Compare map's performance of mpi4py module and Python's multiprocessing module

hostfile

localhost slots=9

julia.py

c = complex(0, 0.65)


def julia(x, y):
    z = complex(x, y)
    n = 255
    while abs(z) < 3 and n > 1:
        z = z**2 + c
        n -= 1
    return n


def julia_line(args):
    k, w, h = args
    x0, x1 = -2.0, +2.0
    y0, y1 = -1.5, +1.5
    dx = (x1 - x0) / w
    dy = (y1 - y0) / h

    line = bytearray(w)
    y = y1 - k * dy
    for j in range(w):
        x = x0 + j * dx
        line[j] = julia(x, y)
    return line

mp.py

# Executed via "python3.7 mp.py"

from multiprocessing import Pool
import sys
import time
from julia import julia_line

if __name__ == '__main__':

    for scale in range(1, 9):
        start_time = time.time()
        with Pool(8) as pool:
            w = 640*2*scale
            h = 480*2*scale
            image = pool.map(julia_line, ((i, w, h) for i in range(h)))
            with open(f'julia-mp-{scale}.pgm', 'wb') as f:
                f.write(b'P5 %d %d %d\n' % (w, h, 255))
                for line in image:
                    f.write(line)
        end_time = time.time()
        duration = end_time - start_time
        print(f"Scale {scale}, Time {duration:>6.6f}s")

mpi.py

# mpi4py version 3.0.2
# Executed via "mpiexec --hostfile hostfile -n 9 python3.7 -m mpi4py.futures mpi.py"

from mpi4py.futures import MPIPoolExecutor
import sys
import time
from julia import julia_line

if __name__ == '__main__':

    for scale in range(1, 9):
        start_time = time.time()
        with MPIPoolExecutor(8) as executor:
            w = 640*2*scale
            h = 480*2*scale
            image = executor.map(julia_line, ((i, w, h) for i in range(h)))
            with open(f'julia-mpi-{scale}.pgm', 'wb') as f:
                f.write(b'P5 %d %d %d\n' % (w, h, 255))
                for line in image:
                    f.write(line)
        end_time = time.time()
        duration = end_time - start_time
        print(f"Scale {scale}, Time {duration:>6.6f}s")