ANN Predicted Apps-Usage Aware Linux Scheduler for Asymmetrical Multi Cluster SoC

摘要

This research introduces an improved Linux scheduler that models and manages several power dissipation problems based on user application usage pattern identified in mobile computing platform. The scheduler is developed and simulated in Android 5.0 with the Linux kernel 3.10 and 3.14. The scheduler monitors multiple system performance metrics at runtime, predicts power dissipation and future workload with an ANN computation block. To exploit DVFS capabilities in ARM asymmetrical SoC designs a three clustered CPU setup had been introduced in this study and the scheduler interacts with the Linux Load Balancing mechanism to passively achieve optimized operating frequency for each application thread.