A. Beguelin, J. Dongarra, A. Geist, R. Manchek, and V. Sunderam. Visualization and Debugging in a Heterogeneous Environment. IEEE Computer, 26(6):88--95, June 1993.  Home/Search   Document Not in Database   Summary   Related Articles   Check

.... the ease at which the APME software can be modified to create code for other parallel and distributive systems, the software was altered to allow the creation of executable representations of SME simulation on a nCube 2 [19] parallel processing system, and on a network ofSUN workstations using PVM [20]. The SME application was used in these cases to verify implementation correctness only, since the size of the SME model is much too small to effectively overcome the large start up communication latency associated with these configurations (the SME application is a relatively small grained ....

A. Beguelin, J. Dongarra and A. Geist, "Visualization and Debugging in a Heterogeneous Environment", Computer, June 1993, . 88-95.

....Servers send the global maximum to their client Selector processes. 2.1 Process topologies in GRADE Figure 1 shows the Application window of the graphical editor GRED in which the process topology of this simple application is defined. We have three Server processes (named Serv[1] Serv[2] and Serv[3] There can be seen six Selectors around the Servers. Three of them connected to Serv[3] named Selector[4] Selector[5] and Selector[6] two of them connected to Serv[2] named Selector[2] and Selector[3] and only one of them connected to Serv[1] The application ....

....topology of this simple application is defined. We have three Server processes (named Serv[1] Serv[2] and Serv[3] There can be seen six Selectors around the Servers. Three of them connected to Serv[3] named Selector[4] Selector[5] and Selector[6] two of them connected to Serv[2] (named Selector[2] and Selector[3] and only one of them connected to Serv[1] The application is executed directly from this environment. At the end each selector process will have acquired the maximum of all integers initially stored in the selector processes. The topology shown can ....

[Article contains additional citation context not shown here]

Beguelin, A., Dongarra, J., Geist, G. and Sunderam, V. (1993) Visualization and debugging in a heterogeneous environment, IEEE Computer, Vol. 26(6).

....provides a comfortable inspection of the details. Examples for performance visualization tools are ParaGraph [19, 20, 21] MulTVision [18] Traceview [28] and PV [26] More information on visualization techniques can be found in [25] and [36] Monitoring tools include ParaGraph [19, 20, 21] Xab [10], and Paradyn [30] The instrumentation is usually build into the runtime system or the operating system, see [23] for an overview. In the area of parallelizing compilers, there is considerable research effort to build modeling tools because such tools are imperative to derive efficient ....

A. Beguelin et al. Visualization and Debugging in a Heterogeneous Environment. IEEE Computer, 26(6):88--95, 1993.

....computational and graphical performances fosters a growing interest in the research community for applying information visualization techniques to various areas of computer science. Some of the concerned areas are program and algorithm visualization [12, 16, 3] debugging and performance analysis [14, 10, 9, 6, 2], information retrieval [11] system management and scientific visualization. However, these techniques do not seem to have been widely applied to the visualization of object oriented systems [8] This paper presents a new proposal in this area. Visibility of complex mechanisms is a very ....

....and we believe that it can be extended to other visualization areas. The rational construction of our virtual images using visual variables is a step in this direction. Beguelin, Dongarra, Geist and Sunderam have worked on visualization applied to the debugging in heterogeneous environment [2]. Their graphical 14 representations are representative of standard solutions for showing concurrency (bargraphs, curves, and 2D call graphs with scrollbars) We believe that this approach is not well suited for the debugging of large distributed programs. 7 Conclusions Through quantitative ....

A. Beguelin, J. Dongarra, Al Geist, V. Sunderam, "Visualization and Debugging in a Heterogeneous Environment", IEEE Computer, vol. 26, num. 6, pp. 88-95, June 1993.

....and send the minimum of the subresults to the previous slave (or to the master in case of ChrSl1 ) via its output port number 2. 5 Related Works A number of other visual programming environments have been developed for parallel applications (e.g. VPE [14] TRAPPER [15] EDPEPPS [6] HENCE [3], PSEE [13] Paralex [2] Most of them are based upon the idea that nodes represent parallel computation, and arcs represent interactions (of some form) among nodes. The problem with the HENCE, Paralex and PSEE approaches is that they force computations to be split into separate processes when ....

A. Beguelin, J. Dongarra, G. Geist, and V. Sunderam. Visualization and debugging in a heterogeneous environment. IEEE Computer, 26(6), June 1993.

....information and provide valuable insight into performance monitoring issues, such as code optimization, debugging, critical path analysis and the identification of performance bottlenecks. The visualization layer for the SuperMSPARC includes tools developed elsewhere (e.g. Pablo [10] Paragraph [2], and Upshot[6] as well as some custom tools designed specifically for tracing MPI based programs. The Pablo and Custom displays are used most often and we will discuss only those two Produce Source Code Instrument the Source Code application specifications analyst requirements Define Probe ....

