National Conference on Contemporary Computing |
Foundation of Computer Science USA |
NCCC2016 - Number 2 |
April 2017 |
Authors: Ghanshyam Verma, Priyanka Tripathi |
21021e84-46ab-4ce5-b189-741213e0b283 |
Ghanshyam Verma, Priyanka Tripathi . Generating Multi-million Data Set using GPGPU Accelerated Models. National Conference on Contemporary Computing. NCCC2016, 2 (April 2017), 4-9.
Generating synthetic data set which is realistic as well as sufficiently large has been a cumbersome task for researchers in the past. Several models have been proposed previously, all adopting heterogeneous approaches, in this work the emphasis is on speeding up the compute time of the data set distribution. Here, Uniform, Poisson and Zipf distributions have been studied and approaches with parallel computation model have been proposed. The models have been verified for speedup using CUDA based implementation on NVIDIA Quadro 2000 GPU. A speed up in the range of 2x to 6x was observed for various range of data sets.