DocumentCode :
1642350
Title :
GPU accelerated relational hash join operation
Author :
Devarajan, Narayanasamy ; Navneeth, S. ; Mohanavalli, S.
Author_Institution :
Dept. of Inf. Technol., SSN Coll. of Eng., Kalavakkam, India
fYear :
2013
Firstpage :
891
Lastpage :
896
Abstract :
Most of the modern applications involves organization and quick processing of data which demands the use of a database management system. GPUs are now considered the future of computing and the promise that they deliver in the field of image processing and graphics rendering, along with their computation capability nominates them to be used for high performance computing. Modern GPUs are now being used, on a larger scale to solve non-graphical problems taking advantage of their availability and affordability. This has opened new avenues for the exploitation of parallel architecture to accelerate massive data processing operations. This paper looks ahead to parallelize hash join algorithm which is an efficient method practiced to carry out join operation in database management systems. The algorithm is parallelized using CUDA and the results on a NVIDIA 9600 GT exhibits a speedup of up to 45x depending on the size of input relations and result set.
Keywords :
graphics processing units; parallel algorithms; parallel architectures; relational databases; CUDA; GPU; NVIDIA 9600 GT; compute unified device architecture; database management systems; graphics processing unit; parallel algorithm; relational hash join operation; Algorithm design and analysis; Database systems; Graphics processing units; Informatics; Instruction sets; Partitioning algorithms; Throughput; CUDA; Databases; GPGPU; Hash join; Parallel join; SQL;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Advances in Computing, Communications and Informatics (ICACCI), 2013 International Conference on
Conference_Location :
Mysore
Print_ISBN :
978-1-4799-2432-5
Type :
conf
DOI :
10.1109/ICACCI.2013.6637294
Filename :
6637294
Link To Document :
بازگشت