Multicore OCaml continuous benchmarking & call for benchmarks

I wrote a short post about the continuous benchmarking infrastructure we’ve deployed for Multicore OCaml + call for benchmark contributions from the community.


How do you write
Parmap.parmap, Parmap.pariter, Parmap.array_parmap, Parmap.array_pariter
in multicore ocaml?

And how to control the number of threads involved in the computation?
Let’s say, from 1 to 16.

PS: note that I am interested by parallelism, not concurrency.

I have a hunch that the full suite of Unicode normalization tests might make a good benchmark candidate.

We discussed a multicore branch exposing only low-level compiler primitives, without user-facing syntax for algebraic effects, to facilitate compatibility with OCaml code depending on the parsetree format. Did you drop the idea, or just haven’t had time to work on it yet? (I don’t see it in the roadmap.)

1 Like