[Python] np.zeros becomes faster with a very large array size [duplicate]

%timeit -n 1000 -r 7 np.zeros(100000000, dtype=bool)


gives 4.22 ms

%timeit -n 1000 -r 7 np.zeros(1000000000, dtype=bool) # 1 more zero = 10 times longer array


gives 3.13 μs (not even milliseconds, microseconds!?)

How can this be possible? Is this hardware dependent?

Continue reading...