i vote sticky so people can see how much asus sucks and we will all boycott their motherboards
Their boards don't suck (at least Intel), their customer service does...AND I don' think this is even a problem with the board perse.
@ Strat: CRAP! I typed a whole post, previewed, saw your post. I thought it was a P5WD2 also...haha. Thanks for pointing it out!
@ Cordial: I did many mem tests on my board (you'll see my SPi @ #4 right in between all those AMD FXs
) I will say that some dividers just SUCK on it. 2:3 ratio in particular was horrible. At some points (closer to max OCs), it would not even boot with 2:3 divider, but would with 1:2 (and even faster mem speed), so I know it was just the divider causing problems.
The first thing I would do is eliminate the mem as the issue and run 1:1 to see what you can OC too. Like I said, I don't know much on the 945 chipset and it just may not be as good for OCing as the 955.
1) You say you had problems around 250FSB. 4GHz is no guarantee on a 640 on air even though most will do it...that could be the hold up. What kind of CPU temps are you seeing?
2) If you are trying to OC at 14x using EIST, I would say don't...at least to start. Whatever is done to lock the proc @ 14x does not seem to always work well in terms of OCing...you might end up with higher FSB, but it will probably max at a lower final clock.
I had a 630 for about a week...it OCed much better and higher FSB at the default 15x than it did on 14x using EIST. Granted, the default multi was closer to 14x than yours, but If you're going for max clock, I would see what it reaches with the 16x for clock speed first and then you'll know about where you limit is clock-wise on the CPU. Then play with 14x and see what you can get, but I'll bet anything you won't be able to boot anywhere near as high a clock as on the default multi.
3) As kick arse as my Mushkin is, it did not work for crap in my old P5AD2-E (I still had it when I bought the mem). I even contacted Mushkin the same day because they wouldn't even run the spec speed in it. I mean it wouldn't do DDR700 in that board with the voltage cranked. I still don't know if it was the chipset (925XE) or what, but it definitely wasn't happy in that board. So it may very well be that the chipset and that mem don't get along to well. If I hadn't experienced it myself, I'd say it was impossible, but I guess mobo manufacturers test mem modules for a reason. So, don't rule out the mem itself as a possible problem in that board. I might try pulling one of the sticks and see if that helps at all.
4) As Strat pointed out, the MCH and DDR voltages will help with higher FSB and mem freq/tighter timings
Like I said, I'd leave the mem @ 1:1 and work on the CPU OC first and see where that becomes an issue. The proc/temps could be the hold up, maybe the 945 chipset just isn't as amenable to higher OCs as the 955 or maybe that particular mem just isn't happy with that chipset. What kind of OCs are others with 945s getting?