[dannyboy]

How does overclocking make your computer unstable? Does the system become more unstable or is it unstable only because perhaps the overclocking made the computer run too hot?
Thanks

BTW - What is a safe speed to overclock an Indigo iBook. Is the method the same as with the older ibooks? http://www.bekkoame.ne.jp/~t-imai/ibooke1.html

[Cipher13]

Its the heat.
Indigo iBook? Whats that, 366?
I wouldn't go higher than 433. Thats pushing it as it is...
Laptops are not the perfect things for overclocking... be careful. They are small and have little room for an increase in heat.
Not sure about the plans being the same though... I assume so.

Cipher13

[Evinyatar]

It is not (just) the heat that makes it unstable. When you overclock you are increasing the frequence of a clock inside your computer that tells the processor when to start it's next instruction. If you increase this frequency the clock will wait not as long to tell the processor to start that instruction which occasionally can result in the clock telling to begin a new instruction when the first hasn't finished yet. This is very bad if for instance the second instruction should process the results of the first. It gets even worse when this happens if it is calculating memory addresses; the processor will simply write to an incorrect address resulting in crashes. Hell, even memory protection can't help you here.
Heat can also be a cause but to a lesser extend. The parts inside the processor are very tight together. Heat can make these parts expand and cause short circuits. We all know that's not good. It can fry your processor if you're not careful.

But just putting in an extra fan isn't going to work, since you cant eliminate the first problem without actually replacing the processor chip.

[Cipher13]

You're right, but usually the matter of instruction ordering getting messed up is more apparent when overclocking the bus than the processor clock, in my experience. That way its pushing more data to the slower proc rather than vice versa.
OC'ing the processor rarely runs into that problem - heat is the main concern...

Cipher13

[This message has been edited by Cipher13 (edited 01-26-2001).]

[BLAZE_MkIV]

The problems with overclocking have nothing to do with instruction ordering. That's what the bus multiplier is for. It tells the processor how many clock cycles it has for each bus cycle so that it checks the bus at the proper time.

There are several reasons you can't just keep increasing the clock speed.

The first is current. The higher the clock speed the more current the processor will draw. I don't know how many amps are drawn each clock cycle but if you increase the clock cycle without increasing the voltage supplied to the processor eventually their won't be enough current.

Second is heat. This is a little more straight forward than you think. The first heat issue is that the chip will melt. The second is that the electrical resistance on non-superconducting materials increases with temperature. That's how electrical thermometers work. Since this affects different materials at a different rate it can upset the balance and timing or the circuits in the processor.

Third is current it's self. As you increase the voltage and the frequency, even if you keep it cool, you run into a problem called electron migration. Instead of travelling along their nice little wires the electrons will pass into the silicon the circuit is laid out on. If memory serves the electrons will actually get stuck in the silicon and with enough of them they can generate enough of a charge to interfere with the circuit.

Fourth is the speed of an electron. The G4 has a 7 stage pipeline i believe where the P4 has a 12 or 14 stage pipeline. The steps for executing a command are broken doen into sections, aka stages, and they go from one stage to the next with each clock cycle. So when instuction x is in stage 2 instruction x+1 is in stage 1. No there's a whole lot of stuff about branch prediction which is a different topic altogether. The affect that this has is that the clock can't advance until the slowest of the stages is complete. This is limited by the speed an electron can travel throught that stages circuit. So the more work you do in the stages the fewer stages there are and the slower the clock spead. The P4 having twice the stages can operate at twice the frequency because they do half as much at each stage.

#include <stddsc>
I'm not a micro processor designer. What I know is from reading around. So I freely admit that some of them may be wrong. Corrections are welcome.

[This message has been edited by BLAZE_MkIV (edited 01-26-2001).]

[This message has been edited by BLAZE_MkIV (edited 01-26-2001).]

[This message has been edited by BLAZE_MkIV (edited 01-26-2001).]