there is an error in theperformance per watt computations..
I hope it isn't in the PTS itself :
- You calculate the performance per watt by dividing the power consumption by the time it took to run. While I agree that the time should be on the denominator, the power should be too ! More power used doesn't make for more performance PER watt (here is the hint...)
Your formula : powerUsed / time = XX Watts/seconds = POWER times performance... has nothing to do with performance PER watts.
CORREST FORMULA : 1[or any number, I chose 10000 for lisibility] /( time * powerUsed) = XX seconds-1.watts-1 (performance is in seconds-1)
By the way I'm not just saying that because you made a mistake, it changes the conclusion (which was very odd) :
no, overclocking doesn't improve (drastically !!!) the performance per watt, it reduces it, like it has always done.
Overclocking is using a CPU out of it's optimal conditions, so of course efficiency decreases, as performance increases.