I already demonstrated that it's not 50/50 and that prefixes are the best method. I have to remind you that we all, at some point, make mistakes, just like you and your closed code, which, by the way, I was the only one who didn’t throw you to the sharks.
I guess we're on our way to shit 10 pages of "why" after all. I thought the demonstration was over once the empirical tests converged to 50% to 50% operations.
This is exactly what I predicted will happen.
Lol
Ok, mister "LOL". Let me LOL a little too:
You are computing in an absurd way that last comparison.
Let's say that the win rate is 1 for sequential and 999 for prefix.
And we have 50 million total ops for sequential and 50 million total ops for prefix.
That doesn't make prefix 999 times faster. Because you are including the amount of ops no matter if the seq or prefix won or lost, which makes no sense whatsoever. You're dividing by the number of wins, but with a total for both scenarios.