Having trouble keeping my Q6600 stable @ 3.2ghz

Mr. Lennon

Diamond Member
Jul 2, 2004
3,492
1
81
CPU is 350x9 with 1.340V
My OCZ 8500 Reaper ram is running with 2.150V
FSB:DRAM 2:3

Temps Idle:
core 0: 42C
core 1: 43C
core 2: 33C
core 3: 38C

Load in Prime 95: 62C


Now I have gamed many hours at this setting. When I run Prime95 though each core will begin to give up. Should I not worry about prime95 since I have yet to have a crash or error during many hours of gaming?
 

AmberClad

Diamond Member
Jul 23, 2005
4,914
0
0
I couldn't get my E6600 stable at 3.2GHz with stock vCore. Needed a small bump (to around 1.38v).

Personally, I wouldn't feel comfortable running a system that isn't Prime stable.
 

taltamir

Lifer
Mar 21, 2004
13,576
6
76
i wouldn't even feel comfortable running a system that is prime stable but hasn't been given a thorough workout in other ways...

Your machine is unstable, simple as that... you might be able to play a specific game fine, but that doesn't mean it is not unstable, it just means that game is underusing your CPU (and thus, probably doesn't even need that OC)
 

Idontcare

Elite Member
Oct 10, 1999
21,118
58
91
Originally posted by: Zeppelin2282
Now I have gamed many hours at this setting. When I run Prime95 though each core will begin to give up. Should I not worry about prime95 since I have yet to have a crash or error during many hours of gaming?

Well if you are curious what part of your OC is failing (you are assuming it is the Q6600 by your topic title, but could be ram, MCH, FSB, etc) then a deeper diagnostic might be helpful.

For example, keep everything the same about your OC but drop your multi from 9 to 8. This will reduce the CPU overclock while keeping everything else just as challenged as before. Now see if Prime95 small FFT passes. If it does then check large FFT.

There are numerous methods for isolating your OC'ed components to determine which is the weakest link, this is just one of those methods.
 

BonzaiDuck

Lifer
Jun 30, 2004
15,726
1,456
126
IdontCare has the insights here -- better than mine -- that I was going to point up.

But -- for further second-hand observations we might be able to offer:

What motherboard are you using? Which chipset?

Which stepping of the Q6600 -- B3 or G0?

I can tell you from my continued use of the B3 stepping -- you can get it stable at 3.2Ghz, but you need to tweak the voltages mentioned by IdontCare, and I had to push the VCORE beyond 1.41V -- most likely between 1.42 and 1.43. Of course, every processor will vary. I also had to run up my CPU_VTT termination voltage to a setting of about 1.45V, although "set," "monitored" and "real" are three different things. And the voltage options will depend on your chipset.

The G0 stepping, from what I've gleaned at this forum since it was released last July 2007, requires less voltage. You might be able to get it to 3.4Ghz, and it is still all "the luck of the draw."

After a while, I chose to bump down the speed to 3.15Ghz. It would run PRIME95 small-FFTs or Blend-Test all day and night. But again -- "which motherboard and chipset?"
 

Mr. Lennon

Diamond Member
Jul 2, 2004
3,492
1
81
Thanks so far for the feedback everyone.

Motherboard is a Abit IX38QuadGT

Q6600 is a GO
 

Mr. Lennon

Diamond Member
Jul 2, 2004
3,492
1
81
Ok after upping the voltage a bit more I have yet to have a core fail on me. Only issue is the heat. I saw the load temps on core 0 get to 75C. Is that temp ok?
 

Idontcare

Elite Member
Oct 10, 1999
21,118
58
91
Originally posted by: Zeppelin2282
Is that temp ok?

Yes it is. Also consider that your chip won't likely be seeing those load temps unless you run Prime95 as an application.

What your data tells you is that your system, at it's current Vcore, is stable running pretty intensive calculations at the current GHz even when the temps are 75C.
 

myocardia

Diamond Member
Jun 21, 2003
9,291
30
91
Originally posted by: Idontcare
What your data tells you is that your system, at it's current Vcore, is stable running pretty intensive calculations at the current GHz even when the temps are 75C.

