Adapting distributed scientific applications to run-time network conditions

Authors:
Masha Sosonkina
Affiliations:
Ames Laboratory and Iowa State University, Ames, IA
Venue:
PARA'04 Proceedings of the 7th international conference on Applied Parallel Computing: state of the Art in Scientific Computing
Year:
2004

Citing 8
Cited 1

LogP: towards a realistic model of parallel computation

PPOPP '93 Proceedings of the fourth ACM SIGPLAN symposium on Principles and practice of parallel programming
Prediction and adaptation in Active Harmony

Cluster Computing
Design of a Tool for Providing Dynamic Network Information to an Application

PaCT '01 Proceedings of the 6th International Conference on Parallel Computing Technologies
Iterative Methods for Sparse Linear Systems

Iterative Methods for Sparse Linear Systems
An Evaluation of Current High-Performance Networks

IPDPS '03 Proceedings of the 17th International Symposium on Parallel and Distributed Processing
MPI: A Message-Passing Interface Standard

MPI: A Message-Passing Interface Standard
System support for bandwidth management and content adaptation in internet applications

OSDI'00 Proceedings of the 4th conference on Symposium on Operating System Design & Implementation - Volume 4
A framework for integrating network information into distributed iterative solution of sparse linear systems

VECPAR'02 Proceedings of the 5th international conference on High performance computing for computational science

Electronic Structure Calculations and Adaptation Scheme in Multi-core Computing Environments

ICCS '09 Proceedings of the 9th International Conference on Computational Science: Part I

Quantified Score

Hi-index	0.00

Visualization

Abstract

High-performance applications place great demands on computation and communication resources of distributed computing platforms. If the availability of resources changes dynamically, the application performance may suffer, which is especially true for clusters. Thus, it is desirable to make an application aware of system run-time changes and to adapt it dynamically to the new conditions. We show how this may be done using a helper tool (middleware NICAN). In our experiments, NICAN implements a packet probing technique to detect contention on cluster nodes while a distributed iterative linear system solver from the pARMS package is executing. Adapting the solver to the discovered network conditions may result in faster iterative convergence.