Striker pwns EVGA with Q6600 G0

[Leeghoofd]

I always have set the value one or two lower than prescribed worked fab till now so try region 2.75 volts or such

[ZeroOne]

The northbridge and dividers on these boards seem flaky, at best, when oc’ing quads. Just thought I’d share my results while on a 4Ghz quest.

Dropped in a G0 Q6600 (to replace an E6600 that was at 3.6, 400x9 / 1066 5-5-5-15 - 3:4 divider - for a YEAR) and am currently battling the dividers and a mysterious heat issue on the striker with this quad. I’ve tried my own settings and also the Linflas settings. Bios 1303.

Here’s where I’m at; the Q6600 (L737A832, VID: 1.3250 packed 10/18/07) is very solid at 3.6ghz (450x8), 1.43750v set and verified with a multimeter at 1.439v idle and 1.441v load. (Did the vdroop pencil mod.) Can run a couple of hours of prime until I get bored, no errors. Will test longer, later. 3Dmark06 passes with no problem. Temps are better than expected with the cores acting in pairs. On water, 22C & 28C idle for the 2 pairs (0&1 and 2&3), and 42C & 48C load. 18C ambient. This thing runs cooler than the lapped E6600 dual core by a couple of degrees!

Here is the problem… sometimes, SuperPi locks the computer. Hard lock, no mouse movement… and the “pulse” of the hard drive light that these 680i’s seem to have, stops. Reset needed. It can run 32M sometimes, and other times, lock at 4M. Seemed pretty random at first, like there was definitely some fine-line that it was on. I’ve tried this with the memory at 1063/5-5-5-15 (11:13 divider) and underclocked ram to 900/5-5-5-15 (1:1 divider, both linked & synced, and unlinked) all at the 1800FSB. Keep in mind that 1066 is known-good/stock on these 2 dominator sticks (2x 1GB), proven for over a year, and the corsair fan is on them too. With so many constants (prime-stable cpu, known-stable memory at speed/timing/volts), the only issue seems to be the northbridge and the balance between voltages and heat building up somewhere. SuperPi is the only indicator of a problem.

If I could stabilize the SuperPi issue, I’d be at a rock solid OC, and be satisfied.

Testing SuperPi with ram underclocked at 900/5-5-5-15, NBvolts at 1.55v.
At 1.50v, SuperPi locked as did the following reboot. Upping NB to 1.55v got me back to windows and stable SuperPi, having just run a 32M and a bunch of 4M’s. After turning the fans all down to a respectable noise level and waiting 5 minutes @ idle… doing another 4M resulted in a lock after about ONE second. Dropped SB voltage to 1.55v (from 1.65v) to try to get rid of some heat buildup in heatpipes. Passed a couple of 4M’s at full fan speed. Turned down fans to low setting again, waited 5 minutes, and it locked after the initial value and first loop completed.

Testing SuperPi at the 1063 5-5-5-15 ram speed on the 11:13 divider and 1.55v NB.
I’m noticing the same issue. At first I thought it was the high ram speed combined with a quad that was overpowering the northbridge… but the same trend seems to be happening as with the 900 ram speed. Passed a 32M with no problems at full fan speed.

It’s funny that with the NB voltage the same as it was for the E6600, it locks NOW when hot with the quad but, with the dual core it never did, even with the fans turned off for a period of time and getting VERY hot. Case temp is the same, as the water removes any heat from the quad without dumping it into the tower. That, and no side panel and a huge 240mm fan! The heatsinks were NOT incredibly hot to the touch when the lockup occurs. They’ve definitely been hotter while experimenting with the E6600 and fan speeds. Maybe it’s some other component on the board overheating with a quad?

Just thought I’d share my experiences. Any suggestions?? I’m still a little uncertain about when to play with 1.2HT and CPU VTT. Would those change any of this? That’s about all I have left to mess with!

UPDATE:
Crashed in Oblivion with only like 25% of each core loaded. Rebooted and ran prime with the BLEND test to stress the memory / NB. Failed in like 3 seconds. Upped NB volts to 1.60, and it failed in about 15 seconds. Upped to 1.70, failed in about 30 seconds. Feeling like I had maxed my safe voltage, I backed down to 425. No boot. Its really hating a lot of the 1700FSB range. This is with the voltages that WORK at 1800FSB. You'd think that going DOWN a bit in speed would be MORE stable at a given voltage, not less!

2nd UPDATE:
Cranking these has bought me some time before crashing out in Primes blend test:
DDRII Controller Ref Voltage:
DDRII Channel A Ref Voltage:
DDRII Channel B Ref Voltage:

Started with +10 and got a little further in. 5min or so... and then it locked. Went to +20mv and it went a little longer. Then to +30mv (the max). Crashed in about the same amout of time as +20mv but messing with these still bought quite a bit of time. I feel like if i LOOK at the computer wrong, or sneeze, or something, it's all going to crash! I think this chipset is bad for my mental health! LOL

[chad]

zero just raise the 1.2ht to 1.7. and cpu to 1.5

[ZeroOne]

zero just raise the 1.2ht to 1.7. and cpu to 1.5

I've got them at 1.65v and 1.5v respectively.


Another update: (I hope this helps someone with a similar problem)
The DDR channel A, B, and controler reference voltages are listed in the bios as something like "ddr-vol/2" (and you can add +10, +20, and +30) and the more I raised them, the more stability I had but it topped at +30 and it still wasn't quite stable.

Even though my memory is KNOWN to work perfectly at 2.20v at the current speed, it is still part of the above equasion. Raising it to 2.2250 and leaving the +30 for the controller and a/b reference voltages, allowed even more juice to those components. (2.250/2 + 30mv). I'm now WAY more stable than before. Been running Spi, and the blend prime with no errors yet.

[my87csx481]

Does anyone have ~412 x 9 stable (with Q6600)? Please post your settings if you don't mind.

Thanks.

[Bloke]

Has anyone tried the official 1401 BIOS yet?

Does it does it work with the latest Intel procies?

[Don_Dan]

Has anyone tried the official 1401 BIOS yet?

Does it does it work with the latest Intel procies?

There is a thread about it, use the search function! ( sorry, that sound more rude than I wanted it to be... )

It seems that the multiplier is still stuck at 6.

Edit: Here you go: http://www.xtremesystems.org/forums/...d.php?t=172725