B.P. Miller, M. Clark, J.K. Hollingsworth, S. Kierstead, S. Lim, and T. Torzewski. IPS-2: The Second Generation of a Parallel Program Measuremnt System. In IEEE Transactions on Parallel and Distributed Systems, volume 1, pages 206-217, April 1990.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....per second. Timing variability causes inconsistency in resultant curves. Therefore, the area under curves might be similar, but the shapes of curves will not necessarily be so. 6 Related Work Our curve fitting signature design was inspired by phase behavioral analysis of parallel codes [3, 4, 5, 6]. Phase behavioral analysis uses geometric curves to characterize performance as a function of time. Unlike our online signature approach, these automatic phase characterization algorithms must scan the entire trace several times. Prophesy [15] is the most similar work. It also uses curve fitting ....

Miller, B. P., Clark, M., Hollingsworth, J., Kierstead, S., Lim, S.-S., and Torzewski, T. IPS-2: The Second Generation of a Parallel Program Measurement System. IEEE Trans. Parallel and Distributed Systems , Vol. 1, No. 2, 1990, pp. 206--217

....multiprocessor. IPS is a performance measurement system for parallel and distributed programs that uses knowledge about the semantics of a program s structure to provide a large amount of easily accessible performance data and analysis techniques that guide programmers to performance bottlenecks [19]. IPS is based on the software instrumentation technique. Quartz is another tool for tuning parallel program performance on shared memory multiprocessors. The principal metric in Quartz is normalized processor time: the total processor time spent in each section of the code divided by the number ....

Barton P. Miller et. al.. IPS-2: The Second Generation of a Parallel Program Measurement System. In IEEE Transactions on Parallel and Distributed Systems, Vol. 1, No. 2, April 1990.

....and performance characterization. Consequently, there is much active research on mechanisms for efficient monitoring in all kinds of systems. The PSpec notion of a log is general enough to fit many existing monitoring systems with little work. For examples of recent work on monitoring, see [3, 12, 20, 24, 28, 32, 33, 38, 43]. General discussions of monitor design may be found in [11] and [18] and in other textbooks on performance measurement. Performance characterization includes work on defining performance metrics, on characterizing workloads, and on techniques for determining values In a system with hard or ....

Barton P. Miller, Morgan Clark, Steven Kierstead, and Sekl-See Lim. ips-2: The second generation of a parallel program measurement system. Computer Sciences Technical Report 783, University of Wisconsin--- Madison, Madison, Wisconsin, August 1988.

....applied to several application kernels on a MIMD system. The performance predictions and error bounds resulting from the level of resource specification are also discussed. 1. Introduction Performance evaluation is an active area of interest especially within the parallel systems community [18, 26, 34]. It is of interest to all parties involved including machine manufactures, algorithm designers, application developers, application users, and performance analysts. Performance evaluation strategies have been proposed and utilised to allow analysis to take place at various stages of application ....

....tuning, capacity planning, and when studying the effects of data dependent parameters. Monitoring tools are widely available that allow the measurement of the execution time of each node of the software execution graph. Some examples of monitoring tools that can be used for this purpose are IPS 2 [26], PABLO [35] and PEPS monitoring tool [24] Because of the available tools it is easy to perform performance studies using component timing as resource information and will result with high accuracy. However, component timing can only be used in the last stage of the software development when the ....

B.P. Miller, M. Clark, J. Hollingsworth, S. Kierstead, S.S. Lim, and T. Torewski, IPS-2: The second generation of a parallel program measurement system, IEEE Trans. on Parallel and Distributed Systems I (2) (1990) 206-217.

....monitoring system during the execution of the target application. Users can enable, disable, and reconfigure the monitoring system to suit their exploration of the application [19, 17, 9, 6] These interactive monitoring systems complement other forms of monitoring that use post mortem analysis [25, 18, 16, 2, 3], but they share with such systems the need to analyze and display monitoring data [20, 11, 12, 4] The interactive monitoring systems # NASA financially supported Vetter with a Graduate Student Researchers Program Fellowship while he was a Ph.D. candidate at Georgia Tech. This work was also ....

