A sub-optimal assignment of application tasks onto heterogeneous systems

Presents a mathematical programming model for finding the sub-optimal assignment of a given application task onto a heterogeneous suite of computers. The proposed model is based on the Cluster-M heterogeneous programming paradigm. Using Cluster-M, an application task is represented in form of a Cluster-M specification which indicates all the concurrent and communicating subtasks at every step of computation. These specifications are then to be mapped onto the underlying heterogeneous suite of computers represented in a Cluster-M representation format. To map every step of Cluster-M specification onto the Cluster-M representation of the heterogeneous suite, we propose to use an optimal synthesis technique called SOS (synthesis of systems). We formulate and solve each of the mapping steps in the form of a mixed integer linear programming model. This leads to a very fast sub-optimal selection and mapping solution.<
>