Allocating processing power to minimize time costs in parallel software systems

Several optimal allocation methods based on the modified computation structure model are proposed. These methods are general, and they achieve optimal processing power allocation in most cases. It is assumed that the underlying computer system has a finite number of processors, all the processors have the same speed and that they communicate with each other through a shared memory. The basic ideas behind these methods are to allocate more processing power to operations having higher time costs and containing more parallel paths. In this way, operations with lower time costs are forced to wait for operations with higher time costs. The time cost can be minimized if all the operations performed in parallel are completed at the same time. It is shown how a computation achieves its minimum time costs while minimizing the required overall processing power