Pipeline-based scheduling for heterogeneous multi-core systems

This paper proposes a pipeline-based scheduling algorithm to improve the throughput for heterogeneous multi-core processors. The algorithm decreases the communication and idle time by assigning most parent-child task pairs on the same processor, and balances the work time of processors through a balance condition. The experimental results show that, on average, this algorithm with the above two techniques achieves 113.99% improvement in throughput compared to some existing algorithms.