We had another development meeting at the end of April where Stephen Dolan was invited to give a progress report on Multicore. Things have been progressing, although of course at a slower pace than anticipated (personally I’m not terribly surprised given the complexity of the whole thing, but that’s what you get for announcing more specific time periods :-). In terms of the original document we are still in the “build-up PRs” and “forward-compatible C API” phase, and things are moving along nicely.
One interesting recent development is the buildup of a comprehensive runtime-benchmarking tool (the not-terribly-easy-to-use interface is at http://ocamllabs.io/multicore/), which makes it possible to get concrete numbers on the performance overhead introduced by the runtime changes. The numbers are not final in any way yet (there is a lot of room for tuning), but it helps evaluate design choices and in fact I understand that the multicore authors have started exploring some alternative choices now that they have numbers to compare options concretely. (Takes time, but gives a stronger implementation overall.)
On the social front: the overall consensus that we want to merge the multicore runtime still stands, there is no worries to be had about that. I think it would be rather foolish to make a statement about this “happening by date XYZ” given the low likelihood of getting such a date right. On the other hand, if you want to help, please feel free to help reviewing any of the open PRs, and/or have a look at the Multicore Roadmap which has lists of tasks still to be done.