I mean running only one worker actor. Yes, it is interesting also to see the improvement respect to the real sequential time (for loop, no actors at all).

I’m measuring how well the actor based solution scale, using the approach that you suggested instead we could obtain a measure of the speedup related only to the problem/algorithm to solve (e.g. for evaluating if an actor based solution make sense at all, respect to a totally sequential computation).

I will do the experiment when I will have some free time, but I guess the real sequential time should be a little bit faster respect to the parallel solution with one worker actors.