As optimization problems become more complicated and extensive, parameterization becomes complex,... more As optimization problems become more complicated and extensive, parameterization becomes complex, resulting in a difficult task requiring significant amounts of time and resources. In this paper we propose a heuristic search algorithm we call MCSA (Montecarlo-Clustering Search Algorithm), which is based on Montecarlo sampling, and a clustering strategy involving two techniques. Our objective is to apply MCSA, an inherently stochastic method, to address optimization problems. To assess its performance, we conducted an evaluation using classical benchmark optimization functions. Additionally, we leveraged the CEC2017 benchmark suite to comprehensively evaluate the algorithm, highlighting the pivotal role of the Exploration stage in our methodology. Subsequently, we extended our methodology to tackle a practical combinatorial problem, the Knapsack problem. This NP-Hard problem holds significant realworld applications in resource allocation, scheduling, planning, logistics, and more. Our contributions lie in parameterizing the Knapsack Problem to align with MCSA's parameters for reference indicators adjustment and achieving high-quality solutions, surpassing 90% in comparison to exhaustive methods such as branch and bound.
Currently, there is an increasing interest about the cloud platform by the High Performance Compu... more Currently, there is an increasing interest about the cloud platform by the High Performance Computing (HPC) community, and the Parallel Input/Output (I/O) for High Performance Systems is not an exception. However, in cloud platform increase the number of parameters that user can select to I/O system. For this reason, we propose a methodology to help the user to select a configuration.
2017 17th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGRID), 2017
Performance evaluation of parallel applications plays an important role in High Performance Compu... more Performance evaluation of parallel applications plays an important role in High Performance Computing (HPC). This is also applied to parallel I/O performance evaluation, which requires understanding the I/O pattern of the application and having knowledge about the performance capacity of the HPC I/O system. In this paper, we present a methodology to evaluate the I/O performance of parallel applications based on the I/O severity degree. We define the I/O severity concept taking into account the I/O requirements of a parallel application, the mapping of I/O processes and the configuration of the I/O subsystem. Requirements are expressed in units denominated I/O phases, which are defined using the temporal and spatial pattern of different files of the application. Our approach is applied to the I/O kernels of scientific applications such as S3DIO, FLASH-IO and BT-IO on the SuperMUC supercomputer. Experimental results show that our methodology allows us to identify if a parallel applica...
2017 International Conference on High Performance Computing & Simulation (HPCS), 2017
Analyzing and understanding an application's Input/Output (I/O) access patterns provides key ... more Analyzing and understanding an application's Input/Output (I/O) access patterns provides key information to gain insight into how the behavior of an application affects its performance in different systems. In this paper, we propose a portable model to represent the I/O behavior of parallel applications that use serial I/O libraries, as part of a more holistic model for I/O of parallel applications. The model allows actions such as replication of the application behavior on different HPC systems and evaluation of the I/O performance without running the real application. In this paper, we evaluate the portability of the proposed model for MADbench2 and ABYSS- P in four HPC systems. We analyze the impact of the parallel system configuration in a Cloud environment for the ABYSS-P application by using the proposed I/O behavior model.
Current parallel scientific applications generate a huge amount of data that must be managed effi... more Current parallel scientific applications generate a huge amount of data that must be managed efficiently for the HPC storage systems. However, the I/O performance depends on the application I/O behavior and the configuration of the underlying I/O system. To understand the I/O behavior in the software stack and its impact on the I/O operations defined in the application logic, we propose a design framework named PIOM-PX, which allows to define an I/O behavior model based on the I/O phases of HPC applications at POSIX-IO level. We validate our framework using the IOR benchmark for four I/O patterns and we analyze the I/O behavior of NAS BT-IO.
Data-Intensive scientific applications use parallel Input/Output (I/O) software to access files. ... more Data-Intensive scientific applications use parallel Input/Output (I/O) software to access files. A tool capable of predicting the behavior of this kind of applications in High Performance Computing (HPC) would be very useful for parallel application developers. On the other hand, Agent-Based Modeling and Simulation (ABMS) has been used to model and simulate problems and complex systems for various science fields. This work presents a preliminary analysis to model a component of the I/O system in HPC using ABMS. Keywords–Agent-Based Modeling and Simulation (ABMS); HPC-I/O System; Parallel File System.
