Well 2c is better than 0c.

Hehe yeah, 4.53Ghz(21x216) passed, about 66C load. 217x21 failed after 27 seconds, gave me a x124 BSOD.

http://valid.canardpc.com/show_oc.php?id=1653112
 

  Best yet?
I believe the rated FSB is directly related to QPI PLL once you go over 4Ghz stability becomes hard. So increase of QPI PLL may be the answer. Along with other tweaks of which will depend on your particular CPU. Speculation of course.
 
Edit:
  Now having no experience with the "930" I really would like to see you try with the default multi.
I have tried to figure out which voltage(s) apply to the multiplier but to no avail. Because I have seen where a chip can OC higher with a lower multi and higher QPI, but I simply do not accept the even odd deal. I really think there is a voltage(s) that affect the efficiency of the multiplier. And it does not necessarily mean a higher voltage simply the right combination of voltages and of course varies from chip to chip. Again all speculation of course. 

Mhm yeah that was a suicide run, highest I've eve had. Then again, it's the only time I've tried with the 930.
 
When I tried OCing with the x22 multi, I just couldn't pass 4.4Ghz. 22x200, it would just freeze every time.

same here halo, ive had 2 930s and both were the same easy to get stable with 22 multi @ 200bclk but not a inch further. 

Yeah, and I spent six hours trying to get 201x22, but nope. That wasn't happening lol.

can you get turbo to kick on @200???? My board never could but 23 multi should be good and stable.  That makes for an easy 4.6

You mean the C3?(Might be C6 or C1E)
 
When I had that turned on it bounced between 22x and 23x.

Hey Reggie, using the 23x multi right now, doing suicide runs for CPUz, what do you think about trying to use 23x for Vantage and Wprime?