Show simple item record

dc.contributor.authorDalcin, Lisandro
dc.contributor.authorMortensen, Mikael
dc.contributor.authorKeyes, David E.
dc.date.accessioned2019-05-21T13:09:36Z
dc.date.available2019-05-21T13:09:36Z
dc.date.issued2019-03-11
dc.identifier.citationDalcin L, Mortensen M, Keyes DE (2019) Fast parallel multidimensional FFT using advanced MPI. Journal of Parallel and Distributed Computing 128: 137–150. Available: http://dx.doi.org/10.1016/j.jpdc.2019.02.006.
dc.identifier.issn0743-7315
dc.identifier.doi10.1016/j.jpdc.2019.02.006
dc.identifier.urihttp://hdl.handle.net/10754/653025
dc.description.abstractWe present a new method for performing global redistributions of multidimensional arrays essential to parallel fast Fourier (or similar) transforms. Traditional methods use standard all-to-all collective communication of contiguous memory buffers, thus necessarily requiring local data realignment steps intermixed in-between redistribution and transform steps. Instead, our method takes advantage of subarray datatypes and generalized all-to-all scatter/gather from the MPI-2 standard to communicate discontiguous memory buffers, effectively eliminating the need for local data realignments. Despite generalized all-to-all communication of discontiguous data being generally slower, our proposal economizes in local work. For a range of strong and weak scaling tests, we found the overall performance of our method to be on par and often better than well-established libraries like MPI-FFTW, P3DFFT, and 2DECOMP&FFT. We provide compact routines implemented at the highest possible level using the MPI bindings for the C programming language. These routines apply to any global redistribution, over any two directions of a multidimensional array, decomposed on arbitrary Cartesian processor grids (1D slabs, 2D pencils, or even higher-dimensional decompositions). The high level implementation makes the code easy to read, maintain, and eventually extend. Our approach enables for future speedups from optimizations in the internal datatype handling engines within MPI implementations.
dc.description.sponsorshipM. Mortensen acknowledges support from the 4DSpace Strategic Research Initiative at the University of Oslo, Norway. L. Dalcin and D.E. Keyes acknowledge support from King Abdullah University of Science and Technology (KAUST), Saudi Arabia and the KAUST Supercomputing Laboratory, Saudi Arabia for the use of the Shaheen supercomputer.
dc.publisherElsevier BV
dc.relation.urlhttps://www.sciencedirect.com/science/article/pii/S074373151830306X
dc.subjectALLTOALLW
dc.subjectFFT
dc.subjectMPI
dc.subjectPencil
dc.subjectSlab
dc.titleFast parallel multidimensional FFT using advanced MPI
dc.typeArticle
dc.contributor.departmentExtreme Computing Research Center
dc.contributor.departmentComputer, Electrical and Mathematical Sciences and Engineering (CEMSE) Division
dc.contributor.departmentApplied Mathematics and Computational Science Program
dc.identifier.journalJournal of Parallel and Distributed Computing
dc.contributor.institutionDepartment of Mathematics, University of Oslo, Oslo, , Norway
kaust.personDalcin, Lisandro
kaust.personKeyes, David E.
dc.date.published-online2019-03-11
dc.date.published-print2019-06


This item appears in the following Collection(s)

Show simple item record