FFT on a new parallel vector processor
A new parallel processing system has been proposed, and a prototype model of the system has been constructed. It is designed to perform parallel vector operations at maximum efficiency. In addition, it can also handle communicating vector operations, and hence exploit irregular parallelism present in many apparently sequential algorithms. The system is therefore suitable for a wide range of algorithms with varying degrees of parallelism. In this paper, we give a brief description of the system, and discuss the implementation of the Cooley-Tukey FFT on this system. We show that the system's versatility allows it to achieve a near maximum degree of parallelism for this algorithm in the asymptotic case.
Unable to display preview. Download preview PDF.
- R.H. Barlow and D.J. Evans, Parallel Algorithms for the Iterative Solution to Linear Systems, Computer J. 25 (1982) 56–60.Google Scholar
- E.O. Brigham, The Fast Fourier Transform (Prentice-Hall 1974).Google Scholar
- J.W. Cooley and J.W. Tukey, An Algorithm for the Machine Calculation of Complex Fourier Series, Math. Comput. 19 (1965) 297–301.Google Scholar
- R.N. Ibbett, P.C. Capon and N.P. Topham, MU6V: A Parallel Vector Processing System, ACM Computer Architecture News 13 (1985) 136–144.Google Scholar
- R.H. Perrot, A Language for Array and Vector Processors, ACM TOPLAS 1 (1979) 177–195.Google Scholar
- H.S. Stone, Parallel Processing with the Perfect Shuffle, IEEE Trans. Comput. 20 (1971) 153–161.Google Scholar
- P.N. Swarztrauber, FFT algorithms for Vector Computers, Parallel Computing 1 (1984) 45–63.Google Scholar