A. Waheed and D. Rover. A structured approach to instrumentation system development and evaluation. In Proc. Supercomputing 95, pages 1--1, 1995.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....[13, 4, 14] produces streams of events that represent observed application state; the user analyzes these streams judge target system behavior. Motivation. Application instrumentation is a primary component of any application specific monitoring system. Across a wide variety of systems [23, 9], instrumentation support may be described as follows. 1) The developer chooses basic instrumentation mechanisms and annotates the source code. 2) At runtime, the monitoring system receives data from this instrumentation. 3) The monitoring server has limited runtime control over the ....

....monitoring system has its own nonportable instrumentation library. In this respect, monitoring assertions could provide a common instrumentation tool for both application developers and tool developers. While these earlier instrumentation techniques propose various instrumentation mechanisms [19, 5, 16, 23], they focus primarily on the design of the monitoring system itself. Research contributions. The novel idea of this work is to generalize instrumentation so that the binding of specific monitoring mechanisms to instrumentation points in target applications is delayed until more is known about ....

[Article contains additional citation context not shown here]

A. Waheed and D. Rover. A structured approach to instrumentation system development and evaluation. In Proc. Supercomputing 95, pages 1--1, 1995.

.... approach to the instrumentation system of the Paradyn parallel performance measurement tool [20] This structured approach is based on modeling and simulating the IS to answer several what if questions regarding possible configurations and scheduling policies to collect and manage runtime data [28]. The Paradyn IS is enhanced based on the initial feedback provided by the modeling and simulation process. Measurement based testing validates the simulation based results and shows more than 60 reduction in the data collection overheads for two applications from the NAS benchmark suite executed ....

.... Reference: http: www.att.com att tj and [6] 24 This study of Paradyn s IS follows previous work by Waheed and Rover to view the IS as enabling technology, or middleware [2] and to establish an approach for characterizing, evaluating, and understanding IS operation, including its overheads [28]. This approach emphasizes a separation of the high level tool requirement and usability issues from the low level design and test issues. We applied this two level approach for modeling and evaluating the Paradyn IS. Acknowledgments We would like to acknowledge the contribution of Bart Miller ....

Waheed, A. and Diane T. Rover, "A Structured Approach to Instrumentation System Development and Evaluation," Proceedings of Supercomputing `95, San Diego, California, Dec. 3--8, 1995.

....data, the user s program is modified to recognize certain events and to record them for display. Note that collecting and displaying data will perturb the user s program [SM93] 2.2. 1 Data Collection and Data Transformation To collect the data we describe an instrumentation system or IS [WR95] An IS deals with the collection and management of performance data. A monitor is a tool used to observe the activities of a system. In general, monitors observe the performance of systems, collect performance statistics, analyze data, and display results. Monitors are classified with respect ....

....refers to the process of writing out data from the main memory to the tracefile. Mainly there are two ways to do so: 2.4. PROFILING METHODOLOGIES 15 ffl FOF: Flushing whenever a buffer on the node is full, ffl FAOF: Flush all the buffers when one of the node buffers is full. Detailed studies [WR95] have shown that the second strategy is the more efficient one. Two things were investigated: ffl Trace stopping time: the time to fill a buffer for flushing, ffl Flushing frequency: ratio of flushes to data collected on one node. It is shown in the study that when one buffer fills, the ....

A. Waheed and D. Rover. A Structured Approach to Instrumentation System Development and Evaluation. In Proceedings SUPERCOMPUTING '95, San Diego, CA, November 1995.

....the instrumented program produces a tracefile that is postprocessed using a graphical interface. Once collected, it is possible to look at the trace data over and over again, always having the same result. The collection and management of performance data is done by an instrumentation system [WR95] It is responsible for inserting the trace calls into the user s program, which can be done statically before compiling like in PICL [GHPW90] or dynamically at run time like in Paradyn [MCC 95] The trace calls, along with the modified communication library, generate entries that go into ....

....of writing out data from the main memory to the tracefile. Traditionally, there have been two ways to do so: Flushing whenever a buffer on the node is full, and flush all the buffers when one of the node buffers is full. Detailed studies have shown that the second strategy is more efficient [WR95] Unfortunately, it is much harder to implement because asynchronous barriers are required. Therefore most existing tools rely on the first strategy. In this paper we introduce a third approach: flush on barriers, which attempts to combine the benefits of both strategies. Collecting all the trace ....

