Digital Signal Processing: World Class Designs

Chapter 8. DSP Processors and Fixed-Point Arithmetic

This chapter presents the basic concepts of DSP hardware and software. And by basic, I mean it starts by referencing von Neumann’s seminal 1946 paper introducing the von Neumann architecture, and contrasts it with the almost-as-old Harvard architecture favored by most DSPs. It then delves into the operation of basic execution units such as the multiply-accumulate (MAC) unit, shifter, and address generation unit (AGU).You may be wondering, “Why do I need to know how a shifter works? Why do I need to know the difference between an opcode and operand?” The reason is that DSP algorithms are computationally intensive, and generally require high levels of optimization to meet performance, power, ...