On Tuesday, 1 May 2018 at 18:46:20 UTC, H. S. Teoh wrote:
Well, yes. Of course the whole idea behind big O is asymptotic
behaviour, i.e., behaviour as n becomes arbitrarily large.
Unfortunately, as you point out below, this is not an accurate
depiction of the real world:
[snip]
The example I like to use is parallel computing. Sure, throwing 8
cores at a problem might be the most efficient with a huge amount
of data, but with a small array there's so much overhead that
it's way slower than a single processor algorithm.