B. Miller, M. Clark, J. Hollingsworth, S. Kierstead, S.-S. Lim, and T. Torzewski. IPS-2: The second generation of a parallel program measurement system. IEEE Trans. Parallel and Distributed Systems, 1:206--217, 1990.

.... to understanding the details of Web transaction performance uses the method of critical path analysis [70] Critical path analysis is the natural tool for understanding distributed applications because it identifies the precise set of activities that determine an application s performance [51, 79, 131]. Prior to our work, CPA has not been applied to TCP transactions. In [105] Paxson discusses the notion of cause effect issues in TCP transactions, which is an idea that also underlies our approach. However, Paxson had difficulty fully developing this notion because he was interested in ....

B. Miller, M. Clark, J. Hollingsworth, S. Kierstead, S. Lim, and T. Torzewski. IPS-2: The second generation of a parallel program measurement system. IEEE Transactions on Parallel and Distributed Systems, 1(2):206--217, April 1990.

....different aspects of the measured values. On the next level process and processor states and interactions are displayed in detail over a time axis. On this level it is possible to identify the sequence of actions of the program that causes the efficiency losses detected before. The IPS 2 system [10] supports a stepwise refinement process in detecting and tracking down performance bottlenecks. It is possible to look at data for the complete system, individual nodes, processes, and procedures. Profile values can be produced to identify the processor states and nodes that cause performance ....

B. P. Miller, M. Clark, J. Hollingsworth, S. Kierstead, S. Lim, and T. Torzewski. IPS-2: The second generation of a parallel program measurement system. IEEE Transactions on Parallel and Distributed Systems, 1(2):206 -- 217, April 1990.

....values but no information on individual events. Detailed traces of system related events with accurate timestamps are more appropriate to understand e.g. why the synchronization of two threads fails in certain cases. 2 Related Work Over the last decade event based performance monitoring tools [5, 6, 8, 9] have shown their value for performance evaluation of complex software systems. Typically, preparing a system for an in depth monitoring means inserting sensors according to a certain monitoring or measurement model. Such a model consists of a (simplified) system model and a specification which ....

B. P. Miller at al., IPS-2: The Second Generation of a Parallel Program Measurement System, IEEE Trans. on Parallel and Distributed Systems, Vol. 1, No. 2, April 1990, pp. 206-217

....in Fig. 1 spends most of the time in function b( and suggest that function for optimization. This is shown in Fig. 2 where the functions a( to d( are found at the top; the other functions shown are only for internal use in the Solaris operating system. The tools and methods described in [3, 8, 18] have critical path analysis and will show that a( is the most beneficial to optimize. However, those tools and methods assume that there is only one thread (or process) scheduled on each processor. Our approach with the extended critical path generalizes the two extreme cases above. Not only ....

....the Extended Critical Path with One Thread per Processor We start to look at the extended critical path algorithm under the condition that the program is executed on a sufficient number of processors and thus no threads have to share a processor at any time. This is the same assumption used in [3, 8, 18]. We assume that explicitly synchronized events happen at the exact same time, such as one thread releasing another thread. This means that a thread starts to execute at exactly the same moment as the other thread releases it. This assumption is realistic, does not change the algorithm in ....

[Article contains additional citation context not shown here]

B. Miller, M. Clark, J. Hollingsworth, S. Kiersead, S.-S. Lim, and T. Torzewski, IPS-2: The second Generation of a Parallel Program Measurement System, IEEE Trans. Parallel Distrib. Systems 1,2 (April 1990), 206#217.

....multiprocessor. IPS is a performance measurement system for parallel and distributed programs that uses knowledge about the semantics of a program s structure to provide a large amount of easily accessible performance data and analysis techniques that guide programmers to performance bottlenecks [19]. IPS is based on the software instrumentation technique. Quartz is another tool for tuning parallel program performance on shared memory multiprocessors. The principal metric in Quartz is normalized processor time: the total processor time spent in each section of the code divided by the number ....

Barton P. Miller et. al.. IPS-2: The Second Generation of a Parallel Program Measurement System. In IEEE Transactions on Parallel and Distributed Systems, Vol. 1, No. 2, April 1990.

