Overlocking Q6600 on XFX 680i SLI

[Mad Professor]

It still fails to boot into windows when running at 3.0Ghz

 

[Big.Wayne]

Hey Mad Professor,

I'm confused by your descriptions . . . when I read this:

9x FSB_1333=: Fails to boot

I assumed the above meant the PC wouldn't boot/POST after adjusting the settings? i.e Black screen/nothing?

Then you say this below?

It still fails to boot into windows when running at 3.0Ghz

Are you describing the same problem both times or are they different?

Take the CPU Core off from [Auto] and manually set it to 1.45vCore and try again, if that doesn't work boost the CPU Core to 1.500 and try again please!

 

[Big.Wayne]

Also in this image from your BIOS on memory timings it is set on [Expert] . . is this the BIOS default or did you manually change that?

 

[Rroff]

You WILL need a hardware vmod to get higher than 3gig stable. These boards can NOT provide sufficent amperage in stock configuration to get most Q6600 chips above 3gig stable (you might get lucky with a really good chip but rare).

You will also find there is a massive amount of FSB holes - anything between 403 and 451 is probably out, also a number around 280 and 340 but 333 should work ok (might need to use 334).

 

[Mad Professor]

Both 1.45v and 1.50v gets as far windows welcome screen and the BSOD.

As for the "memory timing settings" that is default or Expert.
Set to default just grays out all the boxes, so it can't be adjusted.

 

[Big.Wayne]

Hello Mad Professor

As for the "memory timing settings" that is default or Expert.
Set to default just grays out all the boxes, so it can't be adjusted.

When you say "grays out all the boxes" which boxes do you mean? . . . the whole BIOS or just the memory timings? . . . It's a longshot but I would leave that on default for the moment . . .

Both 1.45v and 1.50v gets as far windows welcome screen and the BSOD.

Something is odd here?

8x FSB_1500=: Boot fine, 48-50c @ idle.

9x, 1422 FSB, Boots into OS.

Ok I didn't think the vCore boost would do it . . . at least I hoped so because a q6600 shouldn't need much over 1.35vCore @ 3.0GHz . . .

Can you try lowering that BIOS memory setting so it reads 666MHz DDR and try again please!

 

[Mad Professor]

Setting the memory to 666Mhz still gets as far as the windows welcome screen then BSOD.

 

[Mad Professor]

As said before I can seem to be able to run the processor at 2.4, 2.6, or 3.2, but that seems to be about it.

Q6600 @ 2.4Ghz

Q6600 @ 2.6Ghz

Q6600 @ 3.2Ghz

 

[Big.Wayne]

Hey Mad Professor

Q6600 @ 3.2Ghz

that looks good . . . I've never used an nVidia LGA775 mobo before but something strange is going on if the system BSOD's @ 3.0GHz (9x333) during Windows load but boots fine all the way to desktop @ 3.2GHz (9x355)

How is stability and temps at 3.2GHz?

Prime 95

 

[Mad Professor]

I have not yet done any kind of stress testing at 3.2Ghz yet.

I do have both "Prime 95" and "Intel Burn Test" on my computer

I 1st just wanted to make sure if I am overclocking the right way, etc.

 

[Diggsy]

You might need to loosen the memory timings. I would suggest you set it to 'SPD', or maybe change it to 5-5-5-18.

 

[Harpss1ngh]

I've had similar issues with my old 680i and a q6600 in the past, they're not very good for overclocking quads. I think this may be the same issue in your case.

 

[Mad Professor]

I have started stress testing the system at 3.2Ghz, I will 1st be using the Intel Burn Test program.

With the core voltage set to auto, the system will boot and is stable at idle and light loads, but will BSOD within 10 cycles of Intel Burn Test.

I have then manualy adjusted the core voltage, And my motherboard does have a fair old vdrop on the core voltage.

-----------------------------------------
1.425 = 1.384v, (0.041 vdrop) @ Idle.
1.425 = 1.336v, (0.089 vdrop) @ 100% Load.
------------------------------------------
1.450 = 1.408v, (0.042 vdrop) @ Idle.
1.450 = 1.360v, (0.090 vdrop) @ 100% Load.
------------------------------------------
1.475 = 1.432v, (0.043 vdrop) @ Idle.
1.475 = 1.392v, (0.083 vdrop) @ 100% Load.
------------------------------------------
1.500 = 1.464v, (0.036 vdrop) @ Idle.
1.500 = 1.408v, (0.092 vdrop) @ 100% Load.
------------------------------------------

With the bios core voltage set to, 1.425v, 1.450v, or 1.475v, I would also get a BSOD withing 10 cycles of running Intel Burn Test.

Running the the bios core voltage set to 1.500v, seems to be stable at this point in time, and runs for way more cycles, but I have to abort the test as the cpu core temps go pass intel's 71c max limit.

