Research And Design Of The CORDIC Algorithm Based On Floating Point Numbers

Angle rotation and successive approximation is used in CORDIC algorithm to achieve functional operation with three models, i.e. circle model, linear model and hyperbolic model, which are integrated into a standardized form, so as to transform complicated operation into simple shifting operation and add operation. CORDIC algorithm is especially suitable for hardware realization, for it takes rate, precision, simplicity and high efficiency into account, which are not owned by other algorithm. Due to these exclusive advantages, CORDIC algorithm is widely applied in many fields. How to improve the rate as well as reduce the area and the power dissipation with the premise of ensuring the precision has become the key to the researches on CORDIC algorithm.To meet the requirements above, the CORDIC algorithm is optimized in this paper. On the whole, we realize the function with the combination of iterative method and table look-up scheme. In the actual design, the multiplication is automatically achieved by the multiplier generated from library. In comparison, more adders are used. In this paper, the performances of five kinds of adders are compared with TSMC-0.13μm CMOS technology, e.g. Ripple Carry Adder, Carry Look-ahead Adder, Brent-kung Adder, Kogge-stone Adder and Ladner-fischer Adder, and then the 32-bit Carry Look-ahead Adders with an area of 5168.26μm2 , the power of 313.608mW and the delay time of 6.02ns. Finally, based on the above work, the three kinds of architecture, pipelined, non-pipelined and segmental pipelined, are realized and comprehensively analyzed with TSMC-0.13μm CMOS technology. As the result shows, they have their own advantages and disadvantages, and in specific applications we can choose different implementation architectures according to specific needs.This paper adopts IEEE754 standard, uses Verilog HDL (Hardware Design Language) to carry out RTL language description of floating-point-number-based CORDIC algorithm, analyzes the data stream and the inner structure with Debussy, and applies Modelsim simulation wave to achieve the functional operation of 32-bit single-precision floating point numbers. And then the result of matlab simulation is verified, and compared with that of Modelsim simulation, so as to prove that the algorithm has met the anticipated precision requirement (2^-23). Finally, a high-performance CORDIC algorithm is implemented by Integrated tools.