2020/03/20 by Sumit K. Mandal, Ganapati Bhat, Mandal, Sumit K. +7 · 1 citation
Computer Science · Engineering · #Distributed #Energy Harvesting in Wireless Networks #FOS: Computer and information sciences #FOS: Electrical engineering #Green IT and Sustainability #Machine Learning (cs.LG) #Parallel #Parallel Computing and Optimization Techniques #Systems and Control (eess.SY) #and Cluster Computing (cs.DC) #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.2003.09526
openalex publication_date 2020/03/20 · openalex created_date 2022/07/26 · openalex updated_date 2026/07/28
Mobile platforms must satisfy the contradictory requirements of fast response\ntime and minimum energy consumption as a function of dynamically changing\napplications. To address this need, system-on-chips (SoC) that are at the heart\nof these devices provide a variety of control knobs, such as the number of\nactive cores and their voltage/frequency levels. Controlling these knobs\noptimally at runtime is challenging for two reasons. First, the large\nconfiguration space prohibits exhaustive solutions. Second, control policies\ndesigned offline are at best sub-optimal since many potential new applications\nare unknown at design-time. We address these challenges by proposing an online\nimitation learning approach. Our key idea is to construct an offline policy and\nadapt it online to new applications to optimize a given metric (e.g., energy).\nThe proposed methodology leverages the supervision enabled by power-performance\nmodels learned at runtime. We demonstrate its effectiveness on a commercial\nmobile platform with 16 diverse benchmarks. Our approach successfully adapts\nthe control policy to an unknown application after executing less than 25% of\nits instructions.\n