ABSTRACT
This chapter presents some low-power consumption aspects of the main arithmetic operators and representations of numbers used in high-performance circuits. The optimization of a circuit can be done at different levels: system, algorithm, architecture, circuit, and technology. Addition is the most-used arithmetic operation in microprocessors, digital signal processors, and many digital circuits. The core operation in some computation units such as arithmetic and logic units, but it also plays a part in multipliers, dividers, floating-point units, and address generation units. Multiplication is the most area consuming arithmetic operation in high-performance circuits. There are many research works on low-power design of high-speed multipliers. Three main classes of algorithms are available for the evaluation of the elementary functions: polynomial or rational approximations, shift-and-add methods, and table-based methods. The algorithms based on polynomial or rational approximations do not require specific hardware support. The algorithms used to perform the arithmetic operations in the floating-point number system are more complex than in the fixed-point system.
