parallel_nheads.jl

using BenchmarkTools
using Distributed
using Base.Threads

addprocs(4)
nworkers() # should be 4

function f1()
    nheads = @distributed (+) for i = 1:200000000
        Int(rand(Bool))
    end
    nheads
end

function f2()
    nheads = 0
    for i = 1:200000000
        nheads += Int(rand(Bool))
    end
    nheads
end

function f3()
    wait_list = Task[]
    for i in 1:3
        push!(wait_list, Threads.@spawn f4())
    end
    nheads = f4()

    for i in 1:3
        nheads += Threads.fetch(wait_list[i])
    end

    nheads
end

function f4()
    nheads = 0
    for i in 1:50000000
        nheads += Int(rand(Bool))
    end
    nheads
end

@btime f1()
@btime f2()
@btime f3()

In my setup (4 nworkers, 4 nthreads)
f1: 370.550 ms (824 allocations: 33.41 KiB)
f2: 1.501 s (0 allocations: 0 bytes)
f3: 388.202 ms (30 allocations: 2.44 KiB)