Late-Night Overclocking

[Thrax]

So I was helping Shwaip do some investigating on OCing the e4xxx line of Core 2s, and I was doing some comparisons between the P965 and P35. I was specifically researching chipset strapping and the derivation of the NBCC. The NBCC of the P965 is calculated as such: (STOCK MULTI/SET MULTI)*FSB. The E6400, for example, has an 8x stock multiplier. I was running 6*500 for a while, which means (8/6) = 1.33*500 = 666MHz. That means while my external FSB clock was 500MHz, the northbridge itself was running at a whopping 666MHz. So what if I kicked the multiplier UPWARDS, which is completely counter-intuitive, to 7x? That's (8/7) = 1.14*500, or 570MHz NBCC. This means that I'm putting damn near 100MHz less strain on the internal clock of the northbridge which should, in effect, make EVERYTHING more stable despite defying reason. And hell, I thought, "Let's try it! WOOO MOAR SPEED!"

So I kicked the voltage up to 1.4125 to account for vdroop and gave it a shot:
http://icrontic.com/forum/attachment.php?attachmentid=23726&d=1186461486

//EDIT: This also means that if I had DDR-1200 and a chip that could do 4.2GHz, which I could probably hit with water or phase, 7*570 is not unreasonable.

[Leonardo]

Holy FSB, Batman! Nice job.

[Thelemech]

Even though I did not understand it all it was rather cool to watch part of the action (via IRC) !:bigggrin: That is some mighty overclocking!

[Thrax]

Today, I came into the possession of an E6420 for a few days. It's identical to the E6400 (266MHz stock FSB, 8/7/6 multipliers), but has 4MB of cache instead of 2MB. I'm going to lap it and give this chip a run through; if it can do 3.5GHz, it's a total keeper because the temperature is even between the two cores. That should give me some extra head room that I've not previously had due to the core temp discrepancy. The stepping is the same (B2), but the core stepping says SLA4T -- this is from April 2007 or later, instead of my old SL9S9 from June of 2006. Most people are reporting 3.5-3.7GHz on air, so I should be in for a wild ride. If it sucks, I've got my old 6400 which is no slouch. :)

More info tonight.

[lemonlime]

Very nice :)

[RyderOCZ]

Thrax,

The changing of the multiplier theory about how fast your Northbridge is running has been disproved AFAIK, unless I missed something.

[TheLostSwede]

Aye, high fsb at low multi/different strap is no longer an issue.