Part 2 of this book is concerned with developing Python solutions from the perspective of extracting maximum performance from commonly available hardware. We first discuss using lower-level languages that are closer to the hardware to extract more speed from the CPU. Namely, we concentrate on Cython, a superset of Python that generates efficient C code. We then focus on modern hardware architectures and how they sometimes require counterintuitive approaches to extract maximum performance. Our discussion includes how modern Python libraries, like NumExpr, are designed to take advantage of the hardware.