2017 International Conference on High Performance Computing & Simulation (HPCS), 2017
Handling faults is a growing concern in HPC; higher error rates, larger detection intervals and s... more Handling faults is a growing concern in HPC; higher error rates, larger detection intervals and silent faults are expected in the future. It is projected that, in exascale systems, errors will occur several times a day, and they will propagate to generate errors that will range from process crashes to corrupted results because of undetected errors. In this article, we propose a methodology that improves system reliability against transient faults, when running parallel message-passing applications. The proposed solution, based on process replication, has the goal of helping programmers and users of parallel scientific applications to achieve reliable executions with correct results. This work presents a characterization of the strategy, defining its behavior in the presence of faults and modeling the temporal costs of employing it. As a result, we show its efficacy and viability to tolerate transient faults in HPC systems.
In recent years, the use of public cloud platforms as infrastructure has been gaining popularity ... more In recent years, the use of public cloud platforms as infrastructure has been gaining popularity in many scientific areas and High Performance Computing (HPC) is no exception. These kinds of platforms can be used by system administrators as Test-Bed systems for evaluating and detecting performance inefficiencies in the I/O subsystem, and for taking decisions about the configuration parameters that have influence on the performance of an application, without compromising the performance of the production HPC system. In this paper, we propose a methodology to evaluate parallel applications by using virtual clusters as a test system. Our experimental validation indicates that virtual clusters are a quick and easy solution for system administrators, for analyzing the impact of the I/O system on the I/O kernels of the parallel applications and for taking performance decisions in a controlled environment.
Resumen Este art́ıculo presenta un método de optimización v́ıa simulación que permite mejorar la ... more Resumen Este art́ıculo presenta un método de optimización v́ıa simulación que permite mejorar la predicción de un simulador computacional de cauces de ŕıos aprovechando los valiosos aportes del cómputo paralelo y distribuido. El problema a tratar para lograr este objetivo es la incertidumbre en los parámetros de entrada al modelo, pero buscar parámetros de simulación óptimos es un gran desaf́ıo, ya que el tamaño del espacio de búsqueda hace que sea un problema combinatorio con un alto costo de cómputo. La solución propuesta utiliza técnicas de optimización en el análisis de simulaciones, reduciendo el espacio de búsqueda e identificando regiones que permitan encontrar la configuración óptima en tiempos razonables. Se obtiene una muy buena aceleración del procesamiento que es potenciado al implementar el método con técnicas del cómputo de alto rendimiento (HPC).
Handling faults is a growing concern in HPC. In future exascale systems, it is projected that sil... more Handling faults is a growing concern in HPC. In future exascale systems, it is projected that silent undetected errors will occur several times a day, increasing the occurrence of corrupted results. In this article, we propose SEDAR, which is a methodology that improves system reliability against transient faults when running parallel message-passing applications. Our approach, based on process replication for detection, combined with different levels of checkpointing for automatic recovery, has the goal of helping users of scientific applications to obtain executions with correct results. SEDAR is structured in three levels: (1) only detection and safestop with notification; (2) recovery based on multiple system-level checkpoints; and (3) recovery based on a single valid user-level checkpoint. As each of these variants supplies a particular coverage but involves limitations and implementation costs, SEDAR can be adapted to the needs of the system. In this work, a description of the methodology is presented and the temporal behavior of employing each SEDAR strategy is mathematically described, both in the absence and presence of faults. A model that considers all the fault scenarios on a test application is introduced to show the validity of the detection and recovery mechanisms. An overhead evaluation of each variant is performed with applications involving different communication patterns; this is also used to extract guidelines about when it is beneficial to employ each SEDAR protection level. As a result, we show its efficacy and viability to tolerate transient faults in target HPC environments.
