A weird problem I am having with my overclock

My components are:

Asus Maximus Formula Mobo
E6750 processor
4gb (2 x 2gb) OCZ Reaper PC6400
Zalman 9700 CPU Fan
2 x HD3870's in crossfire
Stacker 832 case with 6 120mm fans

I basically followed the overclocking guide graysky wrote and am very happy with the results. I loosened the memory timing to 5-5-5-15 and brought the chip up to 450 fsb x 8 for a 3.6 total (very happy). I work as a software release manager so I ran many a load test after each incremental backup (17-24 hours each with prime95). I had set the dram voltage to 2.1v as per the instructions on the ram but kept the rest of the volts set to auto while overclocking. After I was stable at 3.6ghz I started fine tuning the volts

1st problem

I ended up OCing the RAM to 900mhz to run in a 1:1 ratio. I cannot seem to get the timings back to 4-4-4-15 now. Is this expected with the overclock? Does it have to run looser now because it is clocked higher?

Problem 2 (the weird part)

It appears that the mobo I have greatly overestimates it's voltages for some things. I looked through ASUS's tool and saw that it set the NB at 1.6v and it was running at aroun 54 degs. That was very high. The other things I saw that were really high was cpu pll volt (1.66v) and FSB Term (1.55v). I lowered the volts manually to:

vcore 1.36525v
cpu pll volt 1.5v
NB 1.41v
FSB term 1.4v

I stress tested these with prime95 the whole time and I can run a test for 24 hours without a crash.

The weird part

If I look at the progress of prime after 17 or so hours it looks like core 0 is running the test much faster than core 1. If the times are correct core 1 is about 2-3 hours behind core 0. What would cause this because it was not the case before I changed the NB, cpu pll volt and FSB term. My complete guess is the FSB term as it may be limiting the flow of data to the second core but I am not sure. Is this even a problem? Tests don't fail just surprised about the results.

Any help would be greatly appreciated

If it didn't fail I think you are just reading way too far into it.

It's probable something else was going on. Do you have an anti-virus or anything at all that is automated that runs at certain times?

ANY load that takes away from Prime95 will slow it down.

Even a 1% load on a single core taking away from Prime95 over time will cause them to not run at the same speed through the test. Especially after 17 hours.

In theory... probably. But I doubt it would cause enough to be noticeable.

LOL