Hello Dmytro,
Thank you for investigation, the results are interesting.
I prepared simpler test [1], which has no recursion, it just
sorts 5 candidates and compare all elements with first (or second)
candidate. I run on array of 200 elements and result is:
e1/e2 267/460 on random data
e1/e2
Hi Vladimir,
There could be many reasons for this.
The verisimilar ones are imprecise time measurement with highly dispersed
results and biased samples.
Another reason is that an attempt to divide whole array into equal-size
partition not always gives us the best result.
And hence choosing "wr
