A Study of Adaptive Co-scheduling Approach for an Opportunistic Software Environment to Execute in Multi-core and Multi-Processor Configurations

Multi-cores and multi-processors loosely coupled are being considered interesting architectures to increase computational power in many organizations. However, these environments brought a new challenge for the scientific community that can be understood as the limited number of applications developed to take advantage of these new configurations. Opportunistic software environments also need to adapt to these new configurations, otherwise it is not expected that these software environments be able to use efficient idle capacity from these architectures. In this paper, it is presented an empirical study of an adaptive coscheduling approach, using an opportunistic package to gather resources from an environment compound by multi-core and multi-processor loosely coupled elements. The proposal approach adopts multiple threads to use idle processing from multi-core and multi-processor resources. Empirical tests proved that the proposal was successful in gathering idle resources from these configurations, enhancing the performance of a demonstration application.