Evolution-Based Real-Time Job Scheduling for Co-Optimizing Processor and Memory Power Savings

Hyokyung Bahn,Kyungwoon Cho

doi:10.1109/access.2020.3017014

Hyokyung Bahn, Kyungwoon Cho

Open Access

https://doi.org/10.1109/access.2020.3017014

Copy DOI

Abstract

With the recent advances in battery-based mobile computing technologies, power-saving techniques in real-time embedded devices are becoming increasingly important. This paper presents a novel job scheduling policy for real-time systems, which aims at minimizing the power consumption of processor and memory without missing the deadline constraints of real-time jobs. To do so, we formulate the power saving techniques of processor voltage/frequency scaling and memory job placement as a unified measure, and show that it is a complex search problem that has the exponential time complexity. Thus, an efficient heuristic based on evolutionary computation is performed to cut down the huge searching space and find a reasonable schedule within the feasible time budget. To evaluate the proposed scheduling policy, we conduct experiments under various workload conditions. Our experimental results show that the proposed policy significantly reduces the energy consumption of real-time systems. Specifically, the average reduction in the energy consumption is 41.7% without deadline misses.

Highlights

Due to the recent advances in IoT (Internet of Things) and mobile computing technologies, reducing the power consumptions in battery-based real-time systems is becoming increasingly important
We propose a novel real-time job scheduling policy that aims at minimizing the power consumption in processor and memory subsystems
OPTIMIZATIONS WITH GENETIC ALGORITHMS we describe the details of our genetic algorithms to optimize the dynamic voltage/frequency scaling of a processor and job placement between DRAM and low-power memory with respect to the minimization of power consumption

Summary

INTRODUCTION

Due to the recent advances in IoT (Internet of Things) and mobile computing technologies, reducing the power consumptions in battery-based real-time systems is becoming increasingly important. The problem becomes the optimization problem of determining the processor’s voltage/frequency mode and the memory location of each job for minimizing the power consumption without missing deadlines. A job τi is represented by < ti, pi, mi >, where ti is the worst case execution time of τi with the default voltage/frequency mode of a processor and DRAM memory placement, pi is the period of τi, and mi is the memory configuration of τi, which is defined as < si, ri, wi >, where si is the size of τi’s memory footprint, and ri and wi are the number of memory read and write operations, respectively, during the execution of τi. The worst case execution time ti of a job is determined by the slower component of processor and memory with the given voltage/frequency mode and the memory medium. Note that LPM does not need the refresh power as it is a non-volatile medium

EXTENDING THE BASIC MODEL

SELECTION OPERATIONS

CROSSOVER AND MUTATION OPERATIONS

PERFORMANCE EVALUATIONS

Findings

CONCLUSION