True, but it also tells me that he might want to invest in a better cooler. Tuniq Towers are pretty cheap these days.:)
 

AmberClad

Diamond Member
Jul 23, 2005
4,914
0
0
O RLY? I would have expect better results than that. The base on it is a little rough though, if you haven't lapped it.
 

chizow

Diamond Member
Jun 26, 2001
9,537
2
0
Looks like you're on your way to getting Prime stable. What I've found is Vcore droop and FSB Term (VTT on NV boards) can have a huge impact on Quad overclocks. If your board has a lot of droop (compare load voltages to settings in BIOS) you may have to crank up Vcore way up to get stable load voltages. This obviously becomes a problem as you hit Vcore ceilings or very hot idle temps to compensate for Vdroop. The other setting, FSB Term/VTT also seem integral for Quad overclocking. I wasn't able to OC my Quad on my NV boards at all without adjusting VTT to max 20%.
 

Mr. Lennon

Diamond Member
Jul 2, 2004
3,492
1
81
Originally posted by: myocardia
Originally posted by: Zeppelin2282
myocardia: I have a XIGMATEK HDT-S1283

Ooh, you shouldn't be hitting those temps, then. You should consider reading this excellent tutorial on How to apply TIM to HDT coolers. It's very informative.

edit: Your Xigmatek is better than my Scythe Ninja, yet here's the temps I get @ 3.2 Ghz.

Oh wow and to think I thought I setup my cooler perfectly. So I soak sand paper in water then rub down the bottom surface? Is there a particular type of sand paper I should buy or does it not matter? Also, whats the best way to completely clean off the MX-2 paste I already have on there?

Thanks again guys.
 

myocardia

Diamond Member
Jun 21, 2003
9,291
30
91
You could lap your Xigmatek if you want. Most "lappers" start with 400 grit, then progress to 800 or 1,000 grit, if I remember correctly. Some actually go as high as 1,500 or 2,000 grit, but it isn't really necessary-- that's what the TIM is for, to fill in the small spots where metal isn't touching metal.

The reason I gave you the link was to let you know that you can't use the "normal" way of applying your TIM, when you have an HDT (heatpipe direct touch) heatsink. If I were you, I'd try applying it the way the guy did at the bottom of the linked page, without lapping, at least at first. You really should be getting at least 8 or 10°C cooler temps with your heatsink, even if it isn't lapped.
 

Drsignguy

Platinum Member
Mar 24, 2002
2,264
0
76
Originally posted by: Zeppelin2282
Originally posted by: myocardia
Originally posted by: Zeppelin2282
myocardia: I have a XIGMATEK HDT-S1283

Ooh, you shouldn't be hitting those temps, then. You should consider reading this excellent tutorial on How to apply TIM to HDT coolers. It's very informative.

edit: Your Xigmatek is better than my Scythe Ninja, yet here's the temps I get @ 3.2 Ghz.

Oh wow and to think I thought I setup my cooler perfectly. So I soak sand paper in water then rub down the bottom surface? Is there a particular type of sand paper I should buy or does it not matter? Also, whats the best way to completely clean off the MX-2 paste I already have on there?

Thanks again guys.


If you do decide to lap your HS, I wouldn't do it the way you described. My suggestion would be getting a flat surface like a piece of glass or mirror and place the sandpaper onto that with tape and Then do the lapping. The whole point of lapping it to get a very flat surface for best results of heat transfer.


Edit: Wasn't aware of your HS being a direct touch heatpipe. Maybe I should have read into that a bit more. Me bad:) Best advice is below post below.
 

Idontcare

Elite Member
Oct 10, 1999
21,118
58
91
Originally posted by: VirtualLarry
I wouldn't lap a HDT cooler.

Yeah I wouldn't recommend it either. I would lap the IHS but I wouldn't want to risk messing around with lapping a hole thru a heatpipe on a direct touch HSF.