J. Vetter, K. Schwan, High Performance Computational Steering of Physical Simulations, Proc. Int'l Parallel Processing Symp., Geneva, pp. 128-132, (1997).  Home/Search   Document Not in Database   Summary   Related Articles   Check

....data upstream in a data flow network is problematic. In SCIRun, this type of user input is mainly used for interactive visualization, such as the positioning of the starting point of a streamline, rather than actual application steering. 2.3. 3 Progress and Magellan Progress [VS95] and Magellan [VS97] have been developed at the Georgia Institute of Technology. Progress stands for Program and Resource Steering System. Magellan is a prototype system designed as the successor of Progress. Scope. The types of steering that can be performed with Progress are model exploration and performance ....

J. Vetter and K. Schwan. High performance computational steering of physical simulations. In Proceedings of the 11th International Parallel Processing Symposium, IPPS 97, pages 128--132, 1997.

....to cope with the complexity added by computational steering, that is the problem of scheduling an application whose computational goals change over time according to potentially arbitrary user behaviors. Computational steering is a difficult problem that has been addressed by several researchers [45, 35, 58, 59, 25]. Those efforts mostly addressed the problems of consistency of state among components of tightly coupled applications. In the limited context of MCell, consistency is not a key issue as the application consists of large sets of tasks which can be stopped and re started independently and without ....

....we will certainly investigate how those systems could be of benefit to the VI. Computational Steering has been an active field of research and several projects have provided models, methodologies, and software for steering scientific applications (SCIRUN [45] VASE [35] Progress [58] Magellan [59], CUMULVS [25] One of the main challenges addressed in these works is the notion of state consistency. Several techniques from the area of distributed systems and fault tolerance have been used successfully to build high performance consistent computational steering environments. Our work is ....

J. Vetter and K. Schwan. High Performance Computational Steering of Physical Simulations. In Proceedings of IPPS'97, pages 128 132, 1997.

....into the simulation, to play what if games, and to terminate uninteresting runs early. Enabling the seamless monitoring and interactive steering of parallel and distributed applications presents many challenges. A significant challenge is the definition and deployment of sensors and actuators [27, 30] to monitor and control applications objects (algorithms and data structures) Defining these interaction interfaces and mechanisms in a generic manner, and co locating them with the application s computational objects can be non trivial. This is because the structure of application computational ....

J. Vetter and K. Schwan. "High Performance Computational Steering of Physical Simulations," International Parallel Processing Symposium (IPPS), IEEE, Geneva, April 1997.

....have become critical for transforming simulations into true research modalities. Enabling seamless interaction and steering high performance parallel distributed applications presents many challenges. A key issue is the definition and deployment of interaction objects with sensors and actuators [16] that will be used to monitor and control the applications. These sensors and actuators must be co located with the computational data structures in order to be able to control individual application data structures. Defining these interfaces in a generic manner and deploying them in distributed ....

....occur when pieces of code are executed. Instrumentation code is placed in the application statically at compile time at such instrumentation points. When the corresponding events occur, steering actions and decisions are taken. Systems that fall under this category include Progress [18] Magellan [16] and the CSE [5] 2) Systems with high level abstractions In order to overcome the shortcomings of event based steering, systems such as the Mirror Object Steering System (MOSS) 4] provide higher level abstractions for steering by translating application level data structures and parameters ....

Vetter J., Schwan K: High Performance Computational Steering of Physical Simulations. IEEE International Parallel Processing Symposium (1997).

..... Many existing applications restrict the interaction to the visualization process (e.g. the direction of view, the zoom factor) A more advanced form of interaction, referred to as computational steering, allows the user to interact with the simulation process. Several systems support steering [10, 11, 15, 17, 26, 27]. But they typically provide only lowlevel interactions and require users to monitor or change the application program s internal variables. We focus on the complete execution of a simulation for a given set of inputs. However, we also examine simulations that produce results during the execution ....

J. Vetter and K. Schwan. High Performance Computational Steering of Physical Simulations. In Proceedings of the 11th IPPS '97, pages 128--132, 1997.

....critical for transforming simulations into true research modalities. Enabling collaborative interaction and steering of high performance parallel distributed applications presents many challenges. A key issue is the definition and deployment of interaction objects with sensors and actuators [1] [2] that can be used to monitor and control the applications. These sensors and actuators must be co located with the computational data structures in order to be able to control individual application data structures. Defining these interfaces in a generic manner and deploying them in distributed ....

J. Vetter and K. Schwan. "High Performance Computational Steering of Physical Simulations," International Parallel Processing Symposium (IPPS), IEEE, Geneva, April 1997.

