Q9450 just won't be stable at 3.2GHz

[Goi]

Hi,

I apologize for the long post in advance.

I've had my Q9450 for since early June, but I didn't start overclocking until maybe a month ago. I've been incrementally and conservatively upping my frequency until I hit a roadblock that required slightly increased voltages.

I'm trying to get it stable at 3.2GHz, but it just won't be stable at 400MHz FSB. 390MHz FSB was fine for over 8 hours of prime95, but 400MHz won't even get an hour. It's always core0/1 that gives the error. Cores 2/3 are fine.

390MHz FSB required only a modest increase in voltage. I can't remember the exact value but it was just a couple notches above the default, way below 1.2V. I'm now up to 1.275V and 400MHz FSB still isn't stable. In fact, this past week or so the system has been getting less stable at a higher voltage than it was at a lower voltage, as in the system would crash when doing slightly intensive stuff like playing videos, or CoD4. Before that it didn't happen even though prime95 failed.

I've kept my memory speeds to 800MHz(DDR2-800) so that shouldn't be the problem.

I've also tried increasing NB voltage 2 notches up from default but that didn't seem to help.

I've read the dual/quad core overclocking FAQ and it seems to suggest that if it's always the same cores failing, it's likely to be a vcore issue and not an NB voltage issue.

Realtemp shows my idle temps in the mid/high 40s and prime95 temps to be in the mid/high 60s. That seems pretty high to me, and I've got a Tuniq Tower. I live in a hot country where ambient temps are about 30 though. Basically summer all year round.

My system specs are listed below
Q9450
DFI LanParty LT X48-T2R
Tuniq Tower w/default fan@full speed
2x2GB Corsair XMS2 DDR800
Leadtek 9800GTX 512MB
2x1TB Samsung F1s, 2 other smaller SATA drives and 2 other smaller IDE drives
FSP Epsilon 600W
Lian Li A7010B full tower

Am I just being paranoid with my voltage and temps, or do I just need to increase the vcore? Any help would be appreciated.

Thanks!

 

[myocardia]

You'll likely need to raise your CPU VTT a notch or two, to be stable @ 400 FSB or above.

 

[Drsignguy]

Hi Goi, here is some info for you to compair:

The Vid on my Q9450 is 1.2500v and to get to 3.4, at current OC, I had to bump up the voltage to 1.31875 to become stable. I used cpu-z to watch the load voltages droop to 1.232v - 1.248v levels and when it reached these values, it has been stable at this point.

Here is the kicker though. Depending on the vdroop your board has is where you need to bump up the voltages accordingly to have the voltages around your chips vid.

My EP35C-DS3R board used a 1.29175v @ 3.4 and my EP45-DS3R board was different, Has a bigger vdroop I've noticed, so the voltage is set at 1.31875v.

I too have the Tuniq and temps @ full load using OCCT, Prime and F@H no higher than 60c, steady at 57c depending on the room temps.

 

[jadeiceman]

I have a Q9450 too, and I remember that I had a load of trouble because I wasn't running my RAM at a high enough voltage. No matter what I did, Prime95 would always fail on some cores. After I changed the voltage, I was able to get 24 hours prime stable on 3.2Ghz with my VID, also 1.2500v. Someone also told me that if one of the pins on the motherboard are crooked or something, the processor will never be stable. If you have the time, you could check that out too. Oh, and remember that Intel released the Tj Max of 45nm procs, so the Tj Max of the Q9450 should be 100c, not 95c. Good luck!

 

[Goi]

OK, I've bumped the Vcore to 1.3V, NB core still 2 notches above stock, and I've also bumped the VTT a few notches from 1.1V to 1.19V. Still no go at 3.2GHz, though it lasts longer into Prime95. I got over 7 hours on cores 1/2, core0 still running after 8+ hours(when I woke up and stopped it), and core 3 failed at 3+ hours.

Weird, cores 2/3 never failed before, it was always cores 0/1. Now core0 is running on yet the other cores fail?

Load temps are now slightly over 70C, which is definitely worrying, especially with a Tuniq Tower! wtf!

Should I go even higher?

 

[dug777]

Those temps would scare me

Are you sure that Tuniq is seated properly, thermal paste correctly applied, that you have some air-flow in your case?

 

[Goi]

I'm beginning to suspect those 3 things too. I was pretty careful when I mounted the Tuniq. As for the thermal paste, I used the default Tuniq thermal paste which is known to be hard to apply, but I thought I did pretty well in getting a thin layer across the entire CPU heat spreader. As for air flow, it's a full tower Lian Li A7010B with the stock fans in there set to max, so shouldn't be a problem? I did notice that the exhaust at the top is being blocked by a metal brace of my table so I shifted the case further out, hopefully that will help a little.