Q6600 Voltage & FSB Questions

[OrbitzXT]

I want to OC my Q6600 a little, it has a Zalman 9700 on it with AS5. I followed a guide I found to OCing on the Gigabyte 965P-DS3 and turned off most the extras as suggested. I set the FSB to 320, and my Memory Divider to 2.5, this way I get the small comfortable OC I want and my Memory runs at 800MHz still. I didn't adjust any voltages at all and the system posted fine and booted into XP. As soon as I'm done with this post I'm going to run Prime95 for awhile. I realize the normal strategy is to set a higher voltage then keep lowering it, but since a 320FSB is my goal and a relatively small OC, I thought I'd work backwards. In the BIOS, the voltage is on default which I think was 1.2375v, but CPU-Z shows as 1.184v. Here is a screenshot for now, but I'd like to hear from Q6600 owners or smart people in general who can estimate what voltage I'll end up needing to be stable.

 

[hat]

Run 334.25FSB so you can have 1337 rated FSB

 

[b1lk1]

There is no set amount of Vcore needed. You have to figure that out. My X3210 is stable @ 366FSB (2.92GHz) with stock Vcore of 1.25V. I haven't tried for more. Either way, if it fails Orthos/Prime then it's time to add more Vcore.

 

[OrbitzXT]

Which Prime95 test is the best to run to stress a CPU and test stability? I did the Small FFTs and so far its been stable about 3 hours at stock voltage and 320 FSB. I plan on letting it go 24 hours to make sure.

 

[newtekie1]

I have a similar setup to you, just a different motherboard, and at 3GHz(333FSB) I need 1.4v to keep it stable plus several other voltage increases on the motherboard.

You need to run multiple instances of Prime95 to test for stability. Since you have 4 cores you need to run it 4 times. Personally I would download SP2004. I unzip it to 4 different folders(you have to do this with Prime95 also). Then I run it from each folder. It is a much better GUI than Prime95, but runs the exact same stability test. Then I set each instance of SP2004 to use a different CPU(0-3). I set the SP2004 window that is testing CPU-0 to "Blend", then I set the rest to CPU.

Here is a screenshot of how it looks when I test my quad-core for stability.

 

[Mussels]

Which Prime95 test is the best to run to stress a CPU and test stability? I did the Small FFTs and so far its been stable about 3 hours at stock voltage and 320 FSB. I plan on letting it go 24 hours to make sure.

small FFT is the best one for CPU testing, but make sure its testing BOTH cores - i remember prime 95 only did one core, not dual (i use Orthos, which is a dual core version of prime 95's stress test)

Really, just watch the temps too - keep em under 60C at load, and you'll have no problems. Things seem to be going smooth for you and you've got the basics covered, so good luck.

 

[OrbitzXT]

The lastest version you only need to run one instance even for quad cores, its been stable about 5 hours now at stock voltage.

http://i65.photobucket.com/albums/h204/OrbitzXT/Prime95.jpg

 

[OrbitzXT]

Also I have another question, do people ever overclock to an extreme number knowing that if they ran a stress test for an extended period of time it'd fail, but they ignore that and use it as is? The thing I'm thinking is, most of us don't do anything that puts our CPUs under the same amount of stress as when they're undergoing a stress test.

Like my system now with it's OC and stock voltage, has been running fine with about an hour of gaming and a few hours of stress testing. If this OC turns out "unstable" after 20 hours or so, what is the harm in just keeping the settings like this? If it takes 20 hours of consecutive 100% load to make it crash, won't it run fine for the normal use? I'm just throwing weird ideas out there trying to understand why most people feel a system has to run completely stable under the heaviest of loads for a day straight or longer.

 

[newtekie1]

The problem with doing that is that the instability can show up at any time and it doesn't have to be under load to fail. Putting it under load just tends to make it fail quicker.

My old P4 system was like this. Overclocked from 3.0GHz to 3.2GHz seemed perfectly stable and would only fail Prime95 after about 24 hours or testing, but it would tend to blue screen whenever it was idle for long periods of time, or crash whenever I played a movie in media player.