Ok, so the Multiplier is locked on the P5B (Even on my Extreme)... I do want to use the multiplier when the bios is fixed but until then, I need to understand something...
Intel reports that their FSB at 1066.
What I have always called the FSB is reported as 266
266mhz X 11 = 2926mhz
So If I raise up the FSB to 300mhz, I drop the memory multiplier down, so that my memory runs slightly above 1ghz (I have Corsairs 1066mhz memory).
I have a feeling that the FSB intel reports is the communication speed between the Chip and the Northside bus. And that when I raise up the FSB speed (from 266->300), the the MOBO will report 1066 FSB increases substantialy. I assume that this will lead to the breakdown of communication between the Northside and the Intel Chip when understrain (such as in a game).
Edit: The higher the FSB, the more likely it IS going to fail... I have also learned that the p5b itself isnt the best choice for OC'ing at this time... This is due to BIOS reasons, and not the board itself. Expect things to change, for new BIOS versions are coming out every week...
------Continue to original post------
Is what I say accurate? Or is there anyother reasons for this irregularity.
What should I call the two different FSB numbers (the 266 and the 1066 number)?
What is the equation that relates these two numbers?
266x4=1064 and not 1066...
So lets assume its 266.49
266.49x4 = 1065.96
Which when rounding the numbers would make it
266 x 4 = 1066 (doh)
Things dont add up though... I can run the chip REALLY cool, run the memory REALLY fast, and have a unreliable machine. There must be a 3rd factor that is causing me to hang up, and why does this 1066 number keep increasing if I drop the memory ratio down?
I read and read about this stuff, and I just have yet to grasp the final aspect that pulls it all together...
Something tells me its the damn northbridge, with its cheap ass alluminum heatsink... but I dont quite understand why the FSB speed of 300 with a low memory multiplier would effect the frontside bus unless it was directly related to the operating levels...
I have done this, and at 300mhz the computer starts to fail on games...
I wonder though, if I drop the multiplier and try the same thing, running the CPU underclocked, and the Memory underclocked... if the machine will still crash in games. I could even toss up the 300 a bit more to make sure that if the issue is something else, that it will become apparent.
Well, what are you temps under full load? It is very strange to run 100% stable in all CPU intensive tests and fail when it becomes GPU intensive. What is the temp of your GPU during gaming? What drivers are you using for which card?
Set FSB to 500 with multiplier = 6
3000ghz (2 Mhz higher)
1000mhz on memory (underclocked)
2000mhz on the reported (usually 1066) fsb
This confirms my point that there is more to Overclocking this processor then
FSB X Multiplier = CPU speed
My problem is that I dont know for certain what this factor is, and that I would really like to know... Again I suspect that second 1066 number... I beleive that since this number is reported as 2000 it wouldnt post.
Now, what is the bottle neck, in this experiment.
Temps on Video Card are all under 50, except the voltage controller which runs as hot at 56.
Video Card is Slightly overclocked (30mghz on the core and 45 on the Memory) and runs fine at default CPU clock speeds for hours and hours (I played a game for 4 hours yesterday, no crashes)
But as I say when the FSB goes up, the machine becomes unstable. The reported vCore also raises. Even if I keep the same CPU speed and LOWER memory speeds, then the reliable 266x11=2.9
I am certain that when Asus gets there ass in line, and unlocks the x6800 then the machine will run fine at higher multipliers... But I want to understand why the machine crashes under these conditions. What is failing...
I know that 300fsb is unreliable at the default 11 multiplier
I tried a 9 multiplier with a FSB of around 333 and again memory running well below 1066. For my 20 minute game test it worked fine, which isnt supprising since it isnt a long time... but the vCore reported 1.36 which is the highest I have seen so far.
So here it goes
11x266=3.0 (1066mhz memory)
1066 Intel reported FSB
11x300=3300 (under 1066 memory, with tests under 900mhz)
1200 Intell Reported FSB
The Multipier is not soo locked on the X6800. Have you updated your Bios? I think its at 0309 if I am correct. I used to have the P5B with my E6600 and I could change the multiplier from 9-6 either way. So I would check their site.
NEWS FLASH - P5B 309 Does NOT recognize the X6800 as having an unlocked multiplier. It supports 6-11 multiplier. I am not happy but am in no rush to overclock... I want to learn more, I just dont know what I dont know... which is a problem because I dont know what to ask, just to show what I see... and hope someone comes up with a answer, even if its brief so that I may start my informational search anew.