Question about power consumption and overclocking

[three4seven]

I apologize in advance if this is a trivial question -

When overclocking, is the CPU power consumption correlated to the clock speed, the voltage, or both (or .. something else)?

For example -

Will a "good batch" i7 running 4ghz @ 1.25v draw less power than another i7 that's running at the same clock (4ghz) running @ 1.35v?
I'm guessing so.. how does one calculate the difference in power draw? Is there a significant difference in power draw?

Will bumping up the frequency to 4.2ghz but keeping the voltage the same equate to the same power draw as when the CPU was clocked at 4ghz?

Thanks in advance for any input.

 

[blivit4]

Power Dissipated = K x F x V^2

where
K = constant dependent on chip design
F = frequency
V = voltage

So, less voltage, less power; less speed, less power.

 

[extide]

^^ Note that the frequency increases heat output roughly linearly (twice the speed is twice the heat), while the heat output goes up by the square of the voltage (double voltage is four times the heat).

 

[drescherjm]

K = constant dependent on chip design

This is the one you can not change, however I am not convinced that this K is the same between two chips even the same stepping. I assume the difference in the flaws (remember all CPUs have flaws from the manufacturing process) will influence this.

 

[extide]

K should be CLOSE for all similar CPU's of the same stepping, but yeah, each one will probably be a different value.

 

[three4seven]

Thanks for all the replies and it's exactly what I've been looking for.

So if my feeble math skills are correct, assuming K and F were the same, the difference between running a processor at 1.2v and 1.3v results in roughly 14% less power draw?

very interesting indeed.

 

[Zero82z]

This is the one you can not change, however I am not convinced that this K is the same between two chips even the same stepping. I assume the difference in the flaws (remember all CPUs have flaws from the manufacturing process) will influence this.

K should be CLOSE for all similar CPU's of the same stepping, but yeah, each one will probably be a different value.

The best numbers we have to go on are the TDP values for different CPUs, so any difference in the capacitances of different CPUs don't really affect anything since we don't know their true power consumption to begin with.

So if my feeble math skills are correct, assuming K and F were the same, the difference between running a processor at 1.2v and 1.3v results in roughly 14% less power draw?

Yes. About 14.8% to be more precise.