How does NumPy handle element-wise operations at the C level?

NumPy executes element-wise operations using precompiled C loops inside ufuncs.

When you perform operations like addition or multiplication, NumPy does not iterate in Python. Instead, it dispatches the operation to a ufunc (universal function), which contains optimized C loops. These loops operate directly on raw memory buffers, reading and writing values sequentially. The execution is further optimized using cache-friendly access patterns, SIMD instructions (when available), and minimal Python overhead. Broadcasting rules are applied before execution to align shapes.