....latency. Our general approach is to study these transactions using the method of critical path analysis [25] Critical path analysis is the natural tool for understanding distributed applications because it identi es the precise set of activities that determine an application s performance [17, 27, 41]. The central observation of critical path analysis as applied to distributed systems is that only some of the component activities in a distributed application are responsible for the overall response time; many other activities may occur in parallel, so that their executions overlap each other, ....

Barton P. Miller, Morgan Clark, Je Hollingsworth, Steven Kierstead, Sek-See Lim, and Timothy Torzewski. IPS-2: The second generation of a parallel program measurement system. IEEE Transactions on Parallel and Distributed Systems, 1(2):206-217, April 1990.

....and there is no gain in performance but in overhead. 4. OTHER RESULTS Mohr affirms in [20] that times reported in literature for software monitoring and normal amount of event instructions (e.g. each entry and exit of each function) vary from 10 up to 500 . Examples are: Bart Miller, IPS 2 [19], 45 and TOPSYS [15] Munich, 200 . In [4] Beier shows that due to the measurements made by the software monitor, the program execution slows down about 2 . However, a software implementation of these mechanisms of the monitoring system reduces the complexity of the hardware monitor by about ....

.... of software instrumentation by using an initial profile to estimate and control the overhead it is introducing [10] Goldberg and Hennessy affirm that generally, Mtool s instrumentation increases program execution time by less than 10 while comparable existing tools (e.g. Quartz [1] IPS 2 [19]) increase it by 40 70 . Using hybrid monitoring and carefully placed event calls, these values could be reduced to 0.1 1 [20, 28, 29] 5. CONCLUSIONS Invasiveness is an important factor to be considered when measuring the performance of a parallel application. Invasiveness represents the ....

[Article contains additional citation context not shown here]

B. Miller, M. Clark, J. Hollingsworth, S. Kierstead, S. Lim, and T. Torzewski. IPS-2: The second generation of a parallel program measurement system. IEEE Transactions on Parallel and Distributed Systems, pages 206--217, April 1990.

....Their system specifies a standard, extensible format for event logs and supports a number of tools for summarizing and presenting information, including auralization and 3 D views on head mounted displays. Miller s group has also created a flexible system for visualizing performance data. IPS 2 [37, 18] uses two kinds of data: a standard trace log and a time histogram that notes the value of some performance metric as it varies over time. Users can see the critical path through their code and usage profiles of several program resources, along with a number of other views. The visualization ....

Miller, B. P., Clark, M., Hollingsworth, J., Kierstead, S., Lim, S., and Torzewski, T. IPS-2: The second generation of a parallel program measurement system. IEEE Transactions on Parallel and Distributed Systems 1, 2 (April 1990), 206--217.

....classes) Moreover, since the internal active instances encapsulate and protect their internal attributes in case of concurrent access, this kind of intra object concurrency minimize problems with internal consistency. 2. 4 Automatic Code Instrumentation In many published monitoring systems [8, 10, 11], the required instrumentation has to be inserted manually. In addition, often events have to be named explicitly and this naming has to be done manually as well. This has several drawbacks. Every location in the program, where an interesting event occurs, has to be spotted and a special statement ....

B. P. Miller at al.: IPS-2: The Second Generation of a Parallel Program Measurement System, IEEE Trans. on Parallel and Distributed Systems, Vol. 1, No. 2, April 1990, pp. 206-217

.... performing work (compute time) the time a processor is stalled waiting for data (memory overhead) the time spent waiting at synchronization events (synchronization overhead) and the time spent in performing work not present in the sequential code (additional work due to parallelization) IPS 2 [42] also uses instrumentation to analyze parallel program performance. It differs from Quartz and Mtool in that it generates a trace of events during application execution, and then analyzes the trace to present useful information to the user. Apart from giving information like those provided by ....

B. P. Miller et al. IPS-2: The Second Generation of a Parallel Program Measurement System. IEEE Transactions on Parallel and Distributed Systems, 1(2):206--217, April 1990.

....: 18 4 Conclusions 21 D3H 6 Rel 1. 0 April 20 1995 Introduction PACT 1 Introduction Early approaches of relating performance engineering activities to the development process of parallel software appeared when parallel machines started to reliably generate computational results [22, 23, 20]. Early work in the performance analysis of parallel systems mainly focused on the mechanics of empirical analysis: measurement and monitoring, automatic probe insertion, trace generation, post execution trace analysis, perturbation analysis [21] and trace visualization [16] It was soon ....

