Design and Implementation of High Speed and Fixed-point FFT Processor
Bo Fu
Abstract
Bo Fu
Abstract
The paper presents the design method of a high speed and fixed-point FFT processor. By optimizing the memory mapping algorithm and the generation of twiddle factors, the method based on CORDIC algorithm can calculate a radix-4 butterfly in one clock cycle and has the maximal degree of parallelism. This paper also introduces a method of twiddle factor generation, which can simultaneously generate three twiddles of one radix-4 butterfly. The implementation of the twiddle factor generation in hardware is simple and don’t need extra ROM resources. The processor has been implemented on a Xilinx chip XCV2P30 and obtains the operating clock frequency at 130MHz. The processor can compute a complex 1024-point FFT with 9.8μs and 16384-point FFT with 221μs. The performance of the FFT processor is better than most available FFT processors.
A significance statement is not available in the OpenAlex record.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
The paper presents the design method of a high speed and fixed-point FFT processor. By optimizing the memory mapping algorithm and the generation of twiddle factors, the method based on CORDIC algorithm can calculate a radix-4 butterfly in one clock cycle and has the maximal degree of parallelism. This paper also introduces a method of twiddle factor generation, which can simultaneously generate three twiddles of one radix-4 butterfly. The implementation of the twiddle factor generation in hardware is simple and don’t need extra ROM resources. The processor has been implemented on a Xilinx chip XCV2P30 and obtains the operating clock frequency at 130MHz. The processor can compute a complex 1024-point FFT with 9.8μs and 16384-point FFT with 221μs. The performance of the FFT processor is better than most available FFT processors.
Key concepts: Twiddle factor, Fast Fourier transform, Computer science, Parallel computing, Split-radix FFT algorithm, Clock rate, Prime-factor FFT algorithm, Computer hardware