### Abstract

We present an algorithm for computing the communication sets in array section movements with block-cyclic (cyclic(k) in HPF) distribution. Our framework can handle multi-level alignments, multi-dimensional arrays, array intrinsic functions, affine indices and axis exchanges in the array subscript. Instead of employing the linear diophantine equation solver, a new algorithm which does not rely on the linear diophantine equation solver to calculate communication sets is proposed We use formal proof and experimental results to show that it is more efficient than previous solution to the same problem. Another important contribution of the paper is that we prove that the compiler is able to compute efficiently the communication sets of block-cyclic distribution as long as the block sizes of the arrays are set to be identical or the lowest common multiple (LCM) of block sizes is not a huge integer We demonstrate it by thorough complexity analyses and extensive experimental results.

Original language | English |
---|---|

Title of host publication | Proceedings - International Conference on Parallel Processing Workshops, ICPPW 2002 |

Editors | Stephan Olariu |

Publisher | Institute of Electrical and Electronics Engineers Inc. |

Pages | 623-631 |

Number of pages | 9 |

ISBN (Electronic) | 0769516807 |

DOIs | |

Publication status | Published - 2002 Jan 1 |

Event | International Conference on Parallel Processing Workshops, ICPPW 2002 - Vancouver, Canada Duration: 2002 Aug 18 → 2002 Aug 21 |

### Publication series

Name | Proceedings of the International Conference on Parallel Processing Workshops |
---|---|

Volume | 2002-January |

ISSN (Print) | 1530-2016 |

### Other

Other | International Conference on Parallel Processing Workshops, ICPPW 2002 |
---|---|

Country | Canada |

City | Vancouver |

Period | 02/8/18 → 02/8/21 |

### Fingerprint

### Keywords

- Block-Cyclic Distributions
- Data Parallel Programs
- Distributed Memory Machines
- HPF Compiler
- Parallelizing Compiler

### ASJC Scopus subject areas

- Software
- Mathematics(all)
- Hardware and Architecture

### Cite this

*Proceedings - International Conference on Parallel Processing Workshops, ICPPW 2002*(pp. 623-631). [1039785] (Proceedings of the International Conference on Parallel Processing Workshops; Vol. 2002-January). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICPPW.2002.1039785

**An efficient algorithm for computing communication sets for data parallel programs with block-cyclic distribution.** / Hwang, Gwan-Hwan.

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

*Proceedings - International Conference on Parallel Processing Workshops, ICPPW 2002.*, 1039785, Proceedings of the International Conference on Parallel Processing Workshops, vol. 2002-January, Institute of Electrical and Electronics Engineers Inc., pp. 623-631, International Conference on Parallel Processing Workshops, ICPPW 2002, Vancouver, Canada, 02/8/18. https://doi.org/10.1109/ICPPW.2002.1039785

}

TY - GEN

T1 - An efficient algorithm for computing communication sets for data parallel programs with block-cyclic distribution

AU - Hwang, Gwan-Hwan

PY - 2002/1/1

Y1 - 2002/1/1

N2 - We present an algorithm for computing the communication sets in array section movements with block-cyclic (cyclic(k) in HPF) distribution. Our framework can handle multi-level alignments, multi-dimensional arrays, array intrinsic functions, affine indices and axis exchanges in the array subscript. Instead of employing the linear diophantine equation solver, a new algorithm which does not rely on the linear diophantine equation solver to calculate communication sets is proposed We use formal proof and experimental results to show that it is more efficient than previous solution to the same problem. Another important contribution of the paper is that we prove that the compiler is able to compute efficiently the communication sets of block-cyclic distribution as long as the block sizes of the arrays are set to be identical or the lowest common multiple (LCM) of block sizes is not a huge integer We demonstrate it by thorough complexity analyses and extensive experimental results.

AB - We present an algorithm for computing the communication sets in array section movements with block-cyclic (cyclic(k) in HPF) distribution. Our framework can handle multi-level alignments, multi-dimensional arrays, array intrinsic functions, affine indices and axis exchanges in the array subscript. Instead of employing the linear diophantine equation solver, a new algorithm which does not rely on the linear diophantine equation solver to calculate communication sets is proposed We use formal proof and experimental results to show that it is more efficient than previous solution to the same problem. Another important contribution of the paper is that we prove that the compiler is able to compute efficiently the communication sets of block-cyclic distribution as long as the block sizes of the arrays are set to be identical or the lowest common multiple (LCM) of block sizes is not a huge integer We demonstrate it by thorough complexity analyses and extensive experimental results.

KW - Block-Cyclic Distributions

KW - Data Parallel Programs

KW - Distributed Memory Machines

KW - HPF Compiler

KW - Parallelizing Compiler

UR - http://www.scopus.com/inward/record.url?scp=84954518547&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84954518547&partnerID=8YFLogxK

U2 - 10.1109/ICPPW.2002.1039785

DO - 10.1109/ICPPW.2002.1039785

M3 - Conference contribution

AN - SCOPUS:84954518547

T3 - Proceedings of the International Conference on Parallel Processing Workshops

SP - 623

EP - 631

BT - Proceedings - International Conference on Parallel Processing Workshops, ICPPW 2002

A2 - Olariu, Stephan

PB - Institute of Electrical and Electronics Engineers Inc.

ER -