A. Waheed and D. Rover. A Structured Approach to Instrumentation System Development and Evaluation. In Proceedings SUPERCOMPUTING '95, San Diego, CA, November 1995.

....on the other hand, will only record a sampling of the information. We will discuss statistical approaches later, but for now will focus on the traditional online monitoring which logs all relevant information. The collection and management of performance data is done by an instrumentation system [WR95] It is responsible for inserting the trace calls into the user s program, which can be done statically before compiling like in PICL [GHPW90] or dynamically at run time like in Paradyn [MCC 95] The trace calls, along with the modified communication library, generate entries that go into ....

....of writing out data from the main memory to the tracefile. Traditionally, there have been two ways to do so: Flushing whenever a buffer on the node is full, and flush all the buffers when one of the node buffers is full. Detailed studies have shown that the second strategy is more efficient [WR95] Unfortunately, it is much harder to implement because asynchronous barriers are required. Therefore most existing tools rely on the first strategy. In this paper we introduce a third approach: flush on barriers, which attempts to combine the benefits of both strategies. Collecting all the trace ....

A. Waheed and D. Rover. A Structured Approach to Instrumentation System Development and Evaluation. In Proceedings SUPERCOMPUTING '95, San Diego, CA, November 1995.

.... system that is found in performance analysis and program steering environments [23] We apply a structured approach that is based on modeling and simulating the IS to answer several what if questions regarding possible configurations and scheduling policies to collect and manage runtime data [28]. The Paradyn IS is enhanced based on the initial feedback provided by the modeling and simulation process. Measurement based testing validates the simulation based results and shows more than 60 percent reduction in the data collection overheads for two applications from the NAS benchmark suite ....

.... steering tool and measure the IS performance [9] This study of Paradyn s IS follows previous work by Waheed and Rover to view the IS as enabling technology, or middleware [2] and to establish an approach for characterizing, evaluating, and understanding IS operation, including its overheads [28]. This approach emphasizes a separa (a ) b) Fig. 16. Paradyn IS testing results related to (a) Paradyn daemon process, and (b) main Paradyn process. TABLE 4 IS SERVICES USED BY TOOLS TO SUPPORT A RANGE OF FUNCTIONS Functionality Representative Tools Description of Key Services ....

# A. Waheed and D.T. Rover, "A Structured Approach to Instrumentation System Development and Evaluation," Proc. Supercomputing `95, San Diego, Calif., Dec. 1995.

.... specifically consider the data collection and management components in state of the art parallel tool environments [35] We favor a structured approach to plan, design, model, evaluate, and implement an IS to address specific requirements imposed by the parallel tool environment that it supports [36]. This paper presents a model to account for the overheads of an IS in a distributed computing environment. The case study presented in this paper focuses on the Paradyn environment [24] and elaborates on the IS design evaluation approach implemented early in the tool development cycle. In order ....

....Unix pipe Figure 17. Arrangement for measurement based validation. Pd) Appl. process) 18 program performance analysis and debugging tools. In this section, we introduce the IS development and evaluation approaches of various representative parallel tools. Additional details can be found in [36]. PICL and ParaGraph [9] have been used with several environments. PICL [5] is a portable library of efficient communication functions that also supports instrumentation. We have modeled and evaluated two performance data management policies supported by PICL in [35] AIMS (Automated ....

[Article contains additional citation context not shown here]

Waheed, A. and Diane T. Rover, "A Structured Approach to Instrumentation System Development and Evaluation," Technical Report, Department of Electrical Engineering, Michigan State University, April 1995.

....access to the above features at the highest level. Vista ISM. Vista is a C framework for development of domain specific instrumentation systems. An instrumentation system (IS) is defined as a set of modules and services that supports the collection and management of program performance data [23]. Figure 8 presents a generic instrumentation system model that guides the design of the Vista framework. It represents a majority of components found in extant ISs. The model defines three components of an IS that supports tool integration: 1) local instrumentation server (LIS) 2) ....

Waheed, A. and Diane T. Rover, "A Structured Approach to Instrumentation System Development and Evaluation," Proc. of Supercomputing `95, IEEE Computer Society Press, 1995 (CD-ROM).

No context found.

A. WAHEED and DIANE T. ROVER, "A Structured Approach to Instrumentation System Development and Evaluation," Proceedings of Supercomputing `95, (San Diego, California, Dec. 4--8, 1995).

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC