P7N Diamond: Q6600 to 4.05ghz!! But stability at 3.6ghz?

Question

Hi all,
I believe I've obtained the first successful non-extreme Kentsfield Quad bench above 4.0ghz on the P7N Diamond. It can be done.
The processor is a watercooled Q6600 G0 Batch L737B277 with a VID of 1.2375. The chipset is using the standard circupipe cooling, but I did buy an Antec Spot Cool to blow air over the NB. It seems to be doing a reasonable job.
Here are my OC settings and voltages:
Beta BIOS 1.1B4
Proc: 4.05ghz (450 x 9)
FSB: 1800 (450 x 4)
RAM: 900 (450 x 2)
RAM Timings: 5-5-5-15-2T (sub-timings on auto)
vCore: Max set in BIOS (haven't benched lower, but it may be possible.)
vDimm: 2.1
NB: 1.55
SB: 1.65
Gtl REF: 65
FSB Term: 35
all other voltages on auto.
Unfortunately, I'm still having a bit of trouble stabilizing at 3.6ghz. I've cornered it down to the same thing every time. The computer will eventually freeze (not blue screen or reboot, but freeze) at around 10 minutes of Prime 95 or between 2 and 5 minutes of OCCT (RAM stress.)
My current settings for 3.6ghz stable are as follows:
Beta BIOS 1.1B4
Proc: 3.6ghz (400 x 9)
FSB: 1600 (400 x 4)
RAM: 800 (400 x 2)
RAM Timings: 5-4-4-13-2T (default EPP timings)
vCore: 1.4375 (+2.000v set in BIOS)
vDimm: 2.1
NB: 1.4
SB: 1.55
Gtl REF: 65
FSB Term: 35
My guess is that the issue lies with the NB or SB, with RAM being the third option. I don't think it's the processor, based upon VID, batch, and the type of errors I'm getting (freezes, nothing else.)
I'm learning more about how NB and SB voltages effect the system, as well as how all of the reference, gtl, and terminator voltages interact (MSI's odd BIOS settings make this all the more difficult.) However, I'm still guessing at places, and without the ability to pinpoint a single voltage that's causing the freezing, it's becoming a bit tiresome. I've tried upping NB to 1.45 and SB 1.6 but can't seem to prevent the freezing. Perhaps its a combination of the two, or perhaps my FSB Terminator voltage is off. If anyone has any ideas on what could be causing this freezing, that would be great going forward.
Seeing that I can bench at 4.05ghz, I should be able to find stability at 3.6ghz. It's just a matter of time, and as always, further explanation of MSI's unique voltage settings always helps.
Thanks everyone!

Answer 1

Here's a brief overview of my 3.6 settings. Cpu is a G0 batch L737A702 with a VID of 1.2875, bios A7510NMS.116, C1E and EIST disabled.
The first screen in cell menu just shows mode to manual, fsb to 1600, and multi at 9. The pci-e freq=107 is just to give me a little help for oc my vid cards so disregard that setting.
Next you see my voltages. One might assume that cpu VID 1.2875v+.225v increase in bios will mean a vcore of 1.5125 but this isn't the case at all as you will see from H/W Monitor in a moment.
The H/W Monitor screen is showing me vcore 1.464v and while I'm sure part of this is vdrop, it also leads me to believe that bios may not be setting the initial voltage correctly or that there is some amount of vdroop before the OS is even loaded. Of course that's purely speculation without actual digital multi-meter measurements.
I can show prime 25.6 stability if need be, I have run it 6hr with similar settings on bios 114 but honestly I'm not really big on hours of stress testing. It cuts into my WCG time too much and I have taken kind of a personal interest in that since my cancer diagnosis 7 weeks ago. 4 cores crunching 24/7 without error and the occasional game of Witcher or Silent Hunter 4 is my idea of stable. You can see the CPU-Z Core Voltage 1.424v 100% load so vdroop is around .04v. You might also note ram timings are tight and CL1. The OCZ Reaper HPC's don't oc worth a damn but they run mighty tight stock. 
The temps are really something I'm not going to discuss too much. The common question seems to be "which app should I trust?". I'll leave this argument to people who know a lot more about intel thermal specs than I do. The author of Real Temp (unclewebb) said something in a post that stuck with me: If you are Prime stable and have left some head room before TjMax, which you have plenty of, then absolute temperatures don't matter one bit. 50C, 55C, 60C, 65C and even 70C doesn't matter while running Prime as long as you are stable. You are within the Intel spec and you aren't going to hurt anything. With no documentation of TjMax by Intel and no user testing of TjMax for your new processor, absolute temperature is no more than just an interesting number no matter what software you are using to report it.
A s-pi 8m run