On the optimization of a pipeline model to integrate a reduced-dimensionality Schrodinger equation for distributed memory architectures

This paper examines the problem of dealing with the integration of a two-di mensional Schrodinger equation using a sector-diabatic coupled-channel appr oach when the coupling matrix is too large to fit into the memory of indivi dual nodes of a multiple-instruction, multiple-data (MIMD) parallel machine . For this case, the distribution of blocks of the coupling matrix among th e different nodes and the use of a pipeline model are investigated. To meas ure performances and to compare them with those measured for a previously o ptimized task farm model, the pipeline model was implemented on a hypercube and its structure optimized. Benchmarks were performed for dimensions of t he coupling matrix manageable by both task farm and pipeline models.