On the Advantages of an Alternative MPI Execution Model for Grids

Sena, Alexandre C.; Nascimento, Aline P.; Silva, J.A. da; Vianna, D.; Boeres, Cristina; Rebello, Vinod E. F.

doi:10.1109/ccgrid.2007.74

Cited by 9 publications

(8 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the master-worker LAM/MPI version, the master process distributes units of workload equivalent to that of a process P to 24 worker processes on demand. While as expected the results show that the AMS execution is slightly slower than that of pure LAM/MPI for a homogeneous environment, this scenario is reversed in shared heterogenous Grids [4].…”

Section: Computational Experimentssupporting

confidence: 65%

“…The fact that resources maybe heterogeneous, shared with local jobs, and fault-prone not only makes this model inefficient but also makes managing the execution of applications extremely complex. A one process per task (1PTask) execution model has been proposed where programs consist of a larger number (determined by the parallelism of the application and not the number of resources) of shorter-lived processes [4]. For this model, we propose to adopt a fault tolerance strategy based solely on message logging.…”

Section: Concepts and Related Workmentioning

confidence: 99%

“…This paper describes and evaluates the integrated self-healing mechanism adopted by the AMS to permit EasyGrid MPI applications to complete their execution in Grids even in the face of failures. The EasyGrid AMS adopts the "one MPI process per application task" grid execution model [4] and therefore is based on MPI implementations which support dynamic process creation such as LAM/MPI. For reasons of deployability, no modifications are made standard distribution or installation of LAM/MPI and no further software, other than the Globus ToolKit 2.x, is required for execution of autonomic EasyGrid MPI applications in a Grid.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Low Cost Self-healing in MPI Applications

Silva

Rebello

2007

Recent Advances in Parallel Virtual Machine and Message Passing Interface

View full text Add to dashboard Cite

Abstract. Writing applications capable of executing efficiently in Grids is extremely difficult and tedious for inexperienced users. The distributed resources are typically heterogeneous, non-dedicated, and are offered without any performance or availability guarantees. Systems capable of adapting the execution of an application to the dynamic characteristics of the Grid are essential. This work describes the strategy used to bestow the self-healing property on autonomic EasyGrid MPI applications to withstand process and resource failures. This paper highlights both the difficulties and the low cost solution adopted to offer fault tolerance in applications based on the standard Grid installation of LAM/MPI.

show abstract

Section: Computational Experimentssupporting

confidence: 65%

Section: Concepts and Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Low Cost Self-healing in MPI Applications

Silva

Rebello

2007

Recent Advances in Parallel Virtual Machine and Message Passing Interface

View full text Add to dashboard Cite

show abstract

“…For the two smaller values of N , the AMS ring algorithm's execution times were, as expected, slightly higher (6.6% and 1.0%) due to overheads of the EasyGrid AMS [9] to manage both the creation and execution of processes and the extra communication incurred by following the alternative execution model [11]. As N increases, this overhead diminishes to the point that AMS ring achieves better results than MPI ring.…”

Section: Homogeneous Environmentsupporting

confidence: 57%

“…To maximise the performance of the N -body ring algorithm when executing in not only physically heterogeneous, but also dynamic and shared environments, this work proposes a new approach based on the execution model specified in [11]. Instead of executing one long running process per processor, a set of smaller processes (called tasks) with precedence relationships among them is defined.…”

Section: Grid-enabling the Ring Algorithmmentioning

confidence: 99%

EasyGrid Enabling of Iterative Tightly-Coupled Parallel MPI Applications

Sena

Nascimento

Boeres

et al. 2008

2008 IEEE International Symposium on Parallel and Distributed Processing With Applications

View full text Add to dashboard Cite

This paper addresses the challenge of how to permit tightly coupled parallel applications, optimised for uniform, stable, static environments, execute equally efficiently in environments which exhibit the complete opposite characteristics. Using the N -body problem as a case study, both the traditional and proposed grid enabled MPI implementations of the popular ring algorithm are analysed. Results with respect to performance show the latter approach to be competitive on a cluster and significantly more effective in heterogeneous and dynamic environments.

show abstract

An autonomic parallel strategy for exhaustive search tree algorithms on shared or heterogeneous systems

Passos,

Rebello

2023

Concurrency and Computation

View full text Add to dashboard Cite

SummaryBacktracking branch‐and‐prune (BP) algorithms and their variants are exhaustive search tree techniques widely employed to solve optimization problems in many scientific areas. However, they characteristically often demand significant amounts of computing power for problem sizes representative of real‐world scenarios. Given that their search domains can often be partitioned, these algorithms are frequently designed to execute in parallel by harnessing distributed computing systems. However, to achieve efficient parallel execution times, an effective strategy is required to balance the nonuniform partition workloads across the available resources. Furthermore, with the increasing integration of servers with heterogeneous resources and the adoption of resource sharing, balancing workloads is becoming complex. This paper proposes a strategy to execute parallel BP algorithms more efficiently on even shared or heterogeneous distributed systems. The approach integrates a self‐adjusting dynamic partitioning method in the BP algorithm with a dynamic scheduler, provided by an application middleware, which manages the parallel execution while addressing any issues of imbalance. Empirical results indicate better scalability with efficiencies above 90% for instances of an application case study for the discretizable molecular distance geometry problem (DMDGP). Improvements of up to 38% were obtained in execution speed‐ups compared to a more traditional parallel BP implementation for DMDGP.

show abstract

On the Advantages of an Alternative MPI Execution Model for Grids

Cited by 9 publications

References 9 publications

Low Cost Self-healing in MPI Applications

Low Cost Self-healing in MPI Applications

EasyGrid Enabling of Iterative Tightly-Coupled Parallel MPI Applications

An autonomic parallel strategy for exhaustive search tree algorithms on shared or heterogeneous systems

Contact Info

Product

Resources

About