Dynamic load balancing in MPI jobs

Authors:
Gladys Utrera;Julita Corbalán;Jesús Labarta
Affiliations:
Departament d'Arquitectura de Computadors, Universitat Politècnica de Catalunya;Departament d'Arquitectura de Computadors, Universitat Politècnica de Catalunya;Departament d'Arquitectura de Computadors, Universitat Politècnica de Catalunya
Venue:
ISHPC'05/ALPS'06 Proceedings of the 6th international symposium on high-performance computing and 1st international conference on Advanced low power systems
Year:
2005

Citing 16
Cited 1

Large-scale parallel programming: experience with BBN butterfly parallel processor

PPEALS '88 Proceedings of the ACM/SIGPLAN conference on Parallel programming: experience with applications, languages and systems
Process control and scheduling issues for multiprogrammed shared-memory multiprocessors

SOSP '89 Proceedings of the twelfth ACM symposium on Operating systems principles
The Performance Implications of Thread Management Alternatives for Shared-Memory Multiprocessors

IEEE Transactions on Computers
Scheduling Support for Concurrency and Parallelism in the Mach Operating System

Computer
The impact of operating system scheduling policies and synchronization methods of performance of parallel applications

SIGMETRICS '91 Proceedings of the 1991 ACM SIGMETRICS conference on Measurement and modeling of computer systems
Analysis of task migration in shared-memory multiprocessor scheduling

SIGMETRICS '91 Proceedings of the 1991 ACM SIGMETRICS conference on Measurement and modeling of computer systems
The implications of cache affinity on processor scheduling for multiprogrammed, shared memory multiprocessors

SOSP '91 Proceedings of the thirteenth ACM symposium on Operating systems principles
A closer look at coscheduling approaches for a network of workstations

Proceedings of the eleventh annual ACM symposium on Parallel algorithms and architectures
Implicit coscheduling: coordinated scheduling with implicit information in distributed systems

ACM Transactions on Computer Systems (TOCS)
An infrastructure for efficient parallel job execution in Terascale computing environments

SC '98 Proceedings of the 1998 ACM/IEEE conference on Supercomputing
Adaptive Load Balancing for MPI Programs

ICCS '01 Proceedings of the International Conference on Computational Science-Part II
Improved Utilization and Responsiveness with Gang Scheduling

IPPS '97 Proceedings of the Job Scheduling Strategies for Parallel Processing
MEMSY - A Modular Expandable Multiprocessor System

Parallel Computer Architectures: Theory, Hardware, Software, Applications
Flexible CoScheduling: Mitigating Load Imbalance and Improving Utilization of Heterogeneous Resources

IPDPS '03 Proceedings of the 17th International Symposium on Parallel and Distributed Processing
Implementing Malleability on MPI Jobs

Proceedings of the 13th International Conference on Parallel Architectures and Compilation Techniques
DITools: application-level support for dynamic extension and flexible composition

ATEC '00 Proceedings of the annual conference on USENIX Annual Technical Conference

Robotic clusters: Multi-robot systems as computer clusters

Robotics and Autonomous Systems

Quantified Score

Hi-index	0.01

Visualization

Abstract

There are at least three dimensions of overhead to be considered by any parallel job scheduling algorithm: load balancing, synchronization, and communication overhead. In this work we first study several heuristics to choose the next to run from a global processes queue. After that we present a mechanism to decide at runtime weather to apply Local process queue per processor or Global processes queue per job, depending on the load balancing degree of the job, without any previous knowledge of it.