Floating-point Triangle Class Of Transcendental Functions Of The Algorithm And Its Hardware Implementation

Computer arithmetic is the theoretical foundation of operations in microprocessor and transcendental function is an important branch of it. Transcendental function can't be expressed in terms of algebra. It includes the exponential function, the trigonometric functions, and the inverse functions of both. It is important to implement transcendental function of high precision with proper hardware cost for high performance microprocessor design.Based on the "National Defense Preliminary Research Project", the dissertation described how to design and implement the floating-point execution Unit. The main work is listed below:1. Analyze and compare different algorithms of transcendental function, finally choose HP CORDIC to implement the computation of transcendental function in Longtium S1 system.2. Present the architecture of floating-point execution unit, and establish the detailed implementation flow of trigonometric function using HP CORDIC.3. Complete the design and coding, and carry out code check, coverage analysis, timing analysis, function verification and joint verification of floating-point execution unit along with the verification plan of Longtium S1.Now, the Longtium S1 has been taped out. The SoC integrates 486DX2 compatible microprocessor, system controller, peripheral controller, watchdog and so on. DOS6.22 and some typical applications can run smoothly on the PC 104 system, which is based on the Longtium S1 and has been verified on a flight control system. The research result is valuable to study high performance floating-point execution unit.