....of modules and presents them in a single 3D view. Salmon has the ability to send messages upstream the data flow network. This allows steering of upstream module parameters by direct manipulation on the objects displayed by the Salmon module. 3. 3 Progress and Magellan Progress [20] and Magellan [21] have been developed at the Georgia Institute of Technology. Progress stands for Program and Resource Steering System. Magellan is a prototype system designed as the successor of Progress. Scope The types of steering that can be performed with Progress are model exploration and performance ....

J. Vetter and K. Schwan. High performance computational steering of physical simulations. In Proceedings of the 11th International Parallel Processing Symposium, IPPS 97, pages 128--132, 1997.

....system in managing these resources. SCIRun manages scheduling and prioritization of threads, mapping of threads to processors, interthread communication, thread stack growth, memory allocation policies, and memory access exception signals. Steering tools and environments, such as Magellan [126] and Pablo [97] that focus on performance steering and algorithm refinement, address some of these issues. They provide mechanisms for performance tuning that can either be controlled by the user developer or automated based upon performance statistics. However, they do not provide a rich set of ....

....which can be altered at runtime through the use of the Progress runtime system. Steerable objects include sensors, actuators, probes, function hooks, complex actions, and synchronization points. Progress uses a client server program model. Developed by the same group, the Magellan Steering System [126] is derived from the Progress system and extends the steering clients and steering servers model used in the initial system. This system uses a specialized specification language, ACSL, which provides commands for monitoring and steering using probes, sensors, and actuators. However, application ....

[Article contains additional citation context not shown here]

VETTER, J., AND SCHWAN, K. High performance computational steering of physical simulations. In Proceedings of the 11th International Parallel Processing Symposium (Apr. 1997), Geneva, Switzerland.

....7 SCIRun. 2. Systems for Performance Optimizations In these systems, optimizing the application s performance is the main goal and different techniques have been used to accomplish this, ranging from fuzzy logic and neural networks in Autopilot [33] to language directed approaches in Magellan [37]. 3. Systems for Application Configuration and Deployment Systems in this category are not as much steering systems as they are for configuring, deploying and monitoring applications. Systems like the WebFlow [12] and Gateway [4] provide powerful user interfaces, usually web based, to configure ....

....actions and decisions are taken. Thus, events are the highest abstraction provided for program steering. Event streams are sufficient for simple cases, but are not an adequate abstraction for finer control of the application. Systems that fall under this category include Progress [35] Magellan [37] and the CSE[34] b) Systems with high level abstractions In order to overcome the shortcomings of event based steering, the MOSS [8] system provides higher level abstractions for steering by allowing application level data structures and parameters to be modified on the fly. Here, the events ....

[Article contains additional citation context not shown here]

Jeffrey Vetter and Karsten Schwan. High Performance Computational Steering of Physical Simulations. IEEE International Parallel Processing Symposium, pages 21--31, April 1997.

....their le access patterns in sucient detail to correctly choose le policies. As a result, some studies have proposed techniques that can automatically classify access patterns [34, 35] and dynamically choose appropriate policies [33, 36] Computational steering. Interactive application steering [37, 36] is studied extensively, particularly in the context of scienti c applications and immersive visualization. Several techniques for automated decision making have been proposed, ranging from decision tables and trees through standard control theory to fuzzy logic. In contrast to other ....

J. Vetter and K. Schwan, \High Performance Computational Steering of Physical Simulations," in Proc. Int'l Parallel Processing Symp., (Geneva), pp. 128-132, 1997.

....This infrastructure allowed us to experiment with closed loop, rule base driven, adaptive file system control policies. The Falcon monitoring and steering system [59, 60] provides mechanisms for on line information capture and analysis, and interactive program steering. Vetter and Schwan [61] report cases where interactive steering allowed an order of magnitude performance improvement in physics computations. Unlike the Falcon system, our adaptive file system and the Autopilot toolkit provides closed loop, automatic control mechanisms besides interactive steering. The diversity in ....

J. Vetter and K. Schwan, "High Performance Computational Steering of Physical Simulations, " in Proc. Int'l Parallel Processing Symp., (Geneva), pp. 128--132, 1997.

....of blocks of source code and arcs representing the control. The program can be stopped at run time between blocks and some user scripts (C like) are executed. At these points, data can be retrieved and modified. The visualization is managed by an external library such as IRIS Explorer. Magellan [27] requires the user to annotate the source code to produce an abstract view of the program, consisting of steering parameters and output data. These objects can be probed, modified at run time, or provide a breakpoint. A master program can control several steered program. A graphical interface can ....

J. Vetter and K. Schwan. High Performance Computational Steering of Physical Simulations. In Proceedings of the 11th IPPS '97, pages 128--132, 1997.

....a particular programming model, i.e. simple iterative computations with a single main loop[4] Application specific steering systems typically solve the consistency problem in an applicationspecific manner. At the other end of the spectrum, more complex schemes for controlling interaction points[5], 6] and detecting changes that affect consistency[7] have been developed. However, few of these systems address the coordination of steering actions across multiple processes in a distributed system. The scripting approach to interactive steering involves breaking up code written in C, C , or ....

