Python 是否在每个循环中为变量创建/分配值比在循环外创建/分配变量需要更多的内存/时间？

Question

For example:

for i in range(5000):
    a = "some words"
    # Other things

vs

a = "some words"
for i in range(5000):
    # Other things

I'd imagine the second would be better, but how much better exactly is it? It might be more efficient to have some code outside a loop but it also feels easier to debug & read when the varaibles are assigned immediately before they are used.

Answer 1

Intuitively, in vanilla cPython in the absence of a just-in-time compiler that might hoist that assignment out of the loop, we would expect option2 to be faster.

See here for more information on " compiler hoisting ".

You can try it for yourself.

import timeit

foo = '''
for i in range(5_000):
    a = "some words"
'''

bar = '''
a = "some words"
for i in range(5_000):
    pass
'''

print(timeit.timeit(foo, number=10_000))
print(timeit.timeit(bar, number=10_000))

Gives me:

1.3565055000000001
0.9187005000000001

Note that this difference in speed is likely meaningless though compared to anything else your app is doing and I would recommend you go with whatever you feel is the easiest to understand and maintain.