Help with QX6700 overclocking

[Bun-Bun]

Ok I have been fighting with this rig for the past 3 weeks and I just can not figure it out on my own.

I have the following:

QX6700 cooled by D-Tek Fuzion (cpu only loop GTX240 RAD)
Asus Striker Extreme
Corsair PC2-8888C4DF (sent back), OCZ PC2-8500 SLI, OCZ PC2-8000 VX2 (OTW)
EVGA 8800GTS
X-Fi Elite Pro
Plextor PX-755SA
2 x 320GB Seagates in Raid 0
Seasonic M12-700 PSU

Ok, before I got the Dominators and the water cooling I had some PC2-6400C5 in here and I had no problems. Then I got the Dominators and all hell broke loose... after corrupting my bios and telling me my CPU was getting 4v... I sent the dominators back and am getting another EVGA 8800GTS and some OCZ Titanium Alphas as there replacement. Until then I have some OCZ PC2-8500 SLI in. If I enable SLI EPP I am not stable... and by not stable I mean I get video artifacting and BSOD's. Turned off SLI EPP and it has seemed to be fine since then. But anyway... onto the overclocking.

Here is the last screenshot I took of my overclock:

http://pics.livejournal.com/bun_bun/pic/000et9xz

Core temp is running in the systray. It went 22 hours 30 minutes like that. Then I loaded media player and listened to some music over the network... got through 3 songs and got BSOD. Weird...

So I remember reading somewhere about Quad cores not taking high mem speeds on high FSB... so I clocked the mem down to 800 from 1050. BSOD takes longer to come but still comes but Super Pi gets a "NOT CONVERGENT IN SQR" error and orthos will get the random error before the BSOD hits. Dropped the mem down to 750... same thing. Tried changing CPU voltages... same thing. Changed HT, VTT, and NB voltages... same thing. I actually have them lower now cause it doesnt seem to affect anything. I beleive I have them set to 1.35, 1.4, and 1.35 for HT, NB, and VTT respectively.

I am going to try again at stock settings and then I am going to try again with the PC2-6400C5 back in... but anyone have any suggestions? To me it seems that the CPU is stable but the mem is not. And I don't think the RAM is bad just that I am missing something or 680i really doesn't like this revision of D9 or something...

[skylinepower]

you can try 4Ghz :p:
what cooling do you use?

[Bun-Bun]

you can try 4Ghz :p:
what cooling do you use?

I posted it in the first post but I am cooling with water. D-Tek Fuzion WB in a CPU only loop with a GTX240 rad and a DCC+ pump. my water to air delta is between 1 and 4°C depending on my ambient. (My ambient fluctuations from 18-28°C... I hate my apartment)

[skylinepower]

sry but i cant see what your temp is :p:
but if its under 50 on load so is it good, try to go so high you get to the temp is 50 load ;)

[Bun-Bun]

sry but i cant see what your temp is :p:
but if its under 50 on load so is it good, try to go so high you get to the temp is 50 load ;)

Fully loaded with dual orthos small FFT @ 3.5GHz the cores are 63,63,60,60°C. My goal was to keep it under 70°C because I had it to 3.2GHz on air and that was over 70°C dual orthos stable. But if I can keep it below 70°C then that means i will never go aboe 60°C because nothing comes close to dual ortho small FFT temps. Even if I fold on all four cores and play games for hours it doesn't come withint 10°C of dual orthos.

It went 6 hours on stock settings so now I am going to try the PC2-6400C5 RAM from my media center.

[Solarfall]

try another bios!!!! some times that may have a huge difference to ocing.

[Bun-Bun]

try another bios!!!! some times that may have a huge difference to ocing.

hmm... ok. I will try Bios 1002 (im using 1004) after im done testing with my PC2-6400C5.

EDIT: Went to go download bios 1002 and there is a new bios released two days ago, 1102 so I am going to try that.

[Solarfall]

hmm... ok. I will try Bios 1002 (im using 1004) after im done testing with my PC2-6400C5.

