DocumentCode :
3206478
Title :
Multi-GPU MapReduce on GPU Clusters
Author :
Stuart, Jeff A. ; Owens, John D.
Author_Institution :
Dept. of Comput. Sci., Univ. of California, Davis, CA, USA
fYear :
2011
fDate :
16-20 May 2011
Firstpage :
1068
Lastpage :
1079
Abstract :
We present GPMR, our stand-alone MapReduce library that leverages the power of GPU clusters for large-scale computing. To better utilize the GPU, we modify MapReduce by combining large amounts of map and reduce items into chunks and using partial reductions and accumulation. We use persistent map and reduce tasks and stress aspects of GPMR with a set of standard MapReduce benchmarks. We run these benchmarks on a GPU cluster and achieve desirable speedup and efficiency for all benchmarks. We compare our implementation to the current-best GPU-MapReduce library (runs only on a solo GPU) and a highly-optimized multi-core MapReduce to show the power of GPMR. We demonstrate how typical MapReduce tasks are easily modified to fit into GPMR and leverage a GPU cluster. We highlight how total and relative amounts of communication affect GPMR. We conclude with an exposition on the types of MapReduce tasks well-suited to GPMR, and why some tasks need more modifications than others to work well with GPMR.
Keywords :
computer graphic equipment; coprocessors; parallel processing; GPMR; GPU cluster; large-scale computing; multiGPU MapReduce; persistent map; stress aspect; Google; Graphics processing unit; Instruction sets; Libraries; Optimization; Pipelines; Programming;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Parallel & Distributed Processing Symposium (IPDPS), 2011 IEEE International
Conference_Location :
Anchorage, AK
ISSN :
1530-2075
Print_ISBN :
978-1-61284-372-8
Electronic_ISBN :
1530-2075
Type :
conf
DOI :
10.1109/IPDPS.2011.102
Filename :
6012914
Link To Document :
بازگشت