....only when the actuator code is executed, at the point in the computation at which the actuator has been inserted into the code. However, such an approach provides no mechanism for the coordination of actions across multiple processes. Systems such as Pablo[10] Falcon[11] Progress[12] Magellan[5] and VASE[13] ensure consistency of steering actions through reliance on the placement of user or programmer defined instrumentation at safe points in the execution of the code. Progress provides a variety of instrumentation object types including sensors, actuators, probes, function hooks, ....

[Article contains additional citation context not shown here]

J. Vetter and K. Schwan, "High performance computational steering of physical simulations," in Proceedings of the 11th International Parallel Processing Symposium, Geneva, Switzerland, Apr. 1997.

....which can be altered at runtime through the use of the Progress runtime system. Steerable objects include sensors, actuators, probes, function hooks, complex actions, and synchronization points. Progress uses a client server program model. Developed by the same group, the Magellan Steering System [11] is derived from the Progress system, and extends the steering clients and steering servers model used in the initial system. This system uses a specialized specification language, ACSL, which provides commands for monitoring and steering using probes, sensors and actuators. However, application ....

J. Vetter and K. Schwan. High performance computational steering of physical simulations. In Proceedings of the 11th International Parallel Processing Symposium. Geneva, Switzerland, Apr. 1997.

....which can be altered at runtime through the use of the Progress runtime system. Steerable objects include sensors, actuators, probes, function hooks, complex actions, and synchronization points. Progress uses a client server program model. Developed by the same group, the Magellan Steering System [16] is derived from the Progress system, and extends the steering clients and steering servers model used in the initial system. This system uses a specialized specification language, ACSL, which provides commands for monitoring and steering using probes, sensors and actuators. However, application ....

J. Vetter and K. Schwan, "High performance computational steering of physical simulations", in Proceedings of the 11th International Parallel Processing Symposium. Apr. 1997, Geneva, Switzerland.

....of modules and presents them in a single 3D view. Salmon has the ability to send messages upstream the data flow network. This allows steering of upstream module parameters by direct manipulation on the objects displayed by the Salmon module. 3. 3 Progress and Magellan Progress [20] and Magellan [21] have been developed at the Georgia Institute of Technology. Progress stands for Program and Resource Steering System. Magellan is a prototype system designed as the successor of Progress. Scope The types of steering that can be performed with Progress are model exploration and performance ....

J. Vetter and K. Schwan. High performance computational steering of physical simulations. In Proceedings of the 11th International Parallel Processing Symposium, IPPS 97, pages 128--132, 1997.

....Urbana, IL. to optimize both the performance of the application in light of infrastructure requirements and the responsiveness of the infrastructure itself. Key insight and contribution. The primary contribution of this work is an analysis of how to use language directed computational steering [12] on a group of existing applications. We also delve into application specific issues about steering these applications. Essentially, this paper has four contributions. First, we present well defined steering maneuvers for each application. Second, we furnish specific details of how we modified ....

....details an evaluation of using steering on several applications with our steering framework. Section 4 presents common issues distilled from the evaluation. Finally, Section 5 summarizes the paper and furnishes some future research directions. 2 Steering overview Figure 1 illustrates Magellan [12]: a system for language directed computational steering. The Magellan system is composed of three basic pieces: steering servers, a steering language, and application instrumentation. The primary advantage of language directed steering is the possible optimization of user requests [13] Because ....

[Article contains additional citation context not shown here]

J. Vetter and K. Schwan, High performance computational steering of physical simulations, in Proc. Int'l Parallel Processing Symp., Geneva, 1997, pp. 128-132.

....and changes in the mapping of components to computational grid elements. Our aim is to use mobile agents to implement some of the data processing tasks of interactive high performance applications. More specifically, while it is unlikely that a high performance simulation like a fluid dynamics[39] or a finite element code will employ mobile agents for the simulation itself, it is desirable to represent as agents many of the computations and data transformations required for their interactive use. Such representations enable end users to interact with their long running simulations from ....

J.S. Vetter and K. Schwan. High performance computational steering of physical simulations. In Proc. IPPS 97, 1997.

....efficiently adapt to changes in user needs or the environment using eager handlers. 1. Introduction End users of high performance codes increasingly desire to interact with their complex applications as they run, perhaps simply to monitor their progress, or to perform tasks like program steering[5][6] or to collaborate with fellow researchers using these applications as computational tools. For instance, in our own past research, we have constructed a distributed scientific laboratory with 3D data visualizations of atmospheric constituents, like ozone, and with parallel computations that ....

