q6600 voltage question

[leo749]

Hi, i have a question regarding my cpu. recently bought a q6600. It has a G0 stepping. I read a lot of forums online which said the q6600 with a G0 stepping is a great overclocker and that it can go to 3Ghz on stock voltage. my problem is, i cant go above 2.7Ghz on stock voltage! Normally you would increase the Vcore a bit, right? But, i see all the forums online where they posted their overclock voltages(even the q6600 overclock guide here at tomshardware has it at 3Ghz at 1.2625V) are around 1.26V. So i checked my BIOS and cpu-z and the BIOS reported current CPU Voltage at 1.288(Actually it kept going back and forth between 1.288 and 1.296V). And cpu-z reports core Vid as 1.313V. All this on a FOXCONN P35a motherboard. My question is is it common or should i be worried( I am a bit worried right now as people report 3ghz at 1.2625 and i cant go beyond 2.7Ghz at 1.288). If its common, can i go ahead and increase the cpu voltage to around 1.34ish and try overclocking again.
Oh and i forgot to mention this! i did run at 2.8Ghz for some time. I changed the BIOS FSB setting to 1248(CPU fsb at 312) and ran prime 95 for abt 4hrs. I felt something was wrong rigt then because my cpu temp never went above 44C on any core at anytime and overall application loading and performance was not as fast as i felt before at 2.7Ghz( When i ram prime 95 for abt 4 hrs before at 2.7 Ghz, core 0 temp was flirting around 50C. and the system felt really fast than it was at 2.4Ghz.).So after i ran Prime95 for abt 4 hrs at 2.8Ghz. I stopped the test, shut down the comp and went out. Came back and switched it on and nothing would happen. the Comp did not post. So cleared the CMOS with jumper settings, loaded back at 2.4Ghz again. Currently running at 2.7Ghz.
So what should i do? Did i do anything wrong? Or is it just that my chip isnt so good at overclocking? Please help guys.

 

[night_wolf_in]

well, my VID is 1.3125(coretemp). and im running it in 1.216 (cupz). n im running it at 3Ghz.

for your case, there is possibility that your chip aint that good, although this possibility is very low. why dont you start with you VID and overclock it to 333FSB. and see if it works fine. if after priming/stressing, your cpu survive. then go and lower your voltage by one step. do the stressing again. and so on till you find the sweet spot.

as i said, try old school, n just start with VID and go lower every time testing it.

 

[MrCommunistGen]

More details could help. What CPU cooler are you using? It probably wouldn't hurt to know what power supply you have either. If you're hitting a wall at 2.7GHz on stock voltage I don't think you'd need a massive voltage spike to hit 3.0GHz. Wait for someone with more overclocking experience to help out before proceeding.

-mcg

 

[leo749]

Do you mean i should increase the cpu Vcore from 1.288 to around 1.32 and try overclocking to 333FSb or try 333FSB at stock voltage? thanks.

 

[leo749]

I use a coolermaster Hyper TX2 cpu cooler and a BFG tech 650 Watt Dual rail power supply with 20 A on each rail. Also mobo is Foxconn P35A and use 2 Gb of Wintec DDR2 800 ram. hope this helps. thanks

 

[cmashwin]

Thats not such a great cooler nor is it a crappy one.. But u shud be able to hit 3Ghz or even 3.2 with that...
I am currently running 3.4Ghz on my q6600 (378X9) @ 1.5v.. Its VID is a bit less @ 1.275.. My temps are under 60 under load (Prime95 for 2 hours).. I use the Thermal right ultra 120 extreme..
Now, coming back to you, if ur target is 3, then dial in 333X9 and up ur voltage to about 1.325 or a step higher and boot.. Also pls mention ur RAM details since it has a direct effect on ur overclocking..
I have heard of overclockers trying to overclock their value 667Mhz ram (ratios above 1:1) and fail and blamed their processors!!

 

[dagger]

Q6600 g0 is fine for up to 1.5v. I run 1.3125vid chips at 3.6ghz on 1.45v. It goes to 4ghz, but won't stay prime95 stable until 1.65v. 3.6ghz is the practical limit.

 

[leo749]

Thanks cmashwin and dagger. My RAm is Wintec DDR2 rated at 800Mhz. I run it at a ratio of 1:1.2( I know its wierd buy my Foxconn mobo allows onlu1:1.2, 1:1.5 and so on. no 1:1). So if i up the fsb to 333 the memory should run at (333*1.2)*2=799.2. Would that be fine taking into consideration it was rated at 800. I shall try upping the Vcore and let you guys know about the results. thanks.

 

[dagger]

Thanks cmashwin and dagger. My RAm is Wintec DDR2 rated at 800Mhz. I run it at a ratio of 1:1.2( I know its wierd buy my Foxconn mobo allows onlu1:1.2, 1:1.5 and so on. no 1:1). So if i up the fsb to 333 the memory should run at (333*1.2)*2=799.2. Would that be fine taking into consideration it was rated at 800. I shall try upping the Vcore and let you guys know about the results. thanks.

Underclocked ram should work fine.

