Home // ICCGI 2015, The Tenth International Multi-Conference on Computing in the Global Information Technology // View article

Energy-Efficient Thread Migration via Dynamic Characterization of Resource Utilization

Authors:
Claudia Alvarado,
Dan Tamir
Apan Qasem

Keywords: energy efficiency; thread scheduling; workload characterization.

Abstract:
The affinity of threads with cores in current chip-multiprocessor systems has a substantial impact on the execution time, latency, and power consumption of multi-threaded workloads. Finding an optimal mapping configuration of threads is a significant challenge as it requires detailed knowledge of each thread’s demands for shared system resources. This paper describes a software-based strategy that makes judicious thread migration decisions founded on careful inspection of dynamic resource utilization. The main novelty of the system reported in this paper is the extensive utilization of hardware performance counters to develop a set of synthesized metrics that capture resource contention among co-running threads. Experimental results with a set of contemporary parallel workloads, show that the system can achieve significant improvements in power consumption and performance over the default scheduling heuristics implemented in the Linux kernel.

Pages: 132 to 137

Copyright: Copyright (c) IARIA, 2015

Publication date: October 11, 2015

Published in: conference

ISSN: 2308-4529

ISBN: 978-1-61208-432-9

Location: St. Julians, Malta

Dates: from October 11, 2015 to October 16, 2015