Pranay Jain, Maximilian Kasper, Göran Köber, Axel Plinge, Dominik Seuß
The study presents a benchmarking framework for optimizing AI models on ARM Cortex processors, balancing energy efficiency and performance for sustainable embedded systems.
This research introduces a framework to help developers optimize AI models for ARM Cortex processors, which are commonly used in embedded systems like smart devices. The focus is on finding the best balance between energy use, accuracy, and resource use. By analyzing different processors, the study shows that the M7 processor is best for quick tasks, the M4 is more energy-efficient for longer tasks, and the M0+ is suitable for simpler tasks. This work aims to guide developers in creating AI systems that are both powerful and energy-efficient.
