That's right -- we first wanted to establish that existing OCaml code wouldn't be adversely impacted. There are various efforts ongoing to build interested (locked and lock-free) concurrent data structures, so that will inform the parallel performance results. Nothing published yet.