Extraction of Parallel Application Signatures for Performance Prediction

2012 IEEE International Conference on Cluster Computing Workshops

Luque

2012

Self Cite

The increase in computational power in processing units and the complexity of scientific applications that use high performance computing require more efficient Input/Output (I/O) systems. To use the I/O systems more efficiently it is necessary to know its performance capacity to determine if it fulfills applications' I/O requirements. Evaluating the I/O system performance capacity is difficult due to the diversity of I/O architectures and the complexity of its I/O software stack. Furthermore, parallel scientific applications have different behavior depending on their access patterns. Then, it is necessary to have some method to evaluate the I/O subsystem capacity taking into account the applications access patterns without executing the application in each I/O subsystem.Here, we propose a methodology to evaluate the I/O subsystem performance capacity through an I/O model of the parallel application independent of the I/O subsystem. This I/O model is composed of I/O phases representing "where" and "when" the I/O operations are performed into application logic.This approach encompasses the I/O subsystem evaluation at I/O library level for the application I/O model. The I/O phases are replicated by benchmark IOR which is executed in the target subsystem. This approach was used to estimate the I/O time of an application in different subsystems. The results show an relative error of estimation lower than 10%. This approach was also utilized to select the I/O subsystem that provide less I/O time for the application.

Section: Related Workmentioning

confidence: 99%

Modeling Parallel Scientific Applications through their Input/Output Phases

Méndez

2012 IEEE International Conference on Cluster Computing Workshops

Luque

2012

Self Cite

“…In [36,38,37], a different approach is proposed. Is based both in the fact that most parallel application have a repetitive behavior and that only partial executions are needed to estimate the total performance of an application.…”

Section: Application Measurementmentioning

confidence: 99%

“…The developed algorithm is inspired by Lamport's algorithm. Once we have the logical trace, it is processed using a technique that searches for similarity [37], to identify and extract the most relevant event sequences (phases) and assign them a weight based on the number of times they occur.…”

Section: Conceptual Basismentioning

confidence: 99%

Including the Workload Effect in the Parallel Program Signature

Canillas

Wong

2011 IEEE International Conference on High Performance Computing and Communications

et al. 2011

This work is dedicated to my parents brothers and to the love of my life. I hope I can give you back in some way all the love received from all of you. AbstractPerformance prediction and application behavior modeling have been the subject of extensive research that aim to estimate applications performance with an acceptable precision. A novel approach to predict the performance of parallel applications is based in the concept of Parallel Application Signatures [36] that consists in extract an application most relevant parts (phases) and the number of times they repeat (weights). Executing these phases in a target machine and multiplying its exeuction time by its weight an estimation of the application total execution time can be made.One of the problems is that the performance of an application depends on the program workload. Every type of workload affects differently how an application performs in a given system and so affects the signature execution time.Since the workloads used in most scientific parallel applications have dimensions and data ranges well known and the behavior of these applications are mostly deterministic, a model of how the programs workload affect its performance can be obtained. We create a new methodology to model how a program's workload affect the parallel application signature.Using regression analysis we are able to generalize each phase time execution and weight function to predict an application performance in a target system for any type of workload within predefined range. We validate our methodology using a synthetic program, benchmarks applications and well known real scientific applications. ResumenLa predicción del rendimiento y el modelado del comportamiento de las aplicaciones son tópicos ampliamente estudiados y se cuentan con numerosos trabajos de investigación que pretenden estimar el rendimiento de la aplicaciones con una precisión aceptable. Un nuevo enfoque para predecir enfoque para predecir el rendimiento de aplicaciones paralelas es el basado en el concepto de las firmas de aplicaciones paralelas [36] que consiste en extraer las partes mas relevantes de una aplicación (fases) y el numero de veces que se repiten (pesos). Ejecutando estas fases en una maquina destino y multiplicando su tiempo de ejecución por su peso, se puede obtener una estimación del tiempo total de ejecución de la aplicación.Uno de los problemas es que el rendimiento de una aplicación depende de la carga de trabajo de esta. Cada tipo de carga de trabajo afecta de manera distinta el rendimiento que tiene una aplicación en un sistema determinado y por lo tanto el tiempo de ejecución de la firma.Dado que las cargas de trabajo de la mayoría de las aplicaciones científicas paralelas, tienen dimensiones y rango de datos bien conocidos y que el comportamiento de estas aplicaciones es generalmente determinista, se puede obtener un modelo de como la carga de trabajo de un programa afecta su rendimiento. Hemos creado una nueva metodología para modelar como la carga de trabajo de un programa afecta a la f...

“…PAS2P [4] instruments a MPI program and executes parallel programs in a base machine, producing a trace log. The collected data is used to characterize the computation and communication behavior of the program.…”

Section: ) Robust Statementioning

confidence: 99%

“…In this work we extend our previous approach by combining the methodology with PAS2P (Parallel Application Signature for Performance Prediction) [4]. PAS2P allows the prediction of the time needed to execute a parallel program based on Message Passing Interface (MPI) by running a small fraction of the program (its signature) and extrapolating the rest of the program execution based on a previously made analysis (briefly described in section III).…”

Section: Introductionmentioning

confidence: 99%

Predicting robustness against transient faults of MPI based programs

Gramacho

Wong

et al. 2016

IJCSE

Self Cite

Abstract-The evaluation of a program's behavior in the presence of transient faults is often a very time consuming work. In order to achieve significant data, thousands of executions are normally required and each execution will have the significant overhead of the fault injection environment. Our previously published methodology reduced significantly the time needed to evaluate the robustness of a program execution by exhaustively analyzing its basic blocks trace instead of using fault injection. In this paper we present an even forward improvement in the evaluation time of parallel programs robustness against transient faults by combining our methodology with PAS2P -a method that strives to describe an application based on its messagepassing activity. The combination of our approach and PAS2P allowed us to predict the robustness of larger parallel programs, reducing in some cases in more than 20 times the time needed to calculate the robustness while obtaining a robustness prediction error of less than 4%.