Why is standard R median function so much slower than a simple C++ alternative?

Question

I made the following implementation of the median in C++ and and used it in R via Rcpp:

// [[Rcpp::export]]
double median2         


        

Answer 1

I'm not sure what "standard" implementation you would be referring to.

Anyway: If there were one, it would, being part of a standard library, certainly not be allowed to change the order of elements in the vector (as your implementation does), so it would definitely have to work on a copy.

Creating this copy would take time and CPU (and significant memory), which would affect the run time.