Energy-Efficient Scheduling Optimization for Parallel Applications on Heterogeneous Distributed Systems

Abstract

Designing energy-efficient scheduling algorithms on heterogeneous distributed systems is increasingly becoming the focus of research. State-of-the-art works have studied scheduling by combining dynamic voltage and frequency scaling (DVFS) technology and turning off the appropriate processors to reduce dynamic and static energy consumptions. However, the methods for turning off processors are ineffective. In this study, we propose a novel method to assign priorities to processors for facilitating effective selection of turned-on processors to decrease static energy consumption. An energy-efficient scheduling algorithm based on bisection (ESAB) is proposed on this basis, and this algorithm directly turns on the most energy-efficient processors depending on the idea of bisection to reduce static energy consumption while dynamic energy consumption is decreased by using DVFS technology. Experiments are performed on fast Fourier transform, Gaussian elimination, and randomly generated parallel applications. Results show that our ESAB algorithm makes a better trade-off between reducing energy consumption and low computation time of task assignment (CTTA) than existing algorithms under different scale conditions, deadline constraints, and degrees of parallelism and heterogeneity.

Online Energy-Aware Scheduling for Deadline-Constrained Applications in Distributed Heterogeneous Systems

“…0, and f k,max = 1 GHz (all the frequency values are discrete with an accuracy of 0.01 GHz). These parameters are set exactly by reference to existing work [3,12,35]. In this experiment, 20 processors are randomly generated for each experiment.…”

“…For the processor's power model, we used the references in the existing articles [12,29,30]. The power consumption manifests in two forms: dynamic and static power consumption.…”

“…Therefore, when scheduling applications in distributed heterogeneous systems, it is necessary to consider both energy consumption and online scheduling issues simultaneously. But most studies are static [11,12] or do not consider energy consumption [13,14]. To tackle these challenges, we propose a dynamic scheduling algorithm for randomly arrived applications, namely, online minimum energy consumption with deadline-constrained scheduling (OMECDS) algorithm.…”

Scheduling energy‐conscious tasks in distributed heterogeneous computing systems

“…This section conducts the performance of the proposed ECGA by comparing with HEFT, 48 MIN-MIN, 50 ECS, 51 MPQGA, 52 ESAB, 9 and MOGA 53 algorithms since all of them have the same model and solve the same problem. The HEFT algorithm first sorts the tasks with descending order of upward rank value into a list and then assigns each task to the processor, which minimizes the earliest finish time.…”

“…Jinlin Song et al 43 proposed an efficient scheduling algorithm to minimize schedule length, which pre-assigns an energy consumption constraint with an average assignment value to every task. Nan Gao et al 9 designed an energy-efficient scheduling algorithm based on bisection, which attempts to lower energy consumption by selecting the most energy-efficient processor to turn on. Longxin Zhang et al 11 present a bi-objective genetic algorithm to find non-dominated solutions, which to pursue low energy consumption and high system reliability.…”