Learning Pareto-Frontier Resource Management Policies for Heterogeneous SoCs: A Information-Theoretic Approach
Event Type
Research Manuscript
Virtual Programs
Hosted in Virtual Platform
Embedded System Design Methodologies
Embedded Systems
DescriptionMobile SoCs are growing in their complexity and heterogeneity to meet the needs of emerging applications. This makes it challenging to optimally manage the resources at runtime to meet the desired trade-offs among target objectives such as performance and energy. In this paper, we propose a novel information-theoretic framework referred as {\em PaRMIS} to create pareto-optimal resource management policies for given target applications and design objectives. Experiments on a commercial heterogeneous SoC using 12 benchmark applications show that PaRMIS achieves better Pareto fronts and is easily usable to optimize complex objectives when compared to state-of-the-art methods.