Fixed-Point Mathematics

Abstract

Many embedded systems applications need to perform floating-point calculations. As mentioned in the previous chapter, however, small-scale microcontrollers may not have hardware support for floating-point calculations with a floating-point unit (FPU). To avoid potentially slow floating-point emulation libraries manipulating 32-bit single-precision

or even 64-bit double-precision

many developers elect to use integer-based fixed-point arithmetic. The first part of this chapter describes fixed-point data types and presents a scalable template class representation for fixed-point. In the second part of this chapter, we will use our fixed-point class to compute some elementary transcendental functions, discuss fixed-point efficiency and develop a specialization of