Sieve of Eratosthenes is faster with list than with numpy.array

sieve

import math
import numpy
from timeit import timeit
HIGHEST = 100000
ARRAY_LEN = HIGHEST + 1 # For 0

def make_sieve():
    sieve = numpy.ones(ARRAY_LEN)
    mark_composites(sieve)
    return sieve

def make_sieve_list():
    sieve = []
    for n in range(ARRAY_LEN): sieve.append(True)
    mark_composites(sieve)
    return sieve

def mark_composites(sieve):
    for prime in range(2, int(math.sqrt(HIGHEST)) + 1):
        for n in range(prime * 2, ARRAY_LEN, prime):
            sieve[n] = False


reps = 1000
print(timeit(make_sieve_list, number=reps))
print(timeit(make_sieve, number=reps))

/Library/Frameworks/Python.framework/Versions/3.4/bin/python3 /Users/daveb/devel/python-lessons/Harder/sieve.py
36.04646955100179
49.208132846993976

Process finished with exit code 0