Originally Posted by MasterKromm
I'm not about to pretend I completely understand what is being said there. I have been giving it some thought though, and I do understand what you and Munchers are saying... I'm just not sure if I am willing to accept it(yet) as inarguably true. Assuming it is truly a FSB wall, and assuming VCCA is constant for all e4300 chips then they should all wall out at very similair FSBs. Unfortunately simply "forcing" the multiplier down may not behave as desired, since its not the original locked multiplier such a comparison isn't really effective. As it stands now I am willing to admit there is definately a CPU FSB wall at some point for all chips. Though I feel that the e4300 is atm limited by overall quality(week/stepping/maybe even where on wafer it was taken) and the NB straps for people who have boards that don't allow them to choose.
**EDIT** oh yeah and good night all I'm beat **EDIT**
Ya you're right the 4300 is limited in some ways. Seems like a very inconsistent chip first of all, more so than some other series anyway, as far as heat goes. Also, it runs at that 800mhz strap, which is limiting to cpu and mobo, and thus overall OC, because of the tighter timings. Basically, either mobo or cpu, one of them will crap out because of the strap, sooner than it would on a 1066 strap cpu. The big "but" here, is that will not always hold true, because no two cpu's are created equally. You have some 4300's having higher walls than others, some 6600's having different walls than others, etc. With all the variances, it becomes a very unexact science, and things get kind of confusing or blurred results ya know?
Which goes to the other thing you said about VCCA, even if it is constant on all 4300's, that doesn't mean they will all react the same, cause again every cpu is different. Sometimes vastly different, even among the same weeks/batches/steppings. Just like the max core clock can vary alot, so can the wall it seems, even among identical spec chips, so voltage may be constant but chip performance from that voltage, is not. It makes sense that the fsb wall would vary similarly to the core clock max, from cpu to cpu. The 4300 on average
, seems to have a lower fsb wall than most 6xxx series chips, from what I've seen. True also you say, certain multis don't behave sometimes, but when I see multi's seem to work for large range, but all bomb out at around same fsb on one cpu, just seems fishy you know
That quote from XS you put up is saying basically what I mean, there's more than just the core under that hood, and everything has it's limits. He's saying what we all know, in detail, and that is: fsb
X cpu multi
= core clock
. The first part of the equation fsb, is controlled by you of course to a point. If mobo can handle it, you likewise try and impose it on cpu, if it too can handle it.
The PLL you could actually call it the multi itself since it pretty much is, and so it's the second part of the equation. It's the receiver of the fsb, and the outputter of the core clock, the middle (multi) man of the deal. Taking the imposed fsb, multiplying it, then outputting the now higher speed core clock. Then third equation part is your core clock. Since all parts of this equation are physical entities, speeds being driven by voltage, and not made up numbers, laws of physics kicks in and every part must have a limit, it must. The PLL is another part of the cpu just like the core, given a voltage and expected to maintain a speed from it, attempting to hold up it's part of this fraqile, larger equation, so doesn't it make sense there'd be a wall really?
--I find that last post in your XS link very interesting, about P35 mobos. Perhaps some mobo's either are programmed to, or just give the higher volts where
the cpu needs it (like PLL supply voltage), or just maybe cleaner voltage or something? Maybe that's why results vary by mobo like muncher saw, further confusing the puzzle. Yet another thing that changes from user to user, affecting the overall result, and ultimately our take on things.