多处理输出比单处理慢

Question

I'm running this code, and it seems running the script using multiprocesses is slower than running it via a single process.

Am I doing something wrong?

from time import time
numbers = [(1963309, 2265973), (2030677, 3814172),
           (1551645, 2229620), (2039045, 2020802)]
start = time()
results = list(map(gcd, numbers))
print(results)
end = time()


print('time is %.3f'%(end - start))

from multiprocessing import Pool


if __name__ == '__main__':
    start = time()
    with Pool(4) as p:
      print(p.map(gcd, numbers))
    end = time()
    #print('time is %.3f'%(end - start))
    print('Took %.3f seconds' % (end - start))

output is

[1, 1, 5, 1]   # single process
time is 0.444
[1, 1, 5, 1]   #multi-processes
Took 0.751 seconds

Thank you.

Answer 1

It's a bad example of multiprocessing advantages. It obvious will be slower because you spent more time on creating and starting each process than on executing your function. Multiprocessing sensible in case you have really big time consuming function so you put it in separate process.

Answer 2

Your timing is unfair. list(map(gcd, numbers)) is an in-process operation and doesn't start a new process.

The Pool class represents a pool of worker processes.

Starting a new process is intuitively slower than simply executing list(map(gcd, numbers)) , relative to list(map(...)) there's so much going on in Pool to start the processes. It's almost like comparing the speed of elephants and cheetahs.