如何在python3中改进这个矩阵生成循环的运行时间？

import numpy as np

n = 1000
m = 1000
matrix = np.zeros((n, m))

from joblib import Parallel, delayed

n = 1000
m = 1000

def generate_row(i):
    row = []
    for j in range(m):
        row.append(i * j)
    return row

matrix = Parallel(n_jobs=-1)(delayed(generate_row)(i) for i in range(n))

n = 1000
m = 1000

matrix = [[i * j for j in range(m)] for i in range(n)]

import numpy as np
import numba as nb

@nb.njit
def generate_matrix(n, m):
    matrix = np.zeros((n, m))
    for i in range(n):
        for j in range(m):
            matrix[i, j] = i * j
    return matrix

n = 1000
m = 1000
matrix = generate_matrix(n, m)