Fundamental rule of overclocking?

Hey all,
Can someone explain to me, the fundamental rule behind overclocking, i.e. why can some people overclock more than others with the same MB and processor. Now before you sigh and laugh , heres the situation.
My spec is as below, but I have seen other setups that have the 8INXP and, say a P4 2.4Ghz, overclocked to 3.0Ghz and up. That's an increase of at least 600mhz, yet I can't get any more that 350mhz without becoming unstable. Now I'm sure they have an SLK900 or summit like it, and I don't (stock), but I do have this which most of you have seen :http://forum.oc-forums.com/vb/showthread.php?s=&threadid=182950&highlight=drainpipe . I get my idle temps around 30*C and load temps around 48*C, which aren't much different than a setup with a fancy heatsink.
So whats the fundamental rule of overclocking? We all know that heat is the enemy, blah blah blah, but it doesn't seem to be the case that the cooler your CPU is, the faster you can get it to go.
So, what is the magic rule?
Thanks!

well, no two pieces of hardware are completely identical, thats one thing you need to bear in mind.
the basic rule would probably be keep it cool.
if your not cool, then your not stable.

There isn't one fundamental rule for overclocking, although the closest might be "All conditions are NOT the same for everyone, and all components(even made by the same manufacturer) are NOT equal in their ability to overclock"

Ram is a fairly big factor.

Adding voltage is another way to increase overclcocking as well.

As you said heat is the enemy

Luck has to do with it too.

I know that they're not all the same, but the ability to overclock over 600mhz, compared to 350mhz is a considerable difference.
And of course there is no one rule that governs overclocking entirely, its just that there must be something that dictates whether one CPU goes further than another. Bit of a weird question i've started here...

voodoomelon said:

I know that they're not all the same, but the ability to overclock over 600mhz, compared to 350mhz is a considerable difference.
And of course there is no one rule that governs overclocking entirely, its just that there must be something that dictates whether one CPU goes further than another. Bit of a weird question i've started here...

Click to expand...

The thing is that the chips may be the exactly the same, but the multiplier is locked to a different number. For instance, they may make the chips, and then lock the multiplier based on what is the most in demand. So you may buy a 2.4ghz, but it may be in reality 2.8ghz.

I have a feeling for instance that the new 800mhz fsb cpus are all identical, yet have different locked multipliers. Hence why they are all hitting the same fsb, but the multiplier is larger for the higher end chips, so they will obviously attain higher mhz. SO when buying a cpu, your basically paying for the multiplier

Total speculation though

Hmm, interesting theory, Intel are probably sending the heavies round to your house right now, for unearthing their dark secret
So all in all, is 350mhz a bad overclock? I can't help thinking I could get more out of it (I once got it to boot at 3.36Ghz, but I couldn't play anything 'cos of crashes.)

basically as i understand it, intel makes all CPUs 800mhz 3 gigs. Since every cpu from every batch is different, there are a signifigant number that cant make this 3 gigs. These are remarked at the lower levels. Of course, the fully capable ones are also remarked to meet with demand. So you may have one that didnt make the cut while some that are able to go higher were droped at that multi because of demand. Just speculation here.

Also, i know you have good cool supply of air to the HS but with a new HSF that is good quality, the heat will get to the air better. It doesnt matter how cold the air is if the HSF is not making adiquate contact to the core to get heat into the HSF. Copper is also a consideration here as it can get more heat away from the HSF. Try reinstalling the HSF and redoing the thermal paste job, and if useing stock white grese, upgrade to AS3.

I you might wnat to try this but then again you might have a bad apple CPU. Just giving you something to think aobut, you need to get the heat off the CPU befoe the cold are can cool it.

Edit: to give you a summary of the fundamental rule, think of it as getting the heat into the HSF and then into the air, rather than just cooling the HSF

What we're talking about here is speed binning, and all CPU manufacturers have used for as long as I can remember. CPU's are tested for their speed potential and, ideally, are binned to their performance level. Then the multipliers are locked and the procs are shipped to a grateful world.

