The basics

The theory behind calculating the power usage for an overclocked CPU is very simple: Power usage is proportional to clock speed, and proportional to the square of the core voltage.

The formula

The formula is also simple. Let's introduce the following variables:
  • Ps is the power usage of the non-overclocked CPU
  • Po is the power usage of the overclocked CPU
  • Fs is the clock speed of the non-overclocked CPU
  • Fo is the clock speed of the overclocked CPU
  • Us is the default voltage of the non-overclocked CPU
  • Uo is the voltage at which the overclocked CPU runs
  • Here is the formula:

    Po = Ps * (Fo/Fs) * (Uo2/Us2)
     

    Where can I find the "power usage of the non-overclocked CPU"?

    At Chris Hare's Processor Electrical Specifications page. Make sure you really use the data of your CPU model, and not of a newer model which could possibly consume less power.
     

    How about an example?

    Ok. We want to calculate the maximum power usage of a Athlon "Thunderbird" 1.33 GHz CPU  overclocked to 1.6GHz using 1,9V voltage. From Chris Hare's Processor Electrical Specifications page, we find out that:
     
  • Ps is 70W (max)
  • Us is 1.75V
  • Also, we know that
  • Fs is 1.33GHz
  • Fo is 1.6GHz
  • and Uo is 1.9V

  • Therefore:

     Po = 70 W * (1.6/1.33) * (1.92/1.752) = 99.26 W

    That's it!
     

    How accurate is this?

    By far accurate enough to estimate the power usage well enough to be able to decide which kind of cooling will be required. This formula does not take factors like FSB speed into account; they also affect power usage, but not to a significant extent.