Self-optimization of large scale wildfire simulations

Jingmei Yang, Huoping Chen, Salim A Hariri, Manish Parashar

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Citations (Scopus)

Abstract

The development of efficient parallel algorithms for large scale wildfire simulations is a challenging research problem because the factors that determine wildfire behavior are complex. These factors make static parallel algorithms inefficient, especially when large number of processors is used because we cannot predict accurately the propagation of the fire and its computational requirements at runtime. In this paper, we propose an Autonomic Runtime Manager (ARM) to dynamically exploit the physics properties of the fire simulation and use them as the basis of our self-optimization algorithm. At each step of the wildfire simulation, the ARM decomposes the computational domain into several natural regions (e.g., burning, unburned, burned) where each region has the same temporal and special characteristics. The number of burning, unburned and burned cells determines the current state of the fire simulation and can then be used to accurately predict the computational power required for each region. By regularly monitoring and analyzing the state of the simulation, and using that to drive the runtime optimization, we can achieve significant performance gains because we can efficiently balance the computational load on each processor. Our experimental results show that the performance of the fire simulation has been improved by 45% when compared with a static portioning algorithm.

Original languageEnglish (US)
Title of host publicationLecture Notes in Computer Science
EditorsV.S. Sunderam, G.D. Albada, P.M.A. Sloot, J.J. Dongarra
Pages615-622
Number of pages8
Volume3514
EditionI
StatePublished - 2005
Event5th International Conference on Computational Science - ICCS 2005 - Atlanta, GA, United States
Duration: May 22 2005May 25 2005

Other

Other5th International Conference on Computational Science - ICCS 2005
CountryUnited States
CityAtlanta, GA
Period5/22/055/25/05

Fingerprint

Fires
Parallel algorithms
Managers
Physics
Monitoring

ASJC Scopus subject areas

  • Computer Science (miscellaneous)

Cite this

Yang, J., Chen, H., Hariri, S. A., & Parashar, M. (2005). Self-optimization of large scale wildfire simulations. In V. S. Sunderam, G. D. Albada, P. M. A. Sloot, & J. J. Dongarra (Eds.), Lecture Notes in Computer Science (I ed., Vol. 3514, pp. 615-622)

Self-optimization of large scale wildfire simulations. / Yang, Jingmei; Chen, Huoping; Hariri, Salim A; Parashar, Manish.

Lecture Notes in Computer Science. ed. / V.S. Sunderam; G.D. Albada; P.M.A. Sloot; J.J. Dongarra. Vol. 3514 I. ed. 2005. p. 615-622.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Yang, J, Chen, H, Hariri, SA & Parashar, M 2005, Self-optimization of large scale wildfire simulations. in VS Sunderam, GD Albada, PMA Sloot & JJ Dongarra (eds), Lecture Notes in Computer Science. I edn, vol. 3514, pp. 615-622, 5th International Conference on Computational Science - ICCS 2005, Atlanta, GA, United States, 5/22/05.
Yang J, Chen H, Hariri SA, Parashar M. Self-optimization of large scale wildfire simulations. In Sunderam VS, Albada GD, Sloot PMA, Dongarra JJ, editors, Lecture Notes in Computer Science. I ed. Vol. 3514. 2005. p. 615-622
Yang, Jingmei ; Chen, Huoping ; Hariri, Salim A ; Parashar, Manish. / Self-optimization of large scale wildfire simulations. Lecture Notes in Computer Science. editor / V.S. Sunderam ; G.D. Albada ; P.M.A. Sloot ; J.J. Dongarra. Vol. 3514 I. ed. 2005. pp. 615-622
@inproceedings{97fea7ffe28a4ae2b710a9176708995a,
title = "Self-optimization of large scale wildfire simulations",
abstract = "The development of efficient parallel algorithms for large scale wildfire simulations is a challenging research problem because the factors that determine wildfire behavior are complex. These factors make static parallel algorithms inefficient, especially when large number of processors is used because we cannot predict accurately the propagation of the fire and its computational requirements at runtime. In this paper, we propose an Autonomic Runtime Manager (ARM) to dynamically exploit the physics properties of the fire simulation and use them as the basis of our self-optimization algorithm. At each step of the wildfire simulation, the ARM decomposes the computational domain into several natural regions (e.g., burning, unburned, burned) where each region has the same temporal and special characteristics. The number of burning, unburned and burned cells determines the current state of the fire simulation and can then be used to accurately predict the computational power required for each region. By regularly monitoring and analyzing the state of the simulation, and using that to drive the runtime optimization, we can achieve significant performance gains because we can efficiently balance the computational load on each processor. Our experimental results show that the performance of the fire simulation has been improved by 45{\%} when compared with a static portioning algorithm.",
author = "Jingmei Yang and Huoping Chen and Hariri, {Salim A} and Manish Parashar",
year = "2005",
language = "English (US)",
volume = "3514",
pages = "615--622",
editor = "V.S. Sunderam and G.D. Albada and P.M.A. Sloot and J.J. Dongarra",
booktitle = "Lecture Notes in Computer Science",
edition = "I",

}

TY - GEN

T1 - Self-optimization of large scale wildfire simulations

AU - Yang, Jingmei

AU - Chen, Huoping

AU - Hariri, Salim A

AU - Parashar, Manish

PY - 2005

Y1 - 2005

N2 - The development of efficient parallel algorithms for large scale wildfire simulations is a challenging research problem because the factors that determine wildfire behavior are complex. These factors make static parallel algorithms inefficient, especially when large number of processors is used because we cannot predict accurately the propagation of the fire and its computational requirements at runtime. In this paper, we propose an Autonomic Runtime Manager (ARM) to dynamically exploit the physics properties of the fire simulation and use them as the basis of our self-optimization algorithm. At each step of the wildfire simulation, the ARM decomposes the computational domain into several natural regions (e.g., burning, unburned, burned) where each region has the same temporal and special characteristics. The number of burning, unburned and burned cells determines the current state of the fire simulation and can then be used to accurately predict the computational power required for each region. By regularly monitoring and analyzing the state of the simulation, and using that to drive the runtime optimization, we can achieve significant performance gains because we can efficiently balance the computational load on each processor. Our experimental results show that the performance of the fire simulation has been improved by 45% when compared with a static portioning algorithm.

AB - The development of efficient parallel algorithms for large scale wildfire simulations is a challenging research problem because the factors that determine wildfire behavior are complex. These factors make static parallel algorithms inefficient, especially when large number of processors is used because we cannot predict accurately the propagation of the fire and its computational requirements at runtime. In this paper, we propose an Autonomic Runtime Manager (ARM) to dynamically exploit the physics properties of the fire simulation and use them as the basis of our self-optimization algorithm. At each step of the wildfire simulation, the ARM decomposes the computational domain into several natural regions (e.g., burning, unburned, burned) where each region has the same temporal and special characteristics. The number of burning, unburned and burned cells determines the current state of the fire simulation and can then be used to accurately predict the computational power required for each region. By regularly monitoring and analyzing the state of the simulation, and using that to drive the runtime optimization, we can achieve significant performance gains because we can efficiently balance the computational load on each processor. Our experimental results show that the performance of the fire simulation has been improved by 45% when compared with a static portioning algorithm.

UR - http://www.scopus.com/inward/record.url?scp=25144444329&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=25144444329&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:25144444329

VL - 3514

SP - 615

EP - 622

BT - Lecture Notes in Computer Science

A2 - Sunderam, V.S.

A2 - Albada, G.D.

A2 - Sloot, P.M.A.

A2 - Dongarra, J.J.

ER -