I'm already here, and as i'm bored and laying in bed (and thanks to a rather numerour series of injections yesterday, unable to get out of bed) i'll give ya quick outline.
You have the FSB (Intel gives out FSB as "Quad Pumped" - ie your CPU has an FSB of 800, which in real terms is 200 - for a 1066 CPU it's 266 MHz etc)
You have a multiplier (This, times the real FSB = Final clock speed)
vCore (the voltage through the CPU)
Ram divider (RAM runs at the same speed as your FSB, however RAM tends to run faster - ie DDR800 RAM runs at 400 MHz, whereas your CPU's FSB runs at 200 MHz - a divider lets you run the RAM at a different speed to the CPU FSB and is usually a series of ratios - ie 1:2 ratio would let you run your CPU FSB at 200 MHz, and the RAM FSB at 400 MHz)
RAM timings (RAM has a set of timings - think of RAM as a train of information - the MHz is the speed between stations, and the timings are the speed the information gets on and off - on Intel MHz means more, on AMD timings mean a little more. Lower numbers are refered to as "Tighter" and are faster, higher timings are "Looser" and slower. They are expressed as a series of numbers - ie 4-4-4-12. The first number is the CAS# latency and the most important, the rest are less imporant. Some hardcore motherboards (DFi are the only ones that springs to mind) have a "second row" of timings that very very very few people understand, but are even less important than the main timings, but ever little bit of performance can be wrung out of the RAM)
RAM voltage (RAM needs more voltage to run higher quality, but you really wont ever need .2v over stock unless your running rediculous clocks or something like DDR1 BH5 sticks (which are famed for their rediculous clock speeds and incredible voltage scaling)
If that hasn't confused you, here goes...
Typically each component has a different limit, and you want to find each one individually - so you set the RAM on a big divider (so that it wont hold you back) and set really loose (Loose = High Timings - Lower Timings=Faster) timings. You then want to find the limit of the CPU - the most accepted way is to find the max FSB of the CPU/mobo by setting the multiplier as low as possible and see how high you can go - you should be able to get up to about 400 FSB. You then use the multiplier to try to get the highest clock, if it boots into windows and works for about 5 minutes chances are it's semi-stable. To check if it's actually stable try something like Prime95 and see if it passes. If something isn't stable, try adding more voltage, on the CPU i'd recommend up by 0.1-0.15 max because voltage is the number one way to heat a CPU up. Once you work out the CPU's max speed thats stable, you then want to get the RAM tied in - slowly up the divider until things become unstable, however the best test of instability is Memtest86.
The great thing about DDR800 is that as it runs at 400 MHz and even value sticks will overclock a bit (maybe 400-430 MHz) means that you don't need amazing RAM to take advantage of the extra speed.
That sounds complicated,i know, but it's simple when you get your head around it. I must confess i never stress test, and overclock in the most random way ever (namely i dont do any of the stuff i said above about dropping the RAM speed - i just find that i think of a clock speed and then try it in various different ways) but that i really should overclock to the way i named above to get the most outta my hardware.