Overclocking becomes possible because the manufacurers become pretty good at consistently making chips that perform in the midlevel of the range or higher. At the same time, the grateful world is actually filled with a bunch of cheap ****ards who wouldn't pay $500 for a CPU at gunpoint. When the two meet a lot of fast CPU's get binned down to a lower speed. Typically, the lowest speed of a processor range has the best potential for overclocking. (Duh.) But while the 2.4B, and the 1.8A before it, often overclocks to three gig and beyond there is always the chance that your 2.4 wound up in that bin because that's all it could pull.

To a lesser extent the same applies to motherboard chipsets and memory modules, and how everything fits together.




BHD

so when AMD leaves the multi unlocked your basically buying a higher end CPU at a lower cost if your lucky you can get a 3000+ for the price of a 2500+ but there is always the risk that the 2500+ you bought was actually only capable of a lil over 1.8 gig thus you dont get the OC that is expected. This is wear stepping comes in if i i am correct, (this whole binning concept is rather new to me) and some batches were really good but had to be binned down because of demand, thus those steppings will oc better. This is also why sometims a softmod 9500np works and sometimes it doesnt - sometimes the extra pipes are just shut off and sometimes they are broken, thus they go into a card that doenst need em. This isnt that usefull of a post but i thought i would put some more examples of binning out and maybee get a lil more insight into stepping (or am i hiting the nail on the head there)

is your power supply any good? maybe it doesn't have enought power to well, give it power

I have a feeling for instance that the new 800mhz fsb cpus are all identical, yet have different locked multipliers. Hence why they are all hitting the same fsb, but the multiplier is larger for the higher end chips, so they will obviously attain higher mhz. SO when buying a cpu, your basically paying for the multiplier

Click to expand...

Hmm, interesting theory, Intel are probably sending the heavies round to your house right now, for unearthing their dark secret

Click to expand...

It's no theory, that's exactly what happens. Although they test them to see if they will run stable under the speed they lock them at. They are all made on the exact same machines, but everything is so small, that things can happen, just like you can buy a car that has 200hp and the same exact engine in a car next to it can have 220hp.

There's no rule here at all, it is pure luck, as much luck as rolling dice. Back in the early days of the Celeron... man those were good little chips to buy because the bus speed was at full instead of half... anyway, I would buy 3 of them and see which one could run the fastest, then return the other 2.

give it a little more voltage. the people that are gettin the same temps as you are prolly dumping alot more heat into their big fancy expensive HSF. (more voltage means more heat)

and maybe your board, or ram is bad. i had a bad NIC that caused mad instability whenever i stepped over stock. check the system without individual parts, and if you dont find a suspect, you got either an unlucky chip or an unlucky board.

So, afetr reading all that stuff you guys posted (very interesting read), i've basically come to the conclusion that a better heatsink is required. I suppose i am pushing my luck a bit with just a stock heatsink fan.
I'm pretty sure that every thing in the comp is in good order, it cost me enough! The ps is excellant, a Antec 400w Smartpower, which gives excellant regular voltages.
Thanks for posting everyone!

Beware: I didn't read the replys.

It goes like this.

If the Athlon XP 1800's are made on the same die with the 1600's, then all 1600's should run at 1800 speed correct?
Nope. Even under this ideal condition where the slower processer is made side by side with the faster ones, nothing is 100%...
Which is why more than half the cores on the wafer became 1600+'s.(Last time I looked at the figures many months ago)

Now... We know every chip has it's own agenda. Take the example a post or two above about early celerons.
Some 300A's could hit 301, some 330, some 425 and some could hit damned near 600 off pielter cooling.

If all the other componets are the same (case/fans, powersupply/draw!, Hard drive tolerance!, memory tolerance) Then either:
A) it needs to be cooler
B) the chip is topped out man

There are actually three rules to overclocking:

1) Knowledge is power

2) Knowledge comes through experience

3) The man who tries the most parts wins

The key thing that you (and most others) are missing completely is the massive complexity involved here. Whatever limits you encounter occur for very real reasons, but they are often reasons that lie beneath a degree of complexity you can barely imagine. Everyone wants a simple answer (one word, please) and most times that answer does not exist. As you gain experience you learn to get more out of what you have, but overclocking is mainly the art of trying a lot of components until you find really willing ones.

What larva said. That's why communication is great. If you don't have experiance, someone else has done it and has input on it.