Overclock.net banner
21 - 23 of 23 Posts
Discussion starter · #22 ·
Thanks for the bump! I'm currently able to pass LinX with "all" memory selected for 17 passes. It craps out on the 18th pass. I'm still at the same settings, only difference is tweaking GTLVREF, which has proven to be difficult. This setup seems to be more stable at a certain lane adjustment, but tweak it a mere 10mV in ANY direction, and it won't pass one single loop. It's very frustrating! Right now I have lane 0 @ +35mV, lane 1 @ +40mV, and lane 2/3 @ +15mV. Tweaking the first two lanes 10mV in either direction results in instability very quickly, but keep them at those settings and they will pass 17 loops. I still need to tweak the other two lanes a bit, and I just might be able to get it stable.

Thanks for all of the help, and please don't hesitate to add any more comments or suggestions. I need it! I WILL get this thing stable at 4.00GHz so help me lol! I'm so very very close!
 
Core 2 Quads are two dual-core dies, connected via the FSB. In my experience, the best OCs are achiveable with a bit more GTL going to the second die. Cores 0 and 2 shlould be die one, while cores 1 and 3 are die two.

The differential between the dies is as important as, if not more important than, the absolute GTL values.

One to three steps higher GTL on the second die worked best for my X3350.
 
21 - 23 of 23 Posts