Q6600 Overclocking - Stuck at x8 Multiplier

[veblen]

I have the following components:

MSI Neo2-FR (v1.8 BIOS)
Q6600 G0 Stepping
Xigmatek S-1283 CPU Heatsink/Fan
8Gb RAM (2x2GB G-Skill rated at DDR2-1000 and 2x2GB Corsair XMS at DDR2-800)
Corsair VX 550W PSU

and am currently running at 8x400 (1:1 divider) with the following voltages:

vcore 1.3875v
VTT 1.35v
NB 1.40v
RAM 2.10v (stock for G-Skill, a tad over for Corsair but no problems thus far)

My chip has a rather high VID (1.3V) so my vcore is not as low as I'd like; nonetheless temps are fine (I hope!) thanks to the S-1283, it idles at 40 and never exceeds 60 at load.

Here is my problem: I seem to be able to overclock only at the 8x multiplier. I've tried pushing the FSB up to 450x8 for 3.6ghz - am able to post but not boot vista x64. I went as far as increasing vcore to 1.45v but when it still failed to boot, I decided not to take the risk.

The highest stable overclock I could get is 417x8 at 1.4 vcore and with ram frequency at 1000MHz (4:5 divider).

I would like to make use my chip's 9x multiplier but I can't seem to get my system stable at 400x9 no matter how much voltage I pump or how relaxed my ram timings are. I've tried lower FSBs, as low as 360x9 but the system is still unstable.

Is there something I am not doing right here? I just think it's odd that I cannot overclock with the 9x multiplier when I keep reading so many posts on various forums about Q6600 overclocking.

Thanks for any help.

 

[Thor86]

Overclocking is luck of the draw. You may just have a limited CPU, but before you give up, get back with some motherboard BIOS settings, especially voltages.

 

[veblen]

Are there additional voltages I should post besides those above? They are:

vcore 1.3875v
vtt 1.35v
nb 1.40v
mem 2.10v
sb 1.8

I am running at 400x8 at 1:1 currently with mem timings at 4-4-4-12.

 

[Gillbot]

I had the same mobo running a Q6600 G0 SLACR and the highest I could get it was 3.4GHz. Watch your Vdroop, mine was horrible with the Q6600 in the socket. I believe I also had the best luck with the 1.9 Beta bios on that board. If you google for Neo2-FR beta bios, you can find it floating around out there. Also, that board never did clock well for me with all 4 ram slots filled. You may want to try using only 2 sticks of ram and see if that improves your OC.

 

[taltamir]

Originally posted by: Gillbot
I had the same mobo running a Q6600 G0 SLACR and the highest I could get it was 3.4GHz. Watch your Vdroop, mine was horrible with the Q6600 in the socket. I believe I also had the best luck with the 1.9 Beta bios on that board. If you google for Neo2-FR beta bios, you can find it floating around out there. Also, that board never did clock well for me with all 4 ram slots filled. You may want to try using only 2 sticks of ram and see if that improves your OC.

this is very similar to my experience. I have massive vdroop with the Q6600 G0 SLACR (0.15+v) and I had practically none (0.02v) with the E8400 wolfdale.

I did not dare go over 1.5v in mobo to get the 3.4ghz stable (much lower after vdroop).
When 3.3ghz was getting into 1.475v I stopped trying higher voltage on it as well. And just jumped down to 3ghz. at which point I was able to slightly underclock my CPU. underclock + overclock from 2.4 to 3.0ghz seemed a mighty enough feat. And the ram and northbridge were scalding hot at 3.3 but are merely hot (and not unbearably so) at 3.0ghz cpu / 800mhz ram.

 

[veblen]

Thanks, gillbot, I'll give 2 sticks of RAM a shot and see if that changes things.

I had forgotten about vdroop but that's true, I did get quite a bit of vdroop; perhaps it will be safe to pump my vcore up to 1.5v to see if that improves my oc.

I just wish I can make use of the 9x multiplier; it's just odd that I can oc only with the multiplier at 8x. Is that normal for oc-ing?

 

[taltamir]

if you did not change your ram ratio / multiplier then you are ocing the heck out of the ram. try lowering it so that the ram ends at below 800mhz, then find a stable CPU OC, it might give you further insight into the capabilities of your CPU.

If you did not change the mem from default multiplier, then it is now at 400 x 3 = 1200mhz... that is one heck of a ram oc.

 

[Thor86]

The GSkill memory are rated for 5-5-5 timings, that could also be your problem with boot ups. Also, try boosting your nb to 1.6. As the others have stated, it could be the limit of your motherboard.

Originally posted by: veblen
Are there additional voltages I should post besides those above? They are:

vcore 1.3875v
vtt 1.35v
nb 1.40v
mem 2.10v
sb 1.8

I am running at 400x8 at 1:1 currently with mem timings at 4-4-4-12.

 

[Gillbot]

When I had my Q6600 at 3.4, I didn't even touch the NB volts. I don't think I moved VTT much either.

 

[taltamir]

same...
dont over volt your southbridge, dont over volt your north bridge, dont overvolt your vtt, and try not overvolting and over clocking your mem so much. Focus on CPU only.

 

[Scotteq]

Another vote for UNDERclocking your RAM until you get your proccy stable. Then go back.

