Vectorization vs apply in pandas

Understand why vectorized NumPy operations outperform row-wise apply in pandas.

You have a DataFrame with columns 'price' and 'quantity' and want a new 'total' column. Which approach is generally the fastest for large DataFrames?