Why NumPy is Fast

Why NumPy is Fast

Speed matters. Think of it like using a sports car instead of a bicycle for a race. NumPy's performance comes from smart design choices.

Vectorization

Vectorization means operating on entire arrays at once instead of element-by-element.

import numpy as np
import time

size = 1000000
arr = np.arange(size)

start = time.time()
result = arr * 2
print(f"Vectorized: {time.time() - start:.4f}s")

start = time.time()
result = np.array([x * 2 for x in arr])
print(f"Loop: {time.time() - start:.4f}s")
    

Vectorized operations can be 10-100x faster than Python loops!

C Under the Hood

NumPy is written in C, which is much faster than Python for numerical operations.

import numpy as np
import sys

py_list = list(range(1000))
np_arr = np.arange(1000)

print(f"Python list size: {sys.getsizeof(py_list)} bytes")
print(f"NumPy array size: {np_arr.nbytes} bytes")
    

Here is the thing - NumPy uses less memory and does operations faster because of C implementation.

Contiguous Memory

NumPy stores data in contiguous memory blocks, which improves cache performance.

import numpy as np

arr = np.array([[1, 2, 3], [4, 5, 6]])
print(f"C contiguous: {arr.flags['C_CONTIGUOUS']}")
print(f"Memory layout: {arr.strides}")
    

One thing that confused me at first was why Python lists are slow. They store pointers to objects, not the objects themselves.

Key Takeaways

• Vectorization operates on entire arrays at once
• NumPy is implemented in C for speed
• Contiguous memory improves cache performance
• NumPy uses less memory than Python lists