 

[Markfw]

Originally posted by: Scotteq
Another vote for UNDERclocking your RAM until you get your proccy stable. Then go back.

Most ram needs 2.1 from what I have seen nowadays, PC-6400 and better that is.

 

[veblen]

Thanks for all the suggestions! Here is what I've managed to do so far:

I've successfully set all my voltages to stock (i.e. nb at 1.25v, vtt at auto, sb at 1.5v), increasing only vcore to 1.3875v to get my oc stable at 400x8. RAM is running at 1:1 (DDR2-800) and mem voltage is at the recommended 2.1v for the G-skill 2x2gb DDR2-1000.

Oh, and my vdroop is 0.02v (1.3875v in BIOS, 1.360v in cpu-z 1.46).

I've taken out the 2x2gb corsair xms and tried 400x9 at the same voltages.

Again, my system posts but blue screens on booting vista x64 with a hardware error. I'm going to start adding voltage incrementally and see if I can finally get my desired 400x9.

Just one question: a higher multiplier allows me to hit a higher oc at lower FSB; aside from putting load on the motherboard if I use the 8x multiplier (instead of the 9x) and a higher FSB, it really shouldn't impact performance much right? It would help me to get over the fact that my chip, for some reason, does not like the 9x multiplier and I should just live with it.



Will report back!

 

[Gillbot]

Load up your system with OCCT and check your droop again. It will get MUCH worse under full load.

 

[taltamir]

@ 400x9 my Q6600 would not run stable on 1.5v, I had to lower the OC, then again I have much bigger vdroop on it, 0.15 almost. (I had 0.02 with the E8400 on the same board).
Good luck.

 

[COPOHawk]

Couple of comments.

1. The Corsair XMS Ram should be at 2.1 (per Corsair's website)...even though it says 1.8. I have put together about 8 boxes in the last 1.5 years using the same Corsair RAM and they were all more stable with RAM at 2.1 volts.

2. Why are you dead set at gettin 3.6 ghz? Just for the heck of it? I got my Q6000 (G0) to boot at 3.6 ghz...then realized that there wasn't much of a difference between that and 3.3 ghz in real practical use...so I set it back down to 3.3 ghz and that is where it has stayed for months now...with running F@H 24/7 (mostly). I can understand the accomplishment of the 3.6 ghz overclock for bragging rights...

3. My best overclocking seems to happen when I keep the RAM unlinked to the CPU...or a low RAM to CPU multiplier. I haven't worked with the MSI mobos though.

 

[veblen]

Thanks for all the comments! And you're right gillbot - stress testing with OCCT resulted in a pretty horrific vdroop before I blue-screened.

Thanks also for your suggestions COPOHawk.

1. I do have my Corsair XMS RAM at 2.1v since that's my G-Skill's recommended voltage so I unwittingly have it set to the correct one for the Corsair.
2. Good point - it is partly for the fun of it just to see how far I can push my chip but on a more practical level, I agree with you - the only difference is in benchmark scores. I'm not so dead set on getting 3.6 ghz as being able to oc to 3.2 ghz using the 9x multipler with a lower FSB. I just cannot seem to oc much on the 9x multiplier relative to the 8x.
3. I've been running my ram at 4:5 (400FSB, DDR2-1000) for a while at 3.2ghz but upon reading a post here a week or so ago, decided to lower it to 1:1 and see if there's any difference. There really isn't much according to Everest so I decided to tighten timings and run at 1:1 instead. One good thing though - I found that my Corsair XMS can run at DDR2-1000!

In any case, after spending more than I should in the BIOS, I think my chip is destined to run at the 8x multi. Until my next upgrade, I should be content! Still, it's great to learn a few things from the veterans here; I now have all my voltages at stock (cutting down my temps) save my mem (2.1v) and cpu (1.375v).

 

[taltamir]

i think part of your problem is that you just up the multiplier without lowering the fsb... the reason you find it easier to OC with 8x is that each mhz of fsb increase produces less of an OC.

If you just up the multi from 8 to 9 on 400mhz you just increased your OC from 3.2ghz to 3.6ghz... a significant change considering the current chips tend to cap somewhere around 3.3 or 3.4ghz (without loadline calibration / unsafe voltages)

A multiplier increase simply increases the speed of CPU operation. And FSB increase increases the speed of the CPU, as well as the northbridge, as well as the physical etched circuit on the mobo connecting the CPU to the northbridge. (well, increasing the CPU alone via multi still increases all of those, just not as much).

So grab a calculator and start doing 100 or even 50mhz increases.
You are at 3.2ghz? aim for 3.3 as your next step., 3300 (target speed in mhz) / 9 (multiplier) = 366.6 (target fsb).
So lower your fsb to 367 and up your multi to 9. Rather then just upping the multi and leaving the FSB at 400.

Overclocking via multiplier is vastly EASIER then overclocking via FSB increase. It just seems harder to you since you are doing it wrong.
Unless you are telling me that you tried 3.3ghz @ 8multi x 412fsb and 3.3ghz @ 9 multi x 366 fsb and that the 8 x 412 was stable at a lower voltage... In which case you really have a case of the fsb overclocking better then the multiplier, but I seriously doubt that is the case as it goes against everything known about overclocking. If that is the case, please let us know.