Array Redistribution in ScaLAPACK Using PVM
Jack J. Dongarra, Loı̈c Prylli, Cyril Randriamaro, Bernard Tourancheau · 1995
Linear algebra on distributed-memory parallel computers raises the problem of data distribution of matrices and vectors among the processes. Block-cyclic distribution works well for most algorithms. The block size must be chosen carefully, however, in order to achieve good efficiency and good load balancing. This choice depends heavily on each operation; hence, it is essential to be able to go from one distribution to another very quickly. We present here the algorithms implemented in the ScaLAPACK library,and we discuss timing results ona network of workstations and on a Cray T3D using PVM. 1 Introduction The problem of data redistributionoccurs when one deals with arrays (from vectors to multidimensional arrays) on parallel distributed-memory computers. Data redistributionapplies both to data-parallel languages such as High Performance Fortran (HPF) and to single-process multiple-data (SPMD) programs with message passing. In the first case the redistribution is implicit in array sta...