Document Type
Conference Proceeding
Language
eng
Format of Original
8 p.
Publication Date
8-2013
Publisher
Institute of Electrical and Electronics Engineers (IEEE)
Source Publication
2013 IEEE International Conference on Green Computing and Communications (GreenCom) and IEEE Internet of Things(iThings) and IEEE Cyber, Physical and Social Computing(CPSCom)
Original Item ID
doi: 10.1109/GreenCom-iThings-CPSCom.2013.49
Abstract
High energy cost has become a salient constraint of the next generation of multicore based supercomputers. One approach that has the potential to conserve energy is to reduce the number of resources allocated for a given parallel application. However, this approach raises the concern that utilizing bounded resources may adversely affect performance. In this paper, we demonstrate that utilizing bounded resources to execute parallel tasks with dependency on multicore systems can actually conserve energy without degrading performance. We achieve this goal by proposing BREES, an energy-efficient scheduling algorithm for multicore systems with bounded resources. The proposed BREES algorithm takes advantage of the Dynamic Voltage Scaling (DVS) algorithm and the task duplication strategy. In addition, a dynamic waiting window (DWW) is implemented in BREES to handle the system hardware heterogeneity. We evaluate the effectiveness of BREES by conducting a series of experiments using both real world and synthetically generated parallel applications on fifteen different multicore processors and four well-known high speed networks.
Recommended Citation
Zong, Ziliang; Bush, Jonathan; Ge, Rong; Li, Xin; and Chen, Zizhong, "Energy-Efficient Scheduling for Multicore Systems with Bounded Resources" (2013). Mathematics, Statistics and Computer Science Faculty Research and Publications. 178.
https://epublications.marquette.edu/mscs_fac/178
Comments
Accepted version. Published as part of the proceedings of the conference, 2013 IEEE International Conference on Green Computing and Communications (GreenCom) and IEEE Internet of Things(iThings) and IEEE Cyber, Physical and Social Computing(CPSCom), 2013: 163-170. DOI. © 2013 IEEE. Used with permission.