ram timings are obvious.
disable Dynamic overclocking and performance stuff (so only ur settings are applied)
set pci to 67 (very imporatant or else you don't have pci lock)
fsb is ovious aswell
Vdim is ram voltage and is straghtforward
Vcore is another thing, you've got reference vcore up to 1.55v and then the percentages stuff. Lock reference vcore to say 1.55 and the percantageges let you increase from there on (IF you want to overvolt of course which can damage your cpu)
as for the rest i do(t know, I myself am figuring out things (for instance i don't kjnow what a relatively safe vcore is for the A64)
actually higher multipliers are always better if you can get stable fsb. of course if you have to lower your fsb below the default then it would be pretty useless. also you have to consider the ram. a high multipler and lower fsb could outperform if the ram is in 1T say, or titgher timings in general.
but really, i wish overclockers would do more real world tests, im kind of tired of PI, 3dmark, and sandra scores. they mean little in the real world, why nto in overclocking too? id liek to see mroe overclockers aiming for more ral world testing, say encoding or gaming.
Since when is a higher multiplier/lower FSB faster than a lower multiplier High FSB? Seems to go against evrything I've ever heard. If so, must be an FX thingy, because otherwise Higher FSB = Higher performance. My Mobile Barton is faster at 12*217 than at 13*200, yet both are 2.6Ghz.
Edit: Ah, on die memory controller, higher multiplier = more bandwidth. I need to read up more on this.
ABIT IS7, P4 2.6C, 1GB Corsair XMS 4000 Pro Series, Radeon 9800 Pro, Santa Cruz, TruePower 430watt<P ID="edit"><FONT SIZE=-1><EM>Edited by Pauldh on 11/08/04 11:08 PM.</EM></FONT></P>
yes trooper the 1T makes a huge difference. For real world tests... I think sandra is ok to measure plain bandwidth and cpu capacity but only as an indication. Sandra is done in 10seconds encoding takes an hour or longer (and I'm not about to encode the same movie twice) gaming however... I can't really give an average but my CS source gets an added 10 fps or so and the spikes in framerates went up from 70 to 100+