DVFS based heterogeneous scheduling for power optimisation and load balancing in HPC clustersaper

Author

Nath, Rintu ; Nagaraju, A.

Author_Institution

Dept. of Comput. Sci. & Eng., Central Univ. of Rajasthan, Ajmer, India

fYear

2015

fDate

19-20 Feb. 2015

Firstpage

1

Lastpage

4

Abstract

As computation intensive scientific applications rely on high performance computing (HPC), energy consumption is a challenging issue. Reducing energy consumption is a major design constraint for modern heterogeneous computing systems. Conventional energy efficient scheduling strategies developed on distributed systems do not sufficiently exploit the system elasticity and adaptability for maximum energy savings. The conventional scheduling algorithm, dynamic power management (DPM) or dynamic voltage and frequency scaling (DVFS) do not simultaneously take account of energy usage and optimal system performance. In this paper, a novel scheduling strategy on DVFS enabled HPC clusters is proposed.

Keywords

parallel processing; power aware computing; resource allocation; DPM; DVFS-based heterogeneous scheduling; HPC clustersaper; computation intensive scientific applications; distributed systems; dynamic power management; dynamic voltage-and-frequency scaling; heterogeneous computing systems; high performance computing; load balancing; optimal system performance; power optimisation; Algorithm design and analysis; Energy consumption; Energy efficiency; Real-time systems; Scheduling; Scheduling algorithms; DPM; DVFS; energy efficiency; load balancing; scheduling;

fLanguage

English

Publisher

ieee

Conference_Titel

Parallel Computing Technologies (PARCOMPTECH), 2015 National Conference on

Conference_Location

Bangalore

Print_ISBN

978-1-4799-6916-6

Type

conf

DOI

10.1109/PARCOMPTECH.2015.7084517

Filename

7084517