A novel function complexity-based code migration policy for reducing power consumption

Embedded system designs has changed greatly owing to rapid developments in both hardware and software technology. Typical designs should consider hardware limitations, such as size, weight, or battery capacity. In other words, the designs are heavily dependent on hardware components. Since hardware components can deteriorate and degenerate, hardware-aware software designs are needed to achieve power-efficient embedded systems. Previous studies usually focus on the microprocessor expecting to reduce power consumed on computation. Besides, entire program execution resulting a lot of memory accesses also consume power. Therefore, it should be considered to minimize overall power consumption for more efficient designs. Modern embedded systems often use heterogeneous memory to benefit from different characteristics of each memory device. This study aims to optimize the power efficiency of heterogeneous memory in embedded systems. We have proposed a detailed function complexity concept whose scale implies the range of power consumption in migrated memory. Afterward, function selection algorithm with function complexity selects a unique function which improve power consumption most after the migration. Several experiments and quantitative analyses with various benchmarks have been performed to validate the proposed algorithm. Consequently, migrating selected complex function successfully minimizes power consumption of an embedded system.