When people say "conventional" sorting algorithms, they're usually talking about sorting algorithms based around pairwise comparison functions.

I note on slide 14 of this presentation, it looks like this is sort of the discriminator for selecting a better partitioning scheme. So it looks to me like this actually leverages a similar principle?

As we've seen in this thread and others, there are some other ways to measure and/or process that have different characteristics. Surely all of these deserve attention! So, thanks very much for sharing this.

Sorry, I was being imprecise. By conventional I meant comparison-based sorting functions that are polymorphic in the element type and thus not allowed to examine individual bytes.

Multikey Quicksort indeed looks like a special case of discrimination, exploiting some of the same principles.