Accelerating Direction-Optimized Breadth First Search on Hybrid Architectures
University of British Columbia
arXiv:1503.04359 [cs.DC], (14 Mar 2015)
@article{sallinen2015accelerating,
title={Accelerating Direction-Optimized Breadth First Search on Hybrid Architectures},
author={Sallinen, Scott and Gharaibeh, Abdullah and Ripeanu, Matei},
year={2015},
month={mar},
archivePrefix={"arXiv"},
primaryClass={cs.DC}
}
Large scale-free graphs are famously difficult to process efficiently: the highly skewed vertex degree distribution makes it difficult to obtain balanced workload partitions for parallel processing. Our research instead aims to take advantage of vertex degree heterogeneity by partitioning the workload to match the strength of the individual computing elements in a hybrid architecture. This paper extends the direction-optimized breadth first search algorithm to work efficiently on hybrid, GPU-accelerated platforms. We present the key graph partitioning, workload allocation, and communication strategies required for massive concurrency and good overall performance. We show that exploiting specialization enables gains as high as 2.4x in terms of time-to-solution and 2.0x in terms of energy efficiency by adding 2 GPUs to a 2 CPU-only baseline, for synthetic graphs with up to 16 Billion undirected edges as well as for real-world graphs. We also show that, for a capped energy envelope, it is more efficient to add a GPU than an additional CPU.
March 18, 2015 by hgpu
