2D Transpose FFT with distributed data files on processors on MPI

• Introduction

  Fast Fourier Transform is important in scientific and technical applications such as convolution, digital signal processing, and solving linear partial differential equations. There have been many works in implementing FFT on parallel computers.

  FFT in this paper assumes each processors reads in part of big data matrix and that data matrix is distributed on several files. This is the concept of bounded memory size M, which is realistic. The size of one file is equal to the memory size M. And this way we can achieve better scalability too.

• Implementation--

  Algorithm and implementation issues

• result(future direction) and source code