Abstract not found

Abstract not found

This paper presents a performance modeling methodology that is faster than traditional cycle-accurate simulation, more sophisticated than performance estimation based on system peak-performance metrics, and is shown to be effective on a class of High Performance Computing benchmarks. The method yields insight into the factors that affect performance on single-processor and parallel computers.

One purpose of the end-user tools described in this paper is to give users a graphical representation of performance information that has been gathered by instrumenting an application with the PAPI library. PAPI is a project that specifies a standard API for accessing hardware performance counters available on most modern microprocessors. These counters exist as a small set of registers that count "events", which are occurrences of specific signals and states related to a processor’s function. Monitoring these events facilitates correlation between the structure of source/object code and the efficiency of the mapping of that code to the underlying architecture. The perfometer tool developed by the PAPI project provides a graphical view of this information, allowing users to quickly see where performance bottlenecks are in their application. Only one function call has to be added by the user to their program to take advantage of perfometer. This makes it quick and simple to add and remove instrumentation from a program. Also, perfometer allows users to change the "event" they are monitoring. Add the ability to monitor parallel applications, set alarms and a Java front-end that can run anywhere, and this gives the user a powerful tool for quickly discovering where and why a bottleneck exists. A number of third-party tools for analyzing performance of message-passing and/or threaded programs have also incorporated support for PAPI so as to be able to display and analyze hardware counter data from their interfaces.

Cycle-accurate simulation is far too slow for modeling the expected performance of full parallel applications on large HPC systems. And just running an application on a system and observing wallclock time tells you nothing about why the application performs as it does (and is anyway impossible on yet-to-be-built systems). Here we present a framework for performance modeling and prediction that is faster than cycle-accurate simulation, more informative than simple benchmarking, and is shown useful for performance investigations in several dimensions.

Empirical performance evaluation of parallel systems and applications can generate significant amounts of performance data and analysis results from multiple experiments as performance is investigated and problems diagnosed. Hence, the management of performance information is a core component of performance analysis tools. To better support tool integration, portability, and reuse, there is a strong motivation to develop performance data management technology that can provide a common foundation for performance data storage, access, merging, and analysis. This paper presents the design and implementation of the Performance Data Management Framework (PerfDMF). PerfDMF addresses objectives of performance tool integration, interoperation, and reuse by providing common data storage, access, and analysis infrastructure for parallel performance profiles. PerfDMF includes an extensible parallel profile data schema and relational database schema, a profile query and analysis programming interface, and an extendible toolkit for profile import/export and standard analysis. We describe the PerfDMF objectives and architecture, give detailed explanation of the major components, and show examples of PerfDMF application.

Within a decade after John von Neumann and colleagues conducted the first experimental weather forecast on the ENIAC computer in the late 1940s, numerical models of the atmosphere became the foundation of modern-day weather forecasting and one of the driving application areas in computer science. This article describes research that is enabling a major shift toward dynamically adaptive responses to rapidly changing environmental conditions. 1521-9615/05/$20.00 © 2005 IEEE Copublished by the IEEE CS and the AIP Each year across the US, mesoscale weather events—flash floods, tornadoes, hail, strong winds, lightning, and localized winter storms—cause hundreds of

The Computational Grid [10] has recently been proposed for the implementation of high-performance applications using widely dispersed computational resources. The goal of a Computational Grid is to aggregate ensembles of shared, heterogeneous, and distributed resources (potentially controlled by separate organizations) to provide computational "power" to an application program. In this paper, we provide a toolkit for the development of Grid applications. The toolkit, called EveryWare, enables an application to draw computational power transparently from the Grid. The toolkit consists of a portable set of processes and libraries that can be incorporated into an application so that a wide variety of dynamically changing distributed infrastructures and resources can be used together to achieve supercomputer-like performance. We provide our experiences gained while building the EveryWare toolkit prototype and the first true Grid application. 1 Introduction Increasingly, the high-perform...

Abstract. LEAD is a large-scale effort to build a service-oriented infrastructure that allows atmospheric science researchers to dynamically and adaptively respond to weather patterns to produce better-than-real time predictions of tornadoes and other ”mesoscale ” weather events. In this paper we discuss an architectural framework that is forming our thinking about adaptability and give early solutions in workflow and monitoring. 7 1

elation with experimental or computational data. Finally, enabling effective remote interaction and visualization necessitates quality-ofservice (QoS) guarantees. Incorporating QoS into these hardware and software systems further increases their complexity. Historically, performance analysis has focused on monolithic applications executing on large, standalone, parallel systems. In such a domain, measurement, postmortem analysis, and code optimization suffice to eliminate performance bottlenecks and optimize applications. Most existing performance analysis systems---for example, SvPablo, 2 Medea, 3 and Paragraph 4 ---use only postmortem analysis. To tune the emerging distributed applications, however, a new generation of online performance measurement and optimization tools must adapt application behavior dynamically as resource availability changes. In addition to providing real-time adaptive control, new perform