IEEE Transactions on Parallel and Distributed Systems, 2018
Executing message-passing parallel applications on a large number of resources in an efficient wa... more Executing message-passing parallel applications on a large number of resources in an efficient way is not a trivial task. Due to the complex interaction between the parallel applications and the HPC system, many applications may suffer performance inefficiencies when they scale. To achieve an efficient use of these large-scale systems using thousands of cores, a point to consider before executing an application is to know its behavior in the system. In this work, we propose a novel methodology called P3S (Prediction of Parallel Program Scalability), which allows us to analyze and predict the scalability of message-passing applications on a given system. The methodology strives to use a bounded analysis time, and a reduced set of resources to predict the application behavior for large-scale. The experimental validation proves that the P3S is able to predict the application scalability with an average accuracy greater than 95 percent using a reduced set of resources.
Cloud Computing offers the possibility of computing resources, allowing remote access to software... more Cloud Computing offers the possibility of computing resources, allowing remote access to software, storage and data processing through the Internet. Infrastructures as a Service (IaaS), it is a flexible space which can be used as an experimental environment, in which experiments can be carried out similar to a real environment, such as in a cluster can be carried out. Before making installations and changes in a production cluster or select resource in the cloud, it is important to analyze the impact of this change. For this reason we propose using the cloud to carry out the study of previous viability. In this paper, we observe the viability of using the cloud to analyze the behavior of the Checkpoint as one of the Fault Tolerance strategies, establishing the differences that exist in the information generated in a real environment (cluster) and a virtual environment (cloud). The results obtained show that due to the variability of the cloud, the impact on the benefits cannot be an...
High performance computing has become a fundamental technology essential for computer simulation.... more High performance computing has become a fundamental technology essential for computer simulation. Modelling and computational simulation provide powerful tools which enable flood event forecasting. In order to reduce flood damage, we have developed a methodology focused on enhancing a flood simulator minimizing the number of errors between simulated and observed results by using a two-phase optimization methodology via simulation. In this research, we implemented this approach to find the best solution or adjusted set of simulator input parameters. As a result of this, we achieved an improvement of up to 14 % which, for example, represents a significant difference of 0.5–1 m of water level along whole Parana River basin. In order to find the adjusted set of input parameters, we reduced the search space using a Monte Carlo + clustering K-Means method. Therefore, an exhaustive search over the reduced search space led us to get a “good solution”. In summary, we propose add an improvement process on the classical computer model output to improve model quality.
Las operaciones de Entrada/Salida en los centros de supercomputación siguen siendo un cuello de b... more Las operaciones de Entrada/Salida en los centros de supercomputación siguen siendo un cuello de botella para el procesamiento en paralelo. Esto provoca que la capacidad de procesamiento de los distintos nodos que componen estos centros se vea limitada. Por este motivo, en el presente trabajo se propone estudiar y evaluar el rendimiento de las principales operaciones paralelas de E/S. Las ejecuciones se realizan utilizando dos tipos de clústeres (fı́sico y virtual), empleando benchmarks especı́ficos de E/S y sistemas de archivos. Además, se propone diseñar, desarrollar e incorporar pequeñas secciones de código a algunas de las funciones más importantes invocadas durante la ejecución de los benchmarks, de manera de obtener medidas de tiempo para cuantificar la ejecución en los distintos puntos del sistema de software de E/S.
