DocumentCode :
3516154
Title :
rCUDA: Reducing the number of GPU-based accelerators in high performance clusters
Author :
Duato, José ; Pena, Antonio J. ; Silla, Federico ; Mayo, Rafael ; Quintana-Ortí, Enrique S.
Author_Institution :
Univ. Politec. de Valencia (UPV), Valencia, Spain
fYear :
2010
fDate :
June 28 2010-July 2 2010
Firstpage :
224
Lastpage :
231
Abstract :
The increasing computing requirements for GPUs (Graphics Processing Units) have favoured the design and marketing of commodity devices that nowadays can also be used to accelerate general purpose computing. Therefore, future high performance clusters intended for HPC (High Performance Computing) will likely include such devices. However, high-end GPU-based accelerators used in HPC feature a considerable energy consumption, so that attaching a GPU to every node of a cluster has a strong impact on its overall power consumption. In this paper we detail a framework that enables remote GPU acceleration in HPC clusters, thus allowing a reduction in the number of accelerators installed in the cluster. This leads to energy, acquisition, maintenance, and space savings.
Keywords :
computer graphic equipment; coprocessors; power aware computing; workstation clusters; GPU-based accelerators; energy consumption; general purpose computing; graphics processing units; high performance clusters; high performance computing; power consumption; rCUDA; Acceleration; Driver circuits; Graphics processing unit; Kernel; Libraries; Runtime; Servers; CUDA; Energy saving; clusters; high performance computing; virtualization;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
High Performance Computing and Simulation (HPCS), 2010 International Conference on
Conference_Location :
Caen
Print_ISBN :
978-1-4244-6827-0
Type :
conf
DOI :
10.1109/HPCS.2010.5547126
Filename :
5547126
Link To Document :
بازگشت