HOWTO: Overclock C2D Quads and C2D Duals - A Guide v1.0

Tat3, thanks for answering buddy. Your post gave me the encouragement to go ahead and try the OC so I went along with it and my PC booted into Windows just fine and I'm now running IBT with 'Maximum' set as my 'Stress Level' and '20' set as the number of times to run. So far it's completed 8 passes successfully (it's also taken core 0 to 67C successfully -- ouch!).

I made the following alterations in bios:

• FSB to 333MHz
• FSB:DRAM divider to 5:6
• PCI-E Frequency locked at 100MHz
• VCore 1.3250V
• Termination Voltage (or something like that) locked at default
• DRAM Voltage locked at default
• DRAM Terminate locked at default
• Northbridge/Southbridge voltages locked at defaults

Anyway, I just wanted to say thanks for answering my questions.

 

No problem. I hope that it's stable

About your DRAM voltage, how high is it ? Not sure how high your RAM is rated, prolly 2.2v... There is sticker in RAM stick which tells how much is it, so if you are planning to tweak your ram settings, be sure that it's not downvolted.

67dec C is not bad, IBT gives temps no application in real world tests will give. Clean heatsink from dust and when you have found clock speed which you want to use, start lowering voltages little by little to lower temps and increase lifespan of that chip.

 

My current voltage is 1.8 -- the same as what it's rated to be run at and the same as it was running at before I OC'd. Do you think I should tweak my RAM to get tighter timings if my PC proves stable? I really don't want to risk damage to my hardware (hell, this 25% OC is bad enough!) so if the gains from tweaking RAM are marginal I'll probably just leave things as they are.

Hehe, I actually did that shortly after I discovered I had forgotten to plug in my HSF fan cable after re-seating my CPU cooler (thread).

PS: 12 passes successful so far.

 

Perhaps it's not even marginal increase from tweaking the ram, but it's hobby for some people and some people even think that it's funny, haha

I had my e6600 (2.4GHz at stock) running at 3GHz for quite some time with stock voltages. Then at some point I ended up increasing voltages and went to 3.6GHz (needed to play around with voltages for some time to get it stable). I was quite happy with that 50% OC. Last summer got my i7 920 (2.66GHz), I think that I installed Windows with stock clocks and then went to OC ended up to 4GHz quite fast. Was happy for some time, then 4,3GHz... Atm running at 4,4GHz with water cooling. Just saying that it's quite addictive hobby and not one of those cheapest ones.

 

Yeah, I hear you.. But then, I'm not really doing this because it's "fun" (although it is) but because I could use a small boost in performance. Also, I've had this Q6600 for 1.5 years now so I think it's about time I get all of my money's worth out of it.

EDIT: Sh*t! IBT failed the last frikkin' pass! Good thing I made it run 20 and not 19..



Any advice on how to proceed? I guess I need to raise the VCore one notch first, and then if it fails again, it's probably not the VCore but something else, right? Argh, what a terrible way to end the day!

 

It did almost pass so I dont think that it needs any huge increases anywhere. Try higher vcore, if it still not pass, increase vcore a bit more.

 

Hey, I just thought I'd write to say that I upped the voltage by exactly one increment (from 1.32500V to 1.33125V) and ran IBT with a full 100 passes eek successfully.. I figured that there would be no point in running only 20 since my system almost passed that even with stock voltage. I think that 100 passes with the maximum stress level should indicate a stable OC, but I will of course run a few Prime95 tests anyway, just to be sure.

Anyways, thanks for your help, Tat3.

EDIT: One question: My OC seemed to raise my Windows Experience Index from 7.1 to 7.3 for my CPU, but also for my RAM! Why did this happen even though I locked in my RAM timings, voltage and speed at their defaults? I even disabled turbo!



EDIT (2): Ran the Prime95 Blend test for 11 hours -- no errors. Also ran the small FFTs test, and some SuperPI tests, and no errors there either. So I guess I can keep this OC.

 

Sorry for late reply, did not notice that you edited your post.

That windows performance index is not very accurate, but I think this would be the answer for your question.

FSB : DRAM 2:3 -> 5:6

 

Hmm.. So does raising the FSB:DRAM provide more performance on its own? I'm confused now; I thought the ratio was only useful for adjusting the RAM speed to a specific, predetermined level with respect to an altered FSB. My ratio of 5:6 keeps the DRAM speed exactly the same as before the OC (also timings are the same). Surely, the only aspects affecting RAM performance are the speed and timings, and not the ratio between RAM speed and FSB..? Or could it be that the faster CPU speed drives data into the RAM faster and hence increases its performance as well?

 

You can try using different ratio and test those with superpi or some other application to see how there is difference.

If you can get it to 1:1 it will run on sunc and provide the best performance, because there is not delay between cpu and ram.
5:6 means smaler delay than with 2:3. (something)

But still

 

OK I see, thanks.

 

You know this guy stole your work and is claiming it's his also says he writes for guru.3D LMAO!

you'll have to fill in the blank in the link ****** without the spaces> e t e k n i x

http://www.*******.com/forum/cooling-overclocking/how-overclock-c2d-c2q-processors-dummies-1744/