scatter-mpi-multiplication.py

import time
from mpi4py import MPI
from functions import line_and_second_matrix_multiplication
from functions import get_matrix

comm = MPI.COMM_WORLD
my_rank = comm.Get_rank()
num_of_processes = comm.Get_size()

second_matrix = get_matrix(num_of_processes, 36.6)

if my_rank == 0:
    first_matrix = get_matrix(num_of_processes, 1.23)
    first_matrix_row = [x for x in first_matrix]

else:
    first_matrix_row = None

if my_rank == 0:
    time_start = time.time()

first_matrix_row = comm.scatter(first_matrix_row, root=0)

if my_rank != 0:
    line = [my_rank, line_and_second_matrix_multiplication(
        first_matrix_row, second_matrix)]
    req = comm.isend(line, dest=0)
    req.wait()
else:
    result_matrix = [[] for i in range(0, num_of_processes)]
    result_matrix[0] = line_and_second_matrix_multiplication(
        first_matrix_row, second_matrix)
    for procid in range(1, num_of_processes):
        req = comm.irecv(source=procid)
        line = req.wait()
        result_matrix[line[0]] = line[1]
    # print(result_matrix)

if my_rank == 0:
    print("Finished in %s seconds" % (time.time() - time_start))

MPI.Finalize