A many processing element framework for the Discrete Fourier Transform

Abstract

For the greater part of the last 30 years of computing history, computational processing has been primarily based on performance improvements of sequential processors. Research has shown that the future no longer lies in the direction of sequential processor performance improvements, but rather on the adoption of parallel processors and the development of suitable architectures. Computing in general has settled on generic suitable-for-all processing architectures, however recent research has since highlighted the importance of adopting many simpler processing elements, with a tailored focus on domain specific designs. In light of this, this work focuses on the development of a parallel framework for the execution of the Discrete Fourier Transform using a recursive algorithm using many simple processing elements. The system described in this work shows a 64 point Discrete Fourier Transform computation in parallel, which achieves a throughput of 2.19GSPS(39.49Gbit/s). The framework proposed promotes adjustable parallelism using the many simple processing elements, and allows simple adoption on existing systems.