threadingscaling.jl

function threadstest1{N}(A::Array{Float64},b::Vector{Float64},k::Vector{Array{Float64,N}})
  @inbounds for i in eachindex(A)
    A[i]+= sin(b[1]*k[1][i] + b[2]*k[2][i] + b[3]*k[3][i] + b[5]*k[5][i]) +
    exp(b[7]*k[7][i] + b[8]*k[8][i] + b[10]*k[10][i] + b[11]*k[11][i]) +
    erf(b[12]*k[12][i] + b[14]*k[14][i] + b[15]*k[15][i] + b[16]*k[16][i]) +
    (b[18]*k[18][i] + b[19]*k[19][i] + b[20]*k[20][i] + b[21]*k[21][i]) +
    (b[22]*k[22][i] + b[23]*k[23][i] + b[24]*k[24][i] + b[25]*k[25][i]) +
    (b[26]*k[26][i] + b[27]*k[27][i] + b[28]*k[28][i] + b[29]*k[29][i]) +
    (b[30]*k[30][i] + b[31]*k[31][i] + b[32]*k[32][i] + b[33]*k[33][i]) +
    (b[34]*k[34][i] + b[35]*k[35][i])
  end
  A
end
function threadstest2{N}(A::Array{Float64},b::Vector{Float64},k::Vector{Array{Float64,N}})
  @inbounds Threads.@threads for i in eachindex(A)
    A[i]+= sin(b[1]*k[1][i] + b[2]*k[2][i] + b[3]*k[3][i] + b[5]*k[5][i]) +
    exp(b[7]*k[7][i] + b[8]*k[8][i] + b[10]*k[10][i] + b[11]*k[11][i]) +
    erf(b[12]*k[12][i] + b[14]*k[14][i] + b[15]*k[15][i] + b[16]*k[16][i]) +
    (b[18]*k[18][i] + b[19]*k[19][i] + b[20]*k[20][i] + b[21]*k[21][i]) +
    (b[22]*k[22][i] + b[23]*k[23][i] + b[24]*k[24][i] + b[25]*k[25][i]) +
    (b[26]*k[26][i] + b[27]*k[27][i] + b[28]*k[28][i] + b[29]*k[29][i]) +
    (b[30]*k[30][i] + b[31]*k[31][i] + b[32]*k[32][i] + b[33]*k[33][i]) +
    (b[34]*k[34][i] + b[35]*k[35][i])
  end
  A
end
using BenchmarkTools
function benchmarkthreadtests()
  const ds = [10 100 1000 2000 5000]
  ts = Array{Any}(length(ds),2)
  for i in eachindex(ds)
    d = ds[i]
    A = rand(d,d); b = rand(35)
    k = [rand(d,d)]
    for j = 1:34
      push!(k,rand(d,d))
    end
    println("d = $d")
    println("Serial")
    ts[i,1] = @benchmark threadstest1($A,$b,$k)
    println("Threads")
    ts[i,2] = @benchmark threadstest2($A,$b,$k)
  end
  ts
end
ts = benchmarkthreadtests()
meantimes = [mean(x.times) for x in ts]
# Result:
#=
5×2 Array{Float64,2}:
 1.24656e5  1.69682e5
 1.63851e6  6.97997e6
 2.13978e8  6.87377e8
 7.20628e8  2.72495e9
 4.91433e9  1.68647e10
=#

This was with 4 threads.

julia> versioninfo()
Julia Version 0.5.0-rc0+99
Commit c10667f* (2016-08-01 07:45 UTC)
Platform Info:
  System: Linux (x86_64-redhat-linux)
  CPU: Intel(R) Xeon(R) CPU E5-2667 v4 @ 3.20GHz
  WORD_SIZE: 64
  BLAS: libopenblas (DYNAMIC_ARCH NO_AFFINITY Haswell)
  LAPACK: libopenblasp.so.0
  LIBM: libopenlibm
  LLVM: libLLVM-3.7.1 (ORCJIT, broadwell)