I wrote some test code at one point to run the method in that link (I never got particularly far with it as more urgent work came up), but at least for small phylogenies, it seemed perfectly feasible speed wise. But if you’re wanting to do any cophylogenetic analysis (such as the stuff in Rafferty and Ives or Hadfield and Nakagawa), the slowdown was too extreme for it to be practical (at least the way I coded it, which was just a modification of Diogo Melo’s animal model with the solution in that thread added).