1 Instituto de Ciencias, Universidad Nacional de G eneral Sarmiento, Bs. As., Argentina agaudi@un... more 1 Instituto de Ciencias, Universidad Nacional de G eneral Sarmiento, Bs. As., Argentina agaudi@ungs .edu.ar 2 Instituto de Investigación en Informática LIDI ( III-LIDI), Universidad Nacional de La Plata, Bs. As ., Argentina – {degiusti, mnaiouf}@lidi.info.unlp.edu.ar 3 Depto. de Arquitectura de Computadores y Sistema s Operativos, Universitat Autónoma de Barcelona, 08 193 Bellaterra (Barcelona) Españ[email protected]
As optimization problems become more complicated and extensive, parameterization becomes complex,... more As optimization problems become more complicated and extensive, parameterization becomes complex, resulting in a difficult task requiring significant amounts of time and resources. In this paper we propose a heuristic search algorithm we call MCSA (Montecarlo-Clustering Search Algorithm), which is based on Montecarlo sampling, and a clustering strategy involving two techniques. Our objective is to apply MCSA, an inherently stochastic method, to address optimization problems. To assess its performance, we conducted an evaluation using classical benchmark optimization functions. Additionally, we leveraged the CEC2017 benchmark suite to comprehensively evaluate the algorithm, highlighting the pivotal role of the Exploration stage in our methodology. Subsequently, we extended our methodology to tackle a practical combinatorial problem, the Knapsack problem. This NP-Hard problem holds significant realworld applications in resource allocation, scheduling, planning, logistics, and more. Our contributions lie in parameterizing the Knapsack Problem to align with MCSA's parameters for reference indicators adjustment and achieving high-quality solutions, surpassing 90% in comparison to exhaustive methods such as branch and bound.
Currently, there is an increasing interest about the cloud platform by the High Performance Compu... more Currently, there is an increasing interest about the cloud platform by the High Performance Computing (HPC) community, and the Parallel Input/Output (I/O) for High Performance Systems is not an exception. However, in cloud platform increase the number of parameters that user can select to I/O system. For this reason, we propose a methodology to help the user to select a configuration.
2017 17th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGRID), 2017
Performance evaluation of parallel applications plays an important role in High Performance Compu... more Performance evaluation of parallel applications plays an important role in High Performance Computing (HPC). This is also applied to parallel I/O performance evaluation, which requires understanding the I/O pattern of the application and having knowledge about the performance capacity of the HPC I/O system. In this paper, we present a methodology to evaluate the I/O performance of parallel applications based on the I/O severity degree. We define the I/O severity concept taking into account the I/O requirements of a parallel application, the mapping of I/O processes and the configuration of the I/O subsystem. Requirements are expressed in units denominated I/O phases, which are defined using the temporal and spatial pattern of different files of the application. Our approach is applied to the I/O kernels of scientific applications such as S3DIO, FLASH-IO and BT-IO on the SuperMUC supercomputer. Experimental results show that our methodology allows us to identify if a parallel applica...
2017 International Conference on High Performance Computing & Simulation (HPCS), 2017
Analyzing and understanding an application's Input/Output (I/O) access patterns provides key ... more Analyzing and understanding an application's Input/Output (I/O) access patterns provides key information to gain insight into how the behavior of an application affects its performance in different systems. In this paper, we propose a portable model to represent the I/O behavior of parallel applications that use serial I/O libraries, as part of a more holistic model for I/O of parallel applications. The model allows actions such as replication of the application behavior on different HPC systems and evaluation of the I/O performance without running the real application. In this paper, we evaluate the portability of the proposed model for MADbench2 and ABYSS- P in four HPC systems. We analyze the impact of the parallel system configuration in a Cloud environment for the ABYSS-P application by using the proposed I/O behavior model.
Current parallel scientific applications generate a huge amount of data that must be managed effi... more Current parallel scientific applications generate a huge amount of data that must be managed efficiently for the HPC storage systems. However, the I/O performance depends on the application I/O behavior and the configuration of the underlying I/O system. To understand the I/O behavior in the software stack and its impact on the I/O operations defined in the application logic, we propose a design framework named PIOM-PX, which allows to define an I/O behavior model based on the I/O phases of HPC applications at POSIX-IO level. We validate our framework using the IOR benchmark for four I/O patterns and we analyze the I/O behavior of NAS BT-IO.
Data-Intensive scientific applications use parallel Input/Output (I/O) software to access files. ... more Data-Intensive scientific applications use parallel Input/Output (I/O) software to access files. A tool capable of predicting the behavior of this kind of applications in High Performance Computing (HPC) would be very useful for parallel application developers. On the other hand, Agent-Based Modeling and Simulation (ABMS) has been used to model and simulate problems and complex systems for various science fields. This work presents a preliminary analysis to model a component of the I/O system in HPC using ABMS. Keywords–Agent-Based Modeling and Simulation (ABMS); HPC-I/O System; Parallel File System.
