This reminds me of this classic StackOverflow question:
Why is it faster to process a sorted array than an unsorted array?
https://stackoverflow.com/questions/11227809/why-is-it-faste...
Why is it faster to process a sorted array than an unsorted array?
https://stackoverflow.com/questions/11227809/why-is-it-faste...
Edit: After running with a wider range of parameters, it seems that the difference is always roughly the same order of magnitude. To investigate further, I included the sort into the second timing to double check and for 3276800 elements it's still a bit faster overall when you sort the array.
#!/usr/bin/env python
import time
import numpy as np
def main(n=32768):
arr = np.random.randint(0, 256, n)
t0 = time.time()
sum1 = do_loop(arr)
t1 = time.time()
arr = np.sort(arr)
t2 = time.time()
sum2 = do_loop(arr)
t3 = time.time()
assert sum1 == sum2
print(" Unsorted execution time: {} seconds".format(t1-t0))
print(" Sorted execution time: {} seconds".format(t3-t2))
def run_many(func):
def wrapper(arg):
for t in range(1000):
func(arg)
return func(arg)
return wrapper
@run_many
def do_loop(arr):
tot = 0
for i in arr:
if i >= 128:
tot += i
return tot
if __name__ == '__main__':
main() import random
...
arr = [random.randint(0, 256) for x in range(n)]
...
arr = sorted(arr)
$ time python2.7 hackernews_13682929.py
Unsorted execution time: 4.33348608017 seconds
Sorted execution time: 4.09405398369 seconds
$ time python3.5 hackernews_13682929.py
Unsorted execution time: 4.4200146198272705 seconds
Sorted execution time: 4.188237905502319 seconds
$ time python2.7 hackernews_13682929_purepython.py
Unsorted execution time: 0.981621026993 seconds
Sorted execution time: 0.832424879074 seconds
$ time python3.5 hackernews_13682929_purepython.py
Unsorted execution time: 1.3005650043487549 seconds
Sorted execution time: 1.157465934753418 seconds
$ time pypy hackernews_13682929_purepython.py
Unsorted execution time: 0.239459037781 seconds
Sorted execution time: 0.0910339355469 seconds
Unsorted execution time: 0.2175428867340088 seconds
Sorted execution time: 1.133354663848877 seconds
Unsorted execution time: 0.21171283721923828 seconds
Sorted execution time: 0.08376479148864746 secondsJust for fun add "randomvariable = 1"(you know, 1 one cycle machine op) in a tight loop and watch tenths of a second added to your "benchmark".
Also, while I'll grant you that using Python implies that convenience, programmer time, and/or development speed is a higher priority than performance, that doesn't at all mean that people who use Python "don't care about performance".