What is your aim with this question?

If you want to use a different sorting algorithm because you think it would be more efficient for your particular data because it has some property as being already partially sorted, then, you should now that the "mergesort" used internally by perl already handles most common special cases. For instance, sorting and almost sorted array is an O(N) operation.

Sort::Key::Radix is a radixsort implementation, that can outperform perl internal mergesort for several key types.