Methodology for efficient Execution of SPMD applications on Multicore Clusters

Abstract

Nowadays, the scientific applications are developed with more complexity and accuracy and these precisions need high computational resources to be executed. Also, the current trend in high performance computing (HPC) is to find clusters composed of Multicore nodes, and these nodes include heterogeneity levels which have to be handled carefully if we want to improve the performance metrics. The integration of these Multicore nodes in HPC (High Performance Computing) has allowed the inclusion of more parallelism within nodes, but this parallelism generates challenges that have to be managed considering some troubles present in these environments that affect the application efficiency and speedup. Aspects associated to number of cores, data locality, shared cache, communications link inside the node, etc are considered relevant when our goal is to improve the performance.