English

Implicit Decomposition for Write-Efficient Connectivity Algorithms

Data Structures and Algorithms 2017-10-10 v1

Abstract

The future of main memory appears to lie in the direction of new technologies that provide strong capacity-to-performance ratios, but have write operations that are much more expensive than reads in terms of latency, bandwidth, and energy. Motivated by this trend, we propose sequential and parallel algorithms to solve graph connectivity problems using significantly fewer writes than conventional algorithms. Our primary algorithmic tool is the construction of an o(n)o(n)-sized "implicit decomposition" of a bounded-degree graph GG on nn nodes, which combined with read-only access to GG enables fast answers to connectivity and biconnectivity queries on GG. The construction breaks the linear-write "barrier", resulting in costs that are asymptotically lower than conventional algorithms while adding only a modest cost to querying time. For general non-sparse graphs on mm edges, we also provide the first o(m)o(m) writes and O(m)O(m) operations parallel algorithms for connectivity and biconnectivity. These algorithms provide insight into how applications can efficiently process computations on large graphs in systems with read-write asymmetry.

Keywords

Cite

@article{arxiv.1710.02637,
  title  = {Implicit Decomposition for Write-Efficient Connectivity Algorithms},
  author = {Naama Ben-David and Guy E. Blelloch and Jeremy T. Fineman and Phillip B. Gibbons and Yan Gu and Charles McGuffey and Julian Shun},
  journal= {arXiv preprint arXiv:1710.02637},
  year   = {2017}
}