[First Look] DFI 680i LT NF-680i LT SLI (56K Warning~~~)

[virtualrain]

I had a brief discussion today with Tony and Raju about quad-core overclocking on this board... here's the outcome...

I'm going to try a new approach. If you read back a page, my previous approach was to set high voltages and a reasonably aggressive FSB and try to tune GTL that way. Well as you can imagine, that's like throwing darts in the dark. I ended up with rather Bizarre GTL settings that seem reasonably stable but I'm going to try it a different way this time.

I'm going to start with some aggressive voltages again, but this time, I'm going to start at 1.4V VTT and suggested GTL values with FSB=300MHz and work up in 10MHz increments. Every time I increase it and it fails stability testing (more on that below), I'm going to tweak GTL2 upwards a bit... since the key is to get Core 2/3 "in sync" with Core 0/1. When I find stable settings, I'll add another 10MHz. This might take awhile.

If my wacky GTL settings are the outcome again, it really means my core dies are not speaking well with each other via the FSB and I likely have a less-than-ideal CPU. I tended to agree that a good part of the problem in overclocking Kentsfields are the dual-die's are both trying to use the same bus and so synchronizing levels is critical. It may take some finesse to get them both talking to an overclocked FSB... that's the way it is.

As for stability, I've been encouraged to use OCCT as it apparently is more reliable and quicker than PRIME. This will be a tough one to let go of!

That's about all for now. Back to the drawing board... I'll report back when I have something useful.