old CPU, and with the 520, on a cold day, I was able to get it from 2.13 up to 3.12 stable under 60°C. Maybe I should try reinstalling it and see what happens (is it possible I damaged something while installing the new CPU)?

Thank you Mindweaver and OneMoar for the fan suggestions!

Sinzia I think you're right about the cure time for the Arctic Silver; my temperatures are a little lower now. While browsing the Internet, they seemed to be between 45 and 48°.