Software Migration for Parallel Execution on a Multicore Tablet: A Case Study

Recent developments in the embedded industry have shown that future products will be based on heterogeneous MPSoCs (Multi-Processor Systems-on-Chip) due to their optimal trade-off among performance, power and flexibility. Programming such MPSoC platforms is a grand challenge for embedded SoC users today. In this paper, we investigate the problem of migrating a legacy code base towards MPSoCs. We present a case study of extending an existing parallelization tool-suite to target a real-life Nvidia Tegra 3 based tablet. A systematic methodology is proposed to integrate differently parallelized parts into existing sequential code. Our experimental results showed that the proposed methodology with tools support is promising to help software migration and improve programmer's productivity.