[amelius]

It's probably in the same sense that Numpy is much faster than doing matrix operations with pure Python arrays and Python for-loops.