parallelization using OpenMP

[ekg]

Would it be straightforward to make the sort parallel, such as using OpenMP? I'm mostly asking if there is a gotcha that I'm missing which would prevent it.

[ekg]

I have done this in my branch, and am interested if it could be pulled in: #14

As you know, a parallel first pass is not possible. An in-place fully parallel MSD radix sort algorithm does exist. It's called PARADIS, and its authors have never provided their implementation to other researchers. It would appear that a reimplementation of this idea would be required to test it.