Efficient implementation of a multidimensional fast fourier transform on a distributedmemory parallel multinode computer
Abstract
The present in invention is directed to a method, system and program storage device for efficiently implementing a multidimensional Fast Fourier Transform (FFT) of a multidimensional array comprising a plurality of elements initially distributed in a multinode computer system comprising a plurality of nodes in communication over a network, comprising: distributing the plurality of elements of the array in a first dimension across the plurality of nodes of the computer system over the network to facilitate a first onedimensional FFT; performing the first onedimensional FFT on the elements of the array distributed at each node in the first dimension; redistributing the onedimensional FFTtransformed elements at each node in a second dimension via "alltoall" distribution in random order across other nodes of the computer system over the network; and performing a second onedimensional FFT on elements of the array redistributed at each node in the second dimension, wherein the random order facilitates efficient utilization of the network thereby efficiently implementing the multidimensional FFT. The "alltoall" redistribution of array elements is further efficiently implemented in applications other than the multidimensional FFT on the distributedmemory parallel supercomputer.
 Inventors:

 Princeton, NJ
 CrotonOnHudson, NY
 Mount Kisco, NY
 Irvington, NY
 Cortlandt Manor, NY
 Bedford Hills, NY
 Issue Date:
 Research Org.:
 Lawrence Livermore National Laboratory (LLNL), Livermore, CA (United States)
 Sponsoring Org.:
 USDOE
 OSTI Identifier:
 921629
 Patent Number(s):
 7315877
 Application Number:
 10/468,998
 Assignee:
 International Business Machines Corporation (Armonk, NY)
 Patent Classifications (CPCs):

F  MECHANICAL ENGINEERING F04  POSITIVE  DISPLACEMENT MACHINES FOR LIQUIDS F04D  NONPOSITIVEDISPLACEMENT PUMPS
F  MECHANICAL ENGINEERING F24  HEATING F24F  AIRCONDITIONING
 DOE Contract Number:
 W7405ENG48
 Resource Type:
 Patent
 Country of Publication:
 United States
 Language:
 English
 Subject:
 97 MATHEMATICS AND COMPUTING
Citation Formats
Bhanot, Gyan V, Chen, Dong, Gara, Alan G, Giampapa, Mark E, Heidelberger, Philip, SteinmacherBurow, Burkhard D, and Vranas, Pavlos M. Efficient implementation of a multidimensional fast fourier transform on a distributedmemory parallel multinode computer. United States: N. p., 2008.
Web.
Bhanot, Gyan V, Chen, Dong, Gara, Alan G, Giampapa, Mark E, Heidelberger, Philip, SteinmacherBurow, Burkhard D, & Vranas, Pavlos M. Efficient implementation of a multidimensional fast fourier transform on a distributedmemory parallel multinode computer. United States.
Bhanot, Gyan V, Chen, Dong, Gara, Alan G, Giampapa, Mark E, Heidelberger, Philip, SteinmacherBurow, Burkhard D, and Vranas, Pavlos M. Tue .
"Efficient implementation of a multidimensional fast fourier transform on a distributedmemory parallel multinode computer". United States. https://www.osti.gov/servlets/purl/921629.
@article{osti_921629,
title = {Efficient implementation of a multidimensional fast fourier transform on a distributedmemory parallel multinode computer},
author = {Bhanot, Gyan V and Chen, Dong and Gara, Alan G and Giampapa, Mark E and Heidelberger, Philip and SteinmacherBurow, Burkhard D and Vranas, Pavlos M},
abstractNote = {The present in invention is directed to a method, system and program storage device for efficiently implementing a multidimensional Fast Fourier Transform (FFT) of a multidimensional array comprising a plurality of elements initially distributed in a multinode computer system comprising a plurality of nodes in communication over a network, comprising: distributing the plurality of elements of the array in a first dimension across the plurality of nodes of the computer system over the network to facilitate a first onedimensional FFT; performing the first onedimensional FFT on the elements of the array distributed at each node in the first dimension; redistributing the onedimensional FFTtransformed elements at each node in a second dimension via "alltoall" distribution in random order across other nodes of the computer system over the network; and performing a second onedimensional FFT on elements of the array redistributed at each node in the second dimension, wherein the random order facilitates efficient utilization of the network thereby efficiently implementing the multidimensional FFT. The "alltoall" redistribution of array elements is further efficiently implemented in applications other than the multidimensional FFT on the distributedmemory parallel supercomputer.},
doi = {},
journal = {},
number = ,
volume = ,
place = {United States},
year = {2008},
month = {1}
}