Multitasking a Navier-Stokes algorithm on the CRAY-2

Abstract

This paper presents the results of mukitasking a Navier-Stokes algorithm on the CRAY-2. The algorithm is a compact difference scheme for the solution of the incompressible, two-dimensional, time-dependent NavierStokes equations. Two implementations of multitasking on the CRAY-2 are considered: macrotasking (parallelism at the subroutine level) and microtasking (parallelism at the do-loop level). These two techniques are briefly described. The implementation of the algorithm is discussed in relation to these techniques, and the results for three problem sizes are presented. The timing results for both techniques are, in general, comparable with differences ranging between 2 % and 14%, depending on the problem size. The best achieved speedup in a dedicated environment is 3.62 for macrotasking and 3.32 for microtasking. The task granularity for both techniques is computed, and the synchronization costs are estimated. For macrotasks of granularity of up to 0.5 msec, microtasking outperformed macrotasking, while the latter outperformed the former for granularity of over one msec.

Escalonamento de processos: uma abordagem dinâmica e incremental para a exploração de características de aplicações paralelas

“…O benchmark LU (lower-upper symmetric gauss-seidel) resolve sistemas de equações lineares através do método LU, empregando um tamanho de bloco igual a 5 >< 5. Esse problema representa os cálculos associados a uma classe de aplicações de dinâmica dos fluidos, chamada de INS3D-LU (Fatoohi, 1989(Fatoohi, , 1992.…”