ABSTRACT
A task-to-processor mapping that balances the workload of the processors will typically increase the overall system throughput and will reduce the computations execution times. In our previous work, we considered the scheduling problem for stochastic tasks, where the time cost distributions of these tasks are known a priori and we developed a heuristic scheduling algorithm that balances the processors loads, based on the load averaging. We developed a new scheduling algorithm using task partitioning whenever needed to improve the results produced by the previous approach. Both algorithms are presented in this paper.
