@Akira1364
It is a very useful piece of code and indeed the C++ code is well documented, but many users, unlike you or me or avk, have no clue.
Nice to see that we have another option and for certain scenario's your code is impressive as shown from avk's benchmarks.
(which is, by itself, a well founded piece of code. It tries to circumvent bias, which is a hard thing to do with benchmarking)
What I still need to test is memory pressure.