 

[CraigN]

i used to have an EVGA 680i A1 revision which i was upgraded too as they original version was no good for quads and also had some chips on the rear of the board which meant you could not fit a retention bracket for a waterblock or large cooler.

However i had my B3 stepping QX6700 at 3.7ghz on it. 3.6-3.7 was its top, i ran a QX6800 on it at 4ghz. Do you have an original revision board ? I also found Crucial Balistix 1066 ram better than OCZ reaper 1066 ram.

I see your ram is running at 800 is there a reason for this by that i mean your not using 1066 ram ?

As your using a locked multiplier CPU you might need to give more juice to your chipset i cannot remember what voltages i used for mine tho im affraid.

http://service.futuremark.com/compare?3dm06=3669527

was 3dm 06 run at 3.6ghz

edi: from that link i can see i was running an FSB of 400 so a QPI of 1600 with a multiplier of 9

 

[Big.Wayne]

Hello Mad Professor

I have started stress testing the system at 3.2Ghz, I will 1st be using the Intel Burn Test program.

Personally I would have started stressing the clock using Prime95 "Small FFTs" instead of IBT/LinX . . . the latter are both good diagnostics but really are worst-case-scenario as they heat up the processor waaay more than many real-world programs could dream of . . .

The advantage of using Prime95 first is you can work out the vCore the processor needs to perform calculations without error and does not factor "Heat" into the equation . . . by using IBT/LinX first it is hard to tell if your errors are vCore (or lack of) or heat related . . .

I'm still curious why the system wouldn't boot at 3.0GHz (9x333) . . . if you had a 333MHz-FSB chip (1333MHz System-Bus) like the newer 45nm ones you would expect that to work fine at stock . . .

This FSB hole business is not something that affects LGA775 boards while operating under 400MHz-FSB and certainly not while operating bang on an official intel FSB such as 266MHz/333MHz/400MHz etc . .

when you set the FSB to 333MHz-FSB (9x333) the board must be goofing or something is not as it seems? . . . the problem is as the system can't get to windows we can't nose at CPU-z and see whats wrong . . . as a suggestion I would download MemTest and burn a bootable CD-ROM . . . you can then load this instead of hammering your O/S and this will also reveal to you some basic diagnostic info such as processor speed and memory frequency that may be different to what you are seeing in BIOS . . .

Lastly if you don't have a backup of your O/S stored safely on another partition/disk be careful you don't bOrked your O/S install . . . enough dodgy overclocks and blue-screens like what you were getting here can corrupt Windows . . .

I'm interested to see what progress you make . . . Memtest should help!

 

[Mad Professor]

I have been running the computer at 3.2Ghz all day, just doing day to day stuff, no problems at all, I have also played a game rounds of Counter Strike, did some work in SolidWorks, also no problems.

I did do a 2hour stress test with the intel burn test, but I had to ziptie my 250mm fan to the side of the open case to move more air around the system, the core temps maxed out at 72c.

While the system was running Intel Burn Test I had a feel of the Akasa AK-CCX-4002HP Venom CPU Cooler, I could touch any part of the cooling fins and it was just warm to the touch, the top of the heat pipes was also warm, but I can't reach the bottom of the Akasa AK-CCX-4002HP Venom CPU Cooler, or the bottom heat pipes.

I have not lapped the cpu or cooler, do you think it might be worth giving this a try to knock a few c off the temps?

As for windows, I am not worried about that, all my data is backed up.

Is it very strange that I can not get the system to boot into windows running at 3.0Ghz

I don't have any blank cd's or dvd's to burn a copy of MemTest, but I will look into seeing if I can get it to boot from a USB Memory Stick.

 

[Mad Professor]

Went onto MemTest site, and they had a ready build USB installer.

Ran MemTest, but it can not detect my chipset.

I let MemTest run a full pass of all the tests, no errors found, this was with my computer overclocked to 3.0Ghz.

I am not sure what info you wanted me to collect.

 

[Mad Professor]

I was running the lastest bios P33 (691N9P33), but after reading on google, that pepole seem to be able to overclock better using the P31 (691N9P31) bios.

So last night I downgraded the bios to P31.

And now the computer will boot into windows at 3Ghz (333x9.0), and this is with all the voltages set to auto.

I have not yet done any long stress tests, only 10 loops of IntelBurnTest.

 

[ShakeNBake]

I have a nForce 750i board and a Q6600, and I have to turn up the voltages of the HB/NB/SB as well as vCore to get a stable overclock.

My Q6600 is a bit knackered now after running at 3.2GHz for 2 years or so.

Trying to squeeze 3.0GHz or 2.7GHz out of it now xD
Seem to be running ok at 9x333 today...