ericeod

It is very possible. When I was testing, I was overcompensating failed Prime95 runs with more vcore. I would eventually get it to pass Prime 8 hours. But I later found that I still had instability (despite passing Prime95). Then I started to research quad cores with NB voltage and finally was able to get stable and bring the vcore down a little. at 438 FSB, I would definitely increase the NB voltage to +0.2v.

__________________
eVGA 680i Overclocking guide .......... ASUS P5E/Rampage Formula Thread.......... 4.4GHz CPU-Z

Blameless

The functional range Intel mentions has it's limit at 1.3625 VID (bios setting; less idle, even less load). You can probably go a little past this without trouble on the vast majority of setups, as very few people run their CPUs very near the tjmax, but only a few chips seem to survive constant 1.45v without sub-ambient cooling.

Having personally damaged a Wolfdale at 1.42v core in less than a week, 1.3625v is my limit for 45nm dual cores. My experiences with the 45nm quads have been different. They seem somewhat more durable.

As for chipset voltages, most of the time these can be increased a great deal beyond what Intel recommends. On good air cooling P35, X38 and X45 boards have been known to run 24/7 for months, with 1.65 vNB without damage. My DFI Blood Iron ran at 1.61 to 1.65 vNB almost constantly with no problems related to that (I did eventually blow a CPU PWM, but never had problems with the NB).

The Vector Kid

Okay, I'll try that next and see what happens, thanks for the advice!
I'm sure you guys have all seen Anandtech's thing on the dead qx9650, which is why I'm skeptical about touching the FSB voltage.

__________________

Blameless

Intel gives the same ranges of fsb (vtt) voltages as it does CPU voltages.

With my X3350 (aka Q9450) at 3.7GHz I only needed 1.25 vtt. The 24/7 safe limit is around 1.35v.

I'm not certain this is the case. I have seen plenty of reports of Quads going well past this, and I was able to get past 465MHz FSB with my Yorkfield on a lowly Blood Iron. The limiting factor for me seemed to be mobo to CPU power delivery and heat (which eventually fried, even though I had them sinked).

ericeod

There are a few boards that can get higher with a quad, but all the ones I've seen squeaking above 465 FSB where using 1.65v+ to the NB. There are however those few boards that have amazing FSB overclocking potential becasue of a nicely binned NB chipset. I had such a board, my NVIDIA 680i A1 board. While many struggled to attain a 450 FSB, I was able to run a 525 FSB on the 680i chipset using 1.45v to the SPP (with a dual core though):



I think microman is the most experienced quad core OCer here on OCN and his highest FSB OC was 467 with a Gigabyte X38 DQ4 motherboard.

As for The Vector Kid, the G(MCH) voltage is not the same as FSB voltage (VTT). That voltage is for the CPU-to-NB link. Anandtech killed theirs running above 1.4v. I think your bios has a FSB overvoltage control, which is the VTT(Its been a while since I messed with the DS3L bios).

__________________
eVGA 680i Overclocking guide .......... ASUS P5E/Rampage Formula Thread.......... 4.4GHz CPU-Z

vix

I think that a lot of rumors and opinions have muddied the waters on this issue. It all depends on ones interpretation of the processor specs. Key is paragraph 1 which I interpret as saying that even though 1.45 is listed as the absolute maximum, 1.45v is also outside of the functional limits and reliability CANNOT be expected at that voltage.

Vix

The Vector Kid

Right right, FSB and MCH are two different voltages. I don't have a problem playing with the MCH (NB) voltage, so as soon as this run passes I'll go back and boost it up. I just don't want to play with the FSB, I'm leaving it at +0.1. Thanks for the info though, I really wanted this thread to pool everyone's experience, and this is a great start!

__________________

Blameless

Even +.1v FSB(vtt) should still be enough for around 450MHz FSB on a 45nm quad, depending on the chip.

The Vector Kid

I don't want to count my chickens before they hatch... but it looks like ericeod is definitely on to something. By kicking the NB voltage up a notch, it looks as if I'm achieving higher OCs with less volts. This is great news for me, because I know my motherboard, and it can definitely take on the higher voltage. I'd rather boost it up than the vCore. Thanks eric, I'll add this advice to the OP.

__________________

Takendown2

i think the NB voltage is very important for OCing, before i need 1.4Vcore to keep my CPU stable at 3.2 now after i raised my NB to 1.4(1.25 stock) i only need 1.365 too keep it at 3.2. Also the conroes can last very long OCed, i brought mine when it came out and OCed it to 3.2 on the fly. also with the NB voltage at 1.4 i could hit 3.44 with 1.4Vcore and 3.52 with 1.45Vcore(8hours prime stable)

__________________
Sorry for my bad english I am not a native english speaker.