We present a novel method to implement the radix-2 fast Fourier transform (FFT) algorithm on field programmable gate arrays (FPGA). The FFT architecture exploits parallelism by having more pipelined units in the stages

and more parallel units within a stage. It has the noticeable advantages of high speed and more efficient resource utilization by employing four ganged butterfly engines (GBEs)

and can be well matched to the placement of the resources on the FPGA. We adopt the decimation-in-frequency (DIF) radix-2 FFT algorithm and implement the FFT processor on a state-of-the-art FPGA. Experimental results show that the processor can compute 1024-point complex radix-2 FFT in about 11 μs with a clock frequency of 200 MHz.