Off-Line Performance Prediction of Message-Passing Applications on Cluster Systems

Applied Parallel Computing. State of the Art in Scientific Computing

Rak

Torella

et al. 2006

Abstract. GRID Application development is a hard task. Good applications should correctly use large distributed systems, whose infrastructure heavily affects the application performance. In this paper we propose a performance oriented approach to GRID application development, founded on the use of a prototype language (MetaPL) for the description of the applications and the use of a heterogeneous system simulation environment (HeSSE) for performance prediction. We developed GRID simulation components for the existing simulation environment (HeSSE) and validated them. After that we extended the MetaPL language in order to explicitly support GRID application features and simulated a simple case study to show how the approach works.

Section: An Example: the Gauss-siedel Applicationmentioning

confidence: 99%

Performance Oriented Development and Tuning of GRID Applications

Applied Parallel Computing. State of the Art in Scientific Computing

Rak

Torella

et al. 2006

“…The adopted simulation framework, HeSSE (Heterogeneous System Simulation Environment), is a component-based simulation environment, which builds simulation models by composition of more specialized simulation models, modeling subsystems of a heterogeneous distributed system [9][10][11]. Simulation components are able to reproduce both the functional (services offered to the system) and behavioral (time spent in performing actions) actions of the different portions of the system.…”

Section: The Hesse/metapl Modeling Techniquementioning

confidence: 99%

“…The configuration needs only to be tuned with parameters taken from the real environment. In [10] we showed how to obtain this information, whether the system is available for measurements, or not.…”

Section: Application and System Modelingmentioning

confidence: 99%

A Performance-Oriented Technique for Hybrid Application Development

Recent Advances in Parallel Virtual Machine and Message Passing Interface

Rak

Torella

et al. 2004

Abstract. In SMP clusters it is not always convenient to switch from pure message-passing code to hybrid software designs that exploit shared memory. This paper tackles the problem of restructuring an existing MPI code through the insertion of OpenMP directives. The choice of the best code is carried out with a performance-oriented approach, predicting the effect of application hybridization in the MetaPL/HeSSE simulation environment, without writing and running any hybrid software. The technique is validated by applying the devised changes to the code, and comparing the predicted results to actual running time measurements.

Self-optimizing MPI Applications: A Simulation-Based Approach

High Performance Computing and Communications

Rak²,

Torella³

et al. 2005

Abstract. Historically, high performance systems use schedulers and intelligent resource managers in order to optimize system usage and application performance. Most of the times, applications just issue requests of resources to the central system. This centralized approach is an unnecessary constraint for a class of potentially flexible applications, whose resource usage may be modulated as a function of the system status. In this paper we propose a tool which, in a way essentially transparent to final users, lets the application to self-tune in function of the status of the target execution environment. The approach hinges on the use of the MetaPL/HeSSE methodology, i.e., on the use of simulation to predict execution times and skeletal descriptions of the application to describe run-time resource usage.