Goal Formulation: Abstracting Dynamic Objectives for Efficient On-chip Resource Allocation
Elham Shamsa, Anil Kanduri, Amir M. Rahmani, Pasi Liljeberg, Axel Jantsch, Nikil D. Dutt · 2018
Run-time resource management of mobile heterogeneous systems is challenging due to the limited energy budget that has to be allocated among diverse workloads. User interaction within these systems alter the performance requirements, which often conflicts with concurrent applications' objectives and system constraints. Current resource allocation approaches focus on optimizing fixed objectives, ignoring the variation in system and applications' constraints at run-time. For adaptive resource allocation, it is necessary to abstract the applications' and system's requirements into goals, which can be dynamically formulated as a weighted combination of different objectives. We highlight the problem by illustrating the limitation of state-of-the-art resource allocation approaches and motivate the need of a goal management solution, which dynamically prioritizes different objectives and switches between them to adapt to the environment.