2017 International Conference on High Performance Computing & Simulation (HPCS), 2017
Handling faults is a growing concern in HPC; higher error rates, larger detection intervals and s... more Handling faults is a growing concern in HPC; higher error rates, larger detection intervals and silent faults are expected in the future. It is projected that, in exascale systems, errors will occur several times a day, and they will propagate to generate errors that will range from process crashes to corrupted results because of undetected errors. In this article, we propose a methodology that improves system reliability against transient faults, when running parallel message-passing applications. The proposed solution, based on process replication, has the goal of helping programmers and users of parallel scientific applications to achieve reliable executions with correct results. This work presents a characterization of the strategy, defining its behavior in the presence of faults and modeling the temporal costs of employing it. As a result, we show its efficacy and viability to tolerate transient faults in HPC systems.
In recent years, the use of public cloud platforms as infrastructure has been gaining popularity ... more In recent years, the use of public cloud platforms as infrastructure has been gaining popularity in many scientific areas and High Performance Computing (HPC) is no exception. These kinds of platforms can be used by system administrators as Test-Bed systems for evaluating and detecting performance inefficiencies in the I/O subsystem, and for taking decisions about the configuration parameters that have influence on the performance of an application, without compromising the performance of the production HPC system. In this paper, we propose a methodology to evaluate parallel applications by using virtual clusters as a test system. Our experimental validation indicates that virtual clusters are a quick and easy solution for system administrators, for analyzing the impact of the I/O system on the I/O kernels of the parallel applications and for taking performance decisions in a controlled environment.
Resumen Este art́ıculo presenta un método de optimización v́ıa simulación que permite mejorar la ... more Resumen Este art́ıculo presenta un método de optimización v́ıa simulación que permite mejorar la predicción de un simulador computacional de cauces de ŕıos aprovechando los valiosos aportes del cómputo paralelo y distribuido. El problema a tratar para lograr este objetivo es la incertidumbre en los parámetros de entrada al modelo, pero buscar parámetros de simulación óptimos es un gran desaf́ıo, ya que el tamaño del espacio de búsqueda hace que sea un problema combinatorio con un alto costo de cómputo. La solución propuesta utiliza técnicas de optimización en el análisis de simulaciones, reduciendo el espacio de búsqueda e identificando regiones que permitan encontrar la configuración óptima en tiempos razonables. Se obtiene una muy buena aceleración del procesamiento que es potenciado al implementar el método con técnicas del cómputo de alto rendimiento (HPC).
Handling faults is a growing concern in HPC. In future exascale systems, it is projected that sil... more Handling faults is a growing concern in HPC. In future exascale systems, it is projected that silent undetected errors will occur several times a day, increasing the occurrence of corrupted results. In this article, we propose SEDAR, which is a methodology that improves system reliability against transient faults when running parallel message-passing applications. Our approach, based on process replication for detection, combined with different levels of checkpointing for automatic recovery, has the goal of helping users of scientific applications to obtain executions with correct results. SEDAR is structured in three levels: (1) only detection and safestop with notification; (2) recovery based on multiple system-level checkpoints; and (3) recovery based on a single valid user-level checkpoint. As each of these variants supplies a particular coverage but involves limitations and implementation costs, SEDAR can be adapted to the needs of the system. In this work, a description of the methodology is presented and the temporal behavior of employing each SEDAR strategy is mathematically described, both in the absence and presence of faults. A model that considers all the fault scenarios on a test application is introduced to show the validity of the detection and recovery mechanisms. An overhead evaluation of each variant is performed with applications involving different communication patterns; this is also used to extract guidelines about when it is beneficial to employ each SEDAR protection level. As a result, we show its efficacy and viability to tolerate transient faults in target HPC environments.