....or from lab office to shop floors or conference rooms. Furthermore, two way interactions occur, such as those where engineers continuously interact via simulations or computational tools, including when jointly steering such computations and sharing alternative views of large scale data sets[5][6] Three concrete instances of such collaborations have been constructed by our group, including an interactively steered simulation of the earth s atmosphere[3] an instance of the hydrology workbench originally developed at the Univ. of Wisconsin[8] and a design workbench used by mechanical ....

J.S. Vetter and K. Schwan, "High performance computational steering of physical simulations", Proceedings of IPPS 97, 1997.

....applications. 3) Empirical evaluations of monitoring assertions and instrumentation signatures demonstrate the utility of delaying mechanism binding until runtime. Sample applications. Two applications demonstrate our ideas. Heat diffusion is a 27 point, 3D stencil, time stepped simulation [22] implemented with kernel level threads for SMP platforms. This simulation exhibits nearest neighbor sharing common to many simulations of physical systems. A more complex example is provided by the Splash 2 OCEAN benchmark. The Ocean benchmark simulates eddy and boundary currents in a cuboidal ....

....research issues. 2. Background: application instrumentation Typical online monitoring systems have four basic components as illustrated in Fig. 1: system software, application, monitoring server, and interactor. Systems that broadly fit this model include Paradyn [17] Falcon [9] Magellan [22], Avatar [20] Vista [23] and various debuggers. We assume no special operating system, special compilers, or hardware. The interactor controls the server; it specifically provides user interface and visualization capabilities. The server is generally a separate process or thread that ....

[Article contains additional citation context not shown here]

J. Vetter and K. Schwan. High performance computational steering of physical simulations. In Proc. Int'l Parallel Processing Symp., pages 128--132, Geneva, 1997.

....and (2) to utilize the configuration mechanisms presented in this paper to develop new technologies for online program configuration. Concerning (2) our group has been investigating the interactive steering of high performance programs, to realize gains in performance and functionality[52]. The essential idea of this technology is to give users the ability to configure remotely (if needed) selected attributes of their application level objects while programs are running, using formulations of configurable object abstractions in distributed systems much like those implemented by CTK ....

J. Vetter and K. Schwan. High Performance Computational Steering of Physical Simulations. In International Parallel Processing Symposium (IPPS), Geneva, April 1997. IEEE.

....on line configuration management [4] and adaptive input output systems. The Autopilot toolkit differs by emphasizing portable performance steering and closed loop adaptive control and by decoupling the steering infrastructure from the policy domain. Likewise, interactive application steering [14] has a long and rich history, particularly in the context of scientific and immersive visualization. By separating performance measurement, control and decision making, Autopilot enables system designers to replace software decision procedures with real time visualization and interactive steering ....

J. Vetter and K. Schwan. High Performance Computational Steering of Physical Simulations. In Proc. Int'l Parallel Processing Symp., pages 128--132, Geneva, 1997.

.... transported using the DataExchange communication infrastructure detailed in [3] 3) information flow, task execution rate, and task execution timings monitored using an on line monitoring facility (examples of such are W3 [7] Chaos MON, and Falcon [4] and (4) an on line steering facility [4] [13] allowing ILI to affect the operation of its tasks. 5. Heuristics ILI s on line adaptation heuristics are applied cyclically in three different steps: 1) Detect State Phase which determines if all tasks are currently meeting their QoS constraints and if not, where the violations are, 2) Predict ....

J. Vetter and K. Schwan. High performance computational steering of physical simulations. IEEE Proceedings of the Internatl Parallel Processing Symposium, pages 128--132, 1997.

....for which Falcon has been developed. The topic of program steering itself is explored in this paper to the extent of defining its requirements on the monitoring system and describing the its low level mechanisms as duals of the monitoring system s mechanisms. More detail on steering appears in [51] as extensions of Falcon s basic steering functionality. For brevity, this paper does not elaborate on two additional aspects of Falcon, which are (1) its support of multiple heterogeneous computing platforms current extensions of Falcon address both single parallel computing platforms running ....

....their runtime behaviors in terms of familiar quantities (e.g. total energy in the MD application) This work differs from related research in program steering (e.g. the Vase system[24] in our focus on steering by human users. Because Falcon enables both algorithmic [44] and interactive [51, 18] program steering, it emphasizes monitoring latencies and overheads more strictly than human interactive systems like Vase. By offering low monitoring latencies, interesting program events may be recognized with suitable delays for corrective actions by adaptation algorithms or human users. The ....

[Article contains additional citation context not shown here]

J. Vetter and K. Schwan. High performance computational steering of physical simulations. In Proc. Int'l Parallel Processing Symp., pages 128--32, 1997.

No context found.

J. Vetter, K. Schwan, High Performance Computational Steering of Physical Simulations, Proc. Int'l Parallel Processing Symp., Geneva, pp. 128-132, (1997).

First 50 documents

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