40% comes from testing the formulas with micro timers, varying sizes of arrays, running a calculate (say 100 times(per time period)), measuring the time and averaging across multiple attempts with the same formulas. (and the 40% 'maybe' unique to my machine, my installation, the way I hold my...