Conclusions and future research work

This book has presented a task scheduling methodology which can effectively explore the performance vs energy trade-off space when mapping dynamical and concurrent tasks onto multiple heterogeneous processors. The principle of this methodology is a two-phase scheduling framework, i.e., the design-time scheduling phase and the run-time scheduling phase. Both a basic (Chapter 4), and a scalable (Chapter 5) design-time scheduling techniques are described in order to perform intra-task (thread node) level design space exploration. These design-time scheduling techniques are different from the existing scheduling algorithms in their systematically explorations of the design trade-off space instead of obtaining a single optimal solution. The fast and effective exploration of large design space based on these design-time scheduling techniques provide system architects the crucial information to perform the system level trade-offs.