Orchestrating parallel detection of strongly connected components on GPUs.

Xuhao Chen,Cheng Chen,Jie Shen,Jianbin Fang,Tao Tang,Canqun Yang,Zhiying Wang
DOI: https://doi.org/10.1016/j.parco.2017.11.001
IF: 0.983
2018-01-01
Parallel Computing
Abstract:•Analysis on graph properties: skewed SCC sizes and dynamically changing structure.•A hybrid parallelism method is proposed to deal with skewed SCC sizes.•Traversal strategies are customized to fit dynamically changing graph structure.•Our method outperforms existing GPU and OpenMP implementations.
What problem does this paper attempt to address?