B. P. Miller, M. Clark, J. Hollingsworth, S. Kierstead, S.-S. Lim, and T. Torzewski. IPS-2: The Second Generation of a Parallel Program Measurement System. IEEE Transactions on Parallel and Distributed Systems, 1(2):206--217, April 1990.

....Irregular Problems, Parallel Simulation, Time Warp, CM 5, Cluster Computing. 1 Introduction Attempts of relating performance engineering activities to the development process of parallel software have been made since parallel machines started to reliably generate computational results [22, 23, 20]. Early work in the performance analysis of parallel systems mainly focused on the mechanics of empirical analysis: measurement and monitoring, automatic probe insertion, trace generation, post execution trace analysis, perturbation analysis [21] and trace visualization [16] It was soon ....

B. P. Miller, M. Clark, J. Hollingsworth, S. Kierstead, S.-S. Lim, and T. Torzewski. IPS-2: The Second Generation of a Parallel Program Measurement System. IEEE Transactions on Parallel and Distributed Systems, 1(2):206--217, April 1990.

....will belong to the dominating ones [6, 7] 1 This work was sponsored by the Polish State Committee of Scientific Research (KBN) under grant no. 8 S503 015 06 Majority of existing monitoring and visualization tools are performance oriented and do not support object oriented applications [18, 14]. The other are mostly limited to a single programming environment [15] or investigate rather centralized than distributed processing paradigm [16, 13] Moreover, 1) they support neither interoperability nor cooperation within a software heterogeneous environment; 2) their architecture is not ....

B.P. Miller, M. Clark, J. Hollingsworth, S. Kierstead, S. Lim and T. Torzewski, IPS-2: The Second Generation of a Parallel Program Measurement System, IEEE Transactions on Parallel and Distributed Systems 1, 2 (April 1990), pp. 206-217.

No context found.

B.P. Miller, M. Clark, J.K. Hollingsworth, S. Kierstead, S. Lim, and T. Torzewski. IPS-2: The Second Generation of a Parallel Program Measuremnt System. In IEEE Transactions on Parallel and Distributed Systems, volume 1, pages 206-217, April 1990.

No context found.

B. P. Miller, M. Clark, J. Hollingsworth, S. Kierstead, S. Lim, and T. Torzewski. IPS-2: The Second Generation of a Parallel Program Measurement System. IEEE Transactions on Parallel and Distributed Systems, 1(2):206-217, April 1990.

No context found.

B. P. Miller, M. Clark, J. Hollingsworth, S. Kierstead, S. Lim, and T. Torzewski. IPS-2: The Second Generation of a Parallel Program Measurement System. IEEE Transactions on Parallel and Distributed Systems, 1(2):206--217, April 1990.

No context found.

B. P. Miller, M. Clark, J. Hollingsworth, S. Kierstead, S. Lim, and T. Torzewski. Ips-2: The second generation of a parallel program measurement system. IEEE Transactions on Parallel and Distributed Systems, 1(2):206 -- 217, April 1990.

No context found.

B. P. Miller, M. Clark, J. Hollingsworth, S. Kierstead, S. Lim, and T. Torzewski. IPS-2: The second generation of a parallel program measurement system. IEEE Transactions on Parallel and Distributed Systems, 1(2):206 -- 217, April 1990.

No context found.

Barton P. Miller, Morgan Clark, Jeff Hollingsworth, Steven Kierstead, SekSee Lim, and Timothy Torzewski. IPS-2: The Second Generation of a Parallel Program Measurement System. IEEE Trans. on Parallel and Distributed Computing, 1(2), April 1990.

No context found.

Barton P. Miller, Morgan Clark, Jeff Hollingsworth, Steven Kierstead, Sek-See Lim, and Timothy Torzewski. IPS-2: The second generation of a parallel program measurement system. IEEE trans. on Parallel Distributed syst., Vol. 1(2):pp.206--217, April 1990.

First 50 documents  Next 50

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