### Abstract

We introduce a new parallel algorithm for approximate breadth-first ordering of an unweighted graph by using bounded asynchrony to parametrically control both the performance and error of the algorithm. This work is based on the k-level asynchronous (KLA) paradigm that trades expensive global synchronizations in the levelsynchronous model for local synchronizations in the asynchronous model, which may result in redundant work. Instead of correcting errors introduced by asynchrony and redoing work as in KLA, in this work we control the amount of work that is redone and thus the amount of error allowed, leading to higher performance at the expense of a loss of precision. Results of an implementation of this algorithm are presented on up to 32,768 cores, showing 2.27x improvement over the exact KLA algorithm and 3.8x improvement over the level-synchronous version with minimal error on several graph inputs.

Original language | English (US) |
---|---|

Title of host publication | Languages and Compilers for Parallel Computing - 29th International Workshop, LCPC 2016, Revised Papers |

Editors | Chen Ding, John Criswell, Peng Wu |

Publisher | Springer-Verlag |

Pages | 40-54 |

Number of pages | 15 |

ISBN (Print) | 9783319527086 |

DOIs | |

State | Published - Jan 1 2017 |

Externally published | Yes |

Event | 29th International Workshop on Languages and Compilers for Parallel Computing, LCPC 2016 - Rochester, United States Duration: Sep 28 2016 → Sep 30 2016 |

### Publication series

Name | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
---|---|

Volume | 10136 LNCS |

ISSN (Print) | 0302-9743 |

ISSN (Electronic) | 1611-3349 |

### Conference

Conference | 29th International Workshop on Languages and Compilers for Parallel Computing, LCPC 2016 |
---|---|

Country | United States |

City | Rochester |

Period | 9/28/16 → 9/30/16 |

### Fingerprint

### Keywords

- Approximate algorithms
- Asynchronous
- Breadth-first search
- Distance query
- Distributed memory
- Parallel graph algorithms

### ASJC Scopus subject areas

- Theoretical Computer Science
- Computer Science(all)

### Cite this

*Languages and Compilers for Parallel Computing - 29th International Workshop, LCPC 2016, Revised Papers*(pp. 40-54). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10136 LNCS). Springer-Verlag. https://doi.org/10.1007/978-3-319-52709-3_4

**Fast approximate distance queries in unweighted graphs using bounded asynchrony.** / Fidel, Adam; Sabido, Francisco Coral; Riedel, Colton; Amato, Nancy Marie; Rauchwerger, Lawrence.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

*Languages and Compilers for Parallel Computing - 29th International Workshop, LCPC 2016, Revised Papers.*Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 10136 LNCS, Springer-Verlag, pp. 40-54, 29th International Workshop on Languages and Compilers for Parallel Computing, LCPC 2016, Rochester, United States, 9/28/16. https://doi.org/10.1007/978-3-319-52709-3_4

}

TY - GEN

T1 - Fast approximate distance queries in unweighted graphs using bounded asynchrony

AU - Fidel, Adam

AU - Sabido, Francisco Coral

AU - Riedel, Colton

AU - Amato, Nancy Marie

AU - Rauchwerger, Lawrence

PY - 2017/1/1

Y1 - 2017/1/1

N2 - We introduce a new parallel algorithm for approximate breadth-first ordering of an unweighted graph by using bounded asynchrony to parametrically control both the performance and error of the algorithm. This work is based on the k-level asynchronous (KLA) paradigm that trades expensive global synchronizations in the levelsynchronous model for local synchronizations in the asynchronous model, which may result in redundant work. Instead of correcting errors introduced by asynchrony and redoing work as in KLA, in this work we control the amount of work that is redone and thus the amount of error allowed, leading to higher performance at the expense of a loss of precision. Results of an implementation of this algorithm are presented on up to 32,768 cores, showing 2.27x improvement over the exact KLA algorithm and 3.8x improvement over the level-synchronous version with minimal error on several graph inputs.

AB - We introduce a new parallel algorithm for approximate breadth-first ordering of an unweighted graph by using bounded asynchrony to parametrically control both the performance and error of the algorithm. This work is based on the k-level asynchronous (KLA) paradigm that trades expensive global synchronizations in the levelsynchronous model for local synchronizations in the asynchronous model, which may result in redundant work. Instead of correcting errors introduced by asynchrony and redoing work as in KLA, in this work we control the amount of work that is redone and thus the amount of error allowed, leading to higher performance at the expense of a loss of precision. Results of an implementation of this algorithm are presented on up to 32,768 cores, showing 2.27x improvement over the exact KLA algorithm and 3.8x improvement over the level-synchronous version with minimal error on several graph inputs.

KW - Approximate algorithms

KW - Asynchronous

KW - Breadth-first search

KW - Distance query

KW - Distributed memory

KW - Parallel graph algorithms

UR - http://www.scopus.com/inward/record.url?scp=85011422305&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85011422305&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-52709-3_4

DO - 10.1007/978-3-319-52709-3_4

M3 - Conference contribution

AN - SCOPUS:85011422305

SN - 9783319527086

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 40

EP - 54

BT - Languages and Compilers for Parallel Computing - 29th International Workshop, LCPC 2016, Revised Papers

A2 - Ding, Chen

A2 - Criswell, John

A2 - Wu, Peng

PB - Springer-Verlag

ER -