IEEE Transactions on Parallel and Distributed Systems, 2018
Executing message-passing parallel applications on a large number of resources in an efficient wa... more Executing message-passing parallel applications on a large number of resources in an efficient way is not a trivial task. Due to the complex interaction between the parallel applications and the HPC system, many applications may suffer performance inefficiencies when they scale. To achieve an efficient use of these large-scale systems using thousands of cores, a point to consider before executing an application is to know its behavior in the system. In this work, we propose a novel methodology called P3S (Prediction of Parallel Program Scalability), which allows us to analyze and predict the scalability of message-passing applications on a given system. The methodology strives to use a bounded analysis time, and a reduced set of resources to predict the application behavior for large-scale. The experimental validation proves that the P3S is able to predict the application scalability with an average accuracy greater than 95 percent using a reduced set of resources.
Cloud Computing offers the possibility of computing resources, allowing remote access to software... more Cloud Computing offers the possibility of computing resources, allowing remote access to software, storage and data processing through the Internet. Infrastructures as a Service (IaaS), it is a flexible space which can be used as an experimental environment, in which experiments can be carried out similar to a real environment, such as in a cluster can be carried out. Before making installations and changes in a production cluster or select resource in the cloud, it is important to analyze the impact of this change. For this reason we propose using the cloud to carry out the study of previous viability. In this paper, we observe the viability of using the cloud to analyze the behavior of the Checkpoint as one of the Fault Tolerance strategies, establishing the differences that exist in the information generated in a real environment (cluster) and a virtual environment (cloud). The results obtained show that due to the variability of the cloud, the impact on the benefits cannot be an...
High performance computing has become a fundamental technology essential for computer simulation.... more High performance computing has become a fundamental technology essential for computer simulation. Modelling and computational simulation provide powerful tools which enable flood event forecasting. In order to reduce flood damage, we have developed a methodology focused on enhancing a flood simulator minimizing the number of errors between simulated and observed results by using a two-phase optimization methodology via simulation. In this research, we implemented this approach to find the best solution or adjusted set of simulator input parameters. As a result of this, we achieved an improvement of up to 14 % which, for example, represents a significant difference of 0.5–1 m of water level along whole Parana River basin. In order to find the adjusted set of input parameters, we reduced the search space using a Monte Carlo + clustering K-Means method. Therefore, an exhaustive search over the reduced search space led us to get a “good solution”. In summary, we propose add an improvement process on the classical computer model output to improve model quality.
Las operaciones de Entrada/Salida en los centros de supercomputación siguen siendo un cuello de b... more Las operaciones de Entrada/Salida en los centros de supercomputación siguen siendo un cuello de botella para el procesamiento en paralelo. Esto provoca que la capacidad de procesamiento de los distintos nodos que componen estos centros se vea limitada. Por este motivo, en el presente trabajo se propone estudiar y evaluar el rendimiento de las principales operaciones paralelas de E/S. Las ejecuciones se realizan utilizando dos tipos de clústeres (fı́sico y virtual), empleando benchmarks especı́ficos de E/S y sistemas de archivos. Además, se propone diseñar, desarrollar e incorporar pequeñas secciones de código a algunas de las funciones más importantes invocadas durante la ejecución de los benchmarks, de manera de obtener medidas de tiempo para cuantificar la ejecución en los distintos puntos del sistema de software de E/S.
1 Instituto de Ciencias, Universidad Nacional de G eneral Sarmiento, Bs. As., Argentina agaudi@un... more 1 Instituto de Ciencias, Universidad Nacional de G eneral Sarmiento, Bs. As., Argentina agaudi@ungs .edu.ar 2 Instituto de Investigación en Informática LIDI ( III-LIDI), Universidad Nacional de La Plata, Bs. As ., Argentina – {degiusti, mnaiouf}@lidi.info.unlp.edu.ar 3 Depto. de Arquitectura de Computadores y Sistema s Operativos, Universitat Autónoma de Barcelona, 08 193 Bellaterra (Barcelona) Españ[email protected]
Uploads
Papers by emilio luque