EDIT: Went to go download bios 1002 and there is a new bios released two days ago, 1102 so I am going to try that.

good luck on that new bios, lets hope that you can oc that chip to even higher. :banana:

[Bun-Bun]

good luck on that new bios, lets hope that you can oc that chip to even higher. :banana:

yes lets hope so... my goal with this system is 3.7 GHz on the processor and my old goal was 1200 on the RAM but... I have sinced abandoned that and will settle for either 1000-1100 @ 4-4-4-12 or something low like 800 MHz with 3-3-3-8-2T or even some combination of 1T.

Ok stock works fine, threw the PC2-6400C5 back in and its even more picky. I dont get BSOD's it just likes to freeze...

But anywho the results are a little more predictible with this ram so I have slowly raising the FSB with them in for now. I had it at 1200FSB just fine and dandy with my previous settings with 1.35, 1.4, 1.35 for HT, NB, and VTT. But at 1300FSB it would just freeze half way through super pi 32M so I tried raising the voltages... raising NV and VTT to 1.4 or 1.45 just caused the system to not boot. At 1.5 volt however it booted once again and now is stable in windows.

http://pics.livejournal.com/bun_bun/pic/000ewz0e

Amazing how my temps don't fluctuate much from 3.0 GHz to 3.5 GHz at the same voltage...

So yeah thats an hour of dual orthos and 4 runs of super pi 1M as where with the lower voltages I could not make it through 1/2 a run of super pi 1M... ok I am stable here.

Next is trying to lower the HT voltage as I feel it is unnessarily high... and then onto 1350FSB.

EDIT: Set HT to 1.35v and it now reads 1.42v in PCProve. System stability was unaffected. I am now going to save these settings to a floppy and try 1350.

[Bun-Bun]

Ok, got it set to 1366 on the FSB now. System seems fine, does a few runs of SuperPi 32M and then it starts to get "not exact round" error and then will eventually BSOD with a STOP PFN_LIST_CORRUPT from what I can tell that indicates still memory problems.

And since I can run Dual Orthos with small FFT's by themselves... I am thinking I just need to keep tweaking the NB, and FSB voltages...

What are safe maxiums for 680i HT, NB, SB, and VTT voltages? Before I start raising them any more I raised the CPU voltage so that it is now reading 1.46 in PCProbe and am stressing it again. I am already at 1.58 for VTT and 1.55 for the NB... 1.55v I thought was kinda the norm but it looks like I might have to raise it above 1.6v and I am a little concerned about pushing it that far with the heatpipe cooler.

What it seems like is happening is every component by itself is fine but when I start to stress the entire system as a whole things go out of whack. Like just super PI is fine... or just dual orthos is fine or just video is fine. But when I do dual orthos, SPI 32M, listen to music over the network, and run rthdrible with ati tool scanning for artifacts... things quickly go downhill... at least it is not as bad as when I had the dominators in here though.

EDIT: Well I couldn't get 1366 FSB stable so I gave up and set it back to the way I had it at 1300 and left it for the night... I woke up to a black screen and restarted the computer and it said a previous POST attempt had failed... so it had tried to restart and failed... blah. So I checked the logs and it didn't even make it through my run of SPI 32M... why would it have been stable earlier but now totally not? That and why wouldn't it be stable at 1300 FSB? I thought 1300 FSB was an easy OC... help? lol

EDIT2: Raising the HT helped... went 2 hours 55 minutes but still. I don't know I give up. Maybe I just have a bunk chip? I am going to put it back to stock and throw hell at it for 36 hours or something... dual orthos, SPI32M, rthdrible, ATI tool... maybe I am stressing it too much? maybe nothing will ever make all of that stable?

[Bun-Bun]

Ok after doing a lot of reading and watching my system run stable on stock settings while puposefully causing it to heat up ...

I have found out that I am not the only one. Early revision EVGA boards were plagued with a wall at 300 FSB... I think Striker is the same way. I see people around here doing 350-380 FSB easy and some people even refering ot specific revisions...

What I need to find out is what revisions are good and where is the revision number on the board? I have a RMA coming back on a Striker and maybe I can get them to hand pick a known good revision if I can figure it out before the ship it out...