Application Phase Behavior-Guided Thermal Management of Embedded Platforms

This letter presents a scheduling policy for multicore embedded applications based on their thermal phase behavior. The thermal phase-guided task allocation allows the scheduler to allocate tasks that run at elevated temperature to low-power cores while others running at normal temperature to high performing cores. The proposed phase-based thermal aware technique shows 15.4 °C reduction in average temperature and 30.5 °C in peak temperature when experimented with PARSEC benchmark applications targeted on the ODROID XU4 embedded platform.

I. Introduction

Increased power density in embedded devices introduces several thermal issues leading to degradation in their performance and life span [10]. Literature suggests that a number of software-based solutions for thermal management in embedded systems have been proposed considering the strict size and design constraints imposed on these systems. Cui et al. [1] proposed decentralized thermal-aware task scheduling for large-scale many core system while Jayaseelan and Mitra [4] proposed to throttle the CPU using hot and cold tasks. Shaik and Baskiyar [9] proposed to put the hot process in the sleep mode until CPU resumes normal temperature. In addition to these solutions, researchers have proposed a number of proactive approaches as well. For example, Henkel et al. [3] have proposed power density-based smart thermal management techniques considering heterogeneity at chip and application level while the same group also proposed dynamic guard band selection for heterogeneous system considering aging effect in transistor due to frequent implementation of dynamic voltage frequency scaling (DVFS) and high voltage of chip in [5]. A number of proactive approaches for thermal management in multicore systems have been proposed by Pathania et al. [6]. Prakash et al. [8] proposed an efficient energy-aware static partition algorithm in conjunction with DVFS to maximize power–performance tradeoff. Their group have also proposed efficient proactive thermal management of CPU–GPU systems as in [7] where the proposal has been to apply a co-operative approach to control the frequency of both CPU and graphics processor unit (GPU) using DVFS.