Nanyang Technological University
Real-time requirements of applications can limit the number of tasks executing in a Multi-Processor System-on-Chip (MPSoC) platform. This demand for efficient mapping algorithm that is capable of maximizing computation performance. This paper introduces a multi-tasking heterogeneous MPSoC platform along with two run-time mapping algorithms to maximize the performance by minimizing the communication overhead. The proposed algorithms have been shown to be capable of upholding the proximity of tasks in order to reduce the communication overhead by mapping the adjacent hardware tasks onto the same reconfigurable processing element whenever possible.