 

[leo749]

I increased the CPU Voltage from 1.288 to 1.34 and increased the FSB from 300 to 333. It booted. Ran Prime95 for about 3-4 hours and the core temperature at full load was at 47-49C on core 0 and core 1 and around 38-39c for core 2 and core 3 constant. Does it mean its stable or should i run it for more time or is there any other application which i should run to make sure its stable? Na dwhy the temp difference between the cores? Does it mean that not all the cores are tested equal?

 

[dagger]

I increased the CPU Voltage from 1.288 to 1.34 and increased the FSB from 300 to 333. It booted. Ran Prime95 for about 3-4 hours and the core temperature at full load was at 47-49C on core 0 and core 1 and around 38-39c for core 2 and core 3 constant. Does it mean its stable or should i run it for more time or is there any other application which i should run to make sure its stable? Na dwhy the temp difference between the cores? Does it mean that not all the cores are tested equal?

That's not right. You have to stress all cores. Older versions of prime95 are single threaded, so you have to run more than one at the same time. Or get a newer version that run on all cores:
http://files.extremeoverclocking.com/file.php?f=103

 

[iluvgillgill]

you need to run it for at least 10 hours. i have seen many reported fail after 8 hours(original recommended length) so 10 or even 12 would be better.

 

[iluvgillgill]

prime 25.6 if what you need to stress the quadi

 

[shadowthor]

+1 for the prime95 25.6. I would aim for 12hrs+ to say its stable. But then I have seen one core crap out at 16hrs.

 

[iluvgillgill]

oh shadowthor you dont wanna tell him that!haha

i prefer 24hours to be honest. but its too long had to say.

 

[leo749]

Thanks guys, the version i had was 24.14. the cpu was always running at 25%. I should have known. Man, I'm so dumb. Anyway i shall install 25.6 now and shall run the stress test for at least 10 hrs and post the results.

Abt my actual question though. Why the partiality for cores 0 and 1 as opposed to cores 2 and 3?

 

[iluvgillgill]

well apparantly Core 0 and 1 is on one chip where as the Core 2 and 3 si on the other hence the name Double Cheese Burger. and just to let you know there will be 1-2C difference in all the cores. that is totally normal.

 

[leo749]

Ok guys.I've benn running the NEW Prime 95 test for about 9Hrs and 22 mins. and it didnt pop out any errors till this point. Does that mean its stable or should i run it for 16-24 hrs as iluvgillgill suggested. The core 0 and core 1 temp were around 59-60C throughout and the core 2 and core 3 were at 54-55c throughout during load. I sthat acceptable range or should i decrease the FSB? Also,i'm a little worried that the difference between cores0 and 1 and cores 2 and 3 is almost 5-6C higher instead of the suggested 1-2C. Does that mean i didnt seat the heatsink properly or is it normal? Thanks guys.

 

[Evilonigiri]

Ok guys.I've benn running the NEW Prime 95 test for about 9Hrs and 22 mins. and it didnt pop out any errors till this point. Does that mean its stable or should i run it for 16-24 hrs as iluvgillgill suggested. The core 0 and core 1 temp were around 59-60C throughout and the core 2 and core 3 were at 54-55c throughout during load. I sthat acceptable range or should i decrease the FSB? Also,i'm a little worried that the difference between cores0 and 1 and cores 2 and 3 is almost 5-6C higher instead of the suggested 1-2C. Does that mean i didnt seat the heatsink properly or is it normal? Thanks guys.

Is that small FFTs or blend? If this is small FFTs, 8hours is good enough for every day use, nothing intensive. If this is blend, continue to run it until it's at least 12hours stable, 24hours recommended.

The core temp difference is normal. Mine differs from 3-4C, and someone had an extreme case of 10-11C difference. I would say you are fine, but the lower difference, the better. Your temps are below 65C, so it's perfectly safe.

 

[leo749]

Its actually In-place Large FFt's (Maximum heat, power consumption and some RAM tested). Its the second option in stress testing. I selected it 'cause it said maximum heat. So, is 9Hrs and 22 Mins at that setting without any errors considered stable or should i need to run it longer? Thanks

 

[iluvgillgill]

dont believe that "max heat". small FFTs stress your CPU!!! the most which is the part you are OCing now. in my OC session i wil OC to a certin point then do SMALL FFTS 24 HOURS then after 24 HOURS BLEND TEST. if all pass then i OC further or settle down.

 

[dagger]

dont believe that "max heat". small FFTs stress your CPU!!! the most which is the part you are OCing now. in my OC session i wil OC to a certin point then do SMALL FFTS 24 HOURS then after 24 HOURS BLEND TEST. if all pass then i OC further or settle down.

Aren't you being a bit too strict? Well, I gues it can't hurt. 😛

 

[iluvgillgill]

YEAH I KNOW!!!LOL

i would let it off for like 12 hours before, but these days the product "quality" seems a bit weak, so it need to be strictly tested.

 

[shadowthor]

24hrs prime95 stable is definitely stable. It seems cores can crap out even at 16, 18, even 23 hrs.

 

[dagger]

24hrs prime95 stable is definitely stable. It seems cores can crap out even at 16, 18, even 23 hrs.

Meh, I doubt that core which crap out at 23 hour will have any effect on realistic usage.