numpy arrays vs core python lists

Asked May 13 '21 at 09:40

Active May 13 '21 at 09:40

Viewed 21 times

I've heard numpy is much faster than core python lists in case of manipulation speed . but when I ran these pieces of code I got oposite. Can anyone explain to me how this happens ?

def pure_py_list():
"""to see speed of normal python lists"""
t1 = time.time()
total = 0
x = range(10000000)
for item in x :
    total+=1
t2 = time.time()
return (t2 - t1)

def numpy_list ():
"""calculating speed of numpy arrays ."""
t1 = time.time()
total = 0
x = np.arange(10000000)
for element in x :
    total +=1
t2 = time.time()
return (t2 - t1)

thanks a lot.

asked May 13 '21 at 09:40

davood asgharzadeh

`numpy` is fast if you use `numpy` methods. – juanpa.arrivillaga May 13 '21 at 09:42
please fix the code indentation – meowulf May 13 '21 at 09:43
replace your for loop with `np.sum(x)` and try again. – Feodoran May 13 '21 at 09:43
You aren't manipulating anything and the range is a generator. – doctorlove May 13 '21 at 09:45
1

Anyway, iterating over a `numpy.ndarray` at the Python level is slow because the array stores primitive numeric types in an underlying buffer. That's why it *can be fast*. But, if you iterate over it, or index into it, it requires the creation of a wrapper object. On the other hand, iterating over a list doesn't create a new object for each element, the elements are already Python objects. – juanpa.arrivillaga May 13 '21 at 09:45
@doctorlove range is not a generator, but actually, iterating over a `range` object is *slower* than iterating over a `list` object. – juanpa.arrivillaga May 13 '21 at 09:46

numpy arrays vs core python lists

0 Answers0