Mar-12-2021, 03:55 PM
After reading a lot on the subject I decided to abandon the conversion to numpy as "dtype=object" is absolutely necessary for manipulating very large numbers (larger than 2**64) and using numpy arrays with dtype=object means that the array in itself is in contiguous blocks of memory, but the content are pointers to the integers so they have to be looked up in the same manner as in python lists. This article explains the details of standard lists in Python.
In any case, I didn't get better performance except in special cases with specific array lengths, so I won't go on with numpy. It is excellent in many cases but I seem to have found a case where it isn't.
In any case, I didn't get better performance except in special cases with specific array lengths, so I won't go on with numpy. It is excellent in many cases but I seem to have found a case where it isn't.