We present the results obtained by using an evolution of our CUDA-based solution for the exploration, via a Breadth First Search, of large graphs. This latest version exploits at its best the features of the Kepler architecture and relies on a combination of techniques to reduce both the number of communications among the GPUs and the amount of exchanged data. The final result is a code that can visit more than 800 billion edges in a second by using a cluster equipped with 4096 Tesla K20X GPUs.
@article{arxiv.1408.1605,
title = {Parallel Distributed Breadth First Search on the Kepler Architecture},
author = {Mauro Bisson and Massimo Bernaschi and Enrico Mastrostefano},
journal= {arXiv preprint arXiv:1408.1605},
year = {2014}
}
Comments
In this revision we adopt a technique to reduce the size of exchanged messages that relies on the use of a bitmap. This change halves, by itself, the total execution time. Now the code reaches 800 GTEPS on 4096 Kepler GPUs. We also made some modifications to the Introduction and to the performance section. Added new references