FPM: A Flexible Programming Model for MPSoC on FPGA

Abstract

This paper proposes a flexible programming model (FPM), which addresses the automatic parallel execution for functional tasks on heterogeneous multiprocessors. Guided by the simply annotated source codes, a front-end source to source compiler is provided to identify the parallel regions and generate the sources codes. A runtime middleware analyzes the inter-task data dependencies and schedules the tasks with renaming techniques automatically. FPM has been verified by the prototype built on state-of-art FPGA. Examples demonstrate that our model can largely ease the burden of programmers as well as uncover the task level parallelism.