Fixed-point math in time-critical C applications

When optimizing C to improve the speed of execution, using fixed-point math as opposed to floating-point emulation can help the user make big improvements. It is the single largest improvement a programmer can make in C to reduce execution time. The paper starts with floating-point equations, and shows the steps necessary to convert them to fixed-point equations. Basic equations are used at the beginning, but then progress to more advanced problems in the fixed-point implementation. The paper also analyzes the assembly output after converting to fixed-point, and data are presented from multiple platforms that show the speed improvement