2013-12-23, 01:00 PM
Grey Wrote:You can skip calculating attacks per second and just find the increase in damage using the delays, since you're basically doing (1000/newDelay)/(1000/oldDelay) - 1 which simplifies to oldDelay/newDelay - 1.
It is necessary to put in some ceiling function for the oldDelay stuff because that expression is for an absolute increase in DPS, whereas a real (CPU-gentle) DPS increase really stems from the expression ceil{[BaseDelay * (10 + x) / 16] / 30} where x is the speed stage in question. Therefore if x is the old speed stage and x' is the new speed stage, the expression for DPS increase will thus be:
DPS increase = ceil{[BaseDelay * (10 + x) / 16] / 30} / ceil{[BaseDelay * (10 + x') / 16] / 30} - 1
Note that because the delay is the period of an attack, whereas what we usually do for DPS calculations is the frequency of the attack (which is 1/period), double-divisions make the old delay on top.
For an example of the calculations (with none of the expressions shown) see this post: http://www.southperry.net/showthread.php...ost1213378. I think it should be a ceiling function, not a round function...?
I'm not sure if speed stage can go negative... maybe?
Hadriel