Python multiprocessing.Pool: Difference between map, apply, map_async, apply_async

· Read in about 1 min · (161 Words)

multiprocessing.Pool  is cool to do parallel jobs in Python. But some tutorials only take Pool.map  for example, in which they used special cases of function accepting single argument.

There are four choices to mapping jobs to process. Here are the differences:

             Multi-args   Concurrence    Blocking     Ordered-results
map          no           yes            yes          yes
apply        yes          no             yes          no
map_async    no           yes            no           yes
apply_async  yes          yes            no           no

In Python 3, a new function starmap can accept multiple arguments.

Note that map and map_async are called for a list of jobs in one time, but apply and apply_async  can only called for one job. However, apply_async execute a job in background therefore in parallel. See examples:

# map
results = pool.map(worker, [1, 2, 3])

# apply
for x, y in [[1, 1], [2, 2]]:
    results.append(pool.apply(worker, (x, y)))

def collect_result(result):
    results.append(result)

# map_async
pool.map_async(worker, jobs, callback=collect_result)

# apply_async
for x, y in [[1, 1], [2, 2]]:
    pool.apply_async(worker, (x, y), callback=collect_result)

Reference