A. Beguelin, J. Dongarra, A. Geist, R. Manchek, and V. Sunderam. Visualization and Debugging in a Heterogeneous Environment. IEEE Computer, 26(6):88--95, June 1993.

....development effort has concentrated on the algorithmic level, often in the form of language design and accompanying tools. Nevertheless, there are a number of comparable projects. Most notable is perhaps the work on modeling applications through functional dependency graphs, exemplified by HENCE [BDGS93] and CODE [BAS89] In this approach, an application is viewed as a collection of functional and indivisible units, connected to each other by pure input output relations. Each unit can be executed on a separate node, and parallelism is achieved by scheduling independent units simultaneously on ....

A. Beguelin, J.J. Dongarra, G.A. Geist, and V.S. Sunderam. Visualization and Debugging in a Heterogeneous Environment. Computer, 26(6), June 1993.

....Program constructed CODE CODE 2.0 [13] is a graphical parallel programming system. The diagramming technique is based largely on the dataflow model but encourages medium to coarse grain parallel programming. Another system that is of interest and very much related to our work is HeNCE [2], which provides an environment for programmers to create parallel applications and execute them over a network of computers. To compare our system with CODE and HeNCE, we quote an example given in [3] The problem is to numerically integrate a function. Figure 14 16 shows the three diagrams ....

Adam Beguelin, Jack Dongarra, Al Geist, and Vaidy Sunderam. Visualization and Debugging in a Heterogeneous Environment. IEEE Computer, 26(6):88--95, June 1993.

....ParaGraph provides views of processor utilization and message passing using a set of views such as Kiviat and Gantt charts. Other significant work on visualizing the operations of message based distributed systems includes Seeplex Seecube[Cou88] HyperView Tapestry[MR90] Belvedere[CHK92] and Xab[BDGS93]. Similarly, systems for visualizing compiler driven (FORTRAN) parallel programs such as SCHEDULE[DS87] and SHMAP[DBKF90] have been developed. Our work focuses specifically on threads programs and programming primitives, and is pioneering in its rigorous depiction of this paradigm. In the next ....

Adam Beguelin, Jack Dongarra, Al Geist, and Vaidy Sunderam. Visualization and debugging in a heterogeneous environment. Computer, 26(6):88--95, June 1993.

....can be performed regularly, whereas in off line environments it can only be performed at the beginning and the end of the application s execution. 5 Related Work A number of other visual programming environments have been developed for parallel applications (e.g. TRAPPER [18] EDPEPPS [4] HENCE [2], PSEE [15] MOGUL [9] Most of them are based upon the idea that nodes represent parallel computation, and arcs represent interactions (of some form) among nodes. The problem with the HENCE and PSEE approaches is that they force computations to be split into separate processes when ....

A. Beguelin, J. Dongarra, G. Geist, and V. Sunderam. Visualization and Debugging in a Heterogeneous Environment. IEEE Computer, 26(6), June 1993.

....objects meanwhile he she investigates the program behavior by PROVE. According to our experience it significantly reduces the time spent on debugging or performance tuning. 7. Related Works A number of other visual parallel programming language and environment have been developed (e.g. HENCE [2],CODE [12] PSEE [10] Paralex [1] TRAPPER [13] Most of them are based upon the idea that nodes represent computation, and arcs represent interactions (of some form) among nodes. The problem with the HENCE, CODE, Paralex and PSEE approaches is that they force computations to be split into ....

A. Beguelin, J. Dongarra, G. Geist, and V. Sunderam. Visualization and debugging in a heterogeneous environment. IEEE Computer, 26(6), June 1993.

....can check all the runtime activities of a particular GRAPNEL icon. 9 Figure 6 Process and Host Communication diagrams of PROVE 6 Related works. A number of other visual programming environments have been developed for parallel applications (e.g. VPE [21] TRAPPER [22] EDPEPPS [9] HENCE [2], CODE [20] PSEE [17] Paralex [1] Most of them are based upon the idea that nodes represent parallel computation and arcs represent interactions (of some form) among nodes. The problem with the HENCE, CODE, Paralex and PSEE approaches is that they force computations to be split into separate ....

A. BEGUELIN, J. DONGARRA, G. GEIST, AND V. SUNDERAM. Visualization and debugging in a heterogeneous environment. IEEE Computer, 26(6), June 1993.

....in parallel where possible. This paper will present results on the use of heterogeneous computing to increase the performance of CSTEM [17] a finite element based application for the design of composite AGTE components. A publicly available programming tool for heterogeneous computing, HeNCE [3, 4], was used to create a heterogeneous version of CSTEM. The most difficult problem associated with heterogeneous distributed computing, as well as parallel computing in general, is the mapping and scheduling problem. This problem assigns the individual code fragments, or tasks, to the set of ....

A. Beguelin, J. Dongarra, A. Geist, and V. Sunderam, "Visualization and debugging in a heterogeneous environment," IEEE Computer, vol. 26, pp. 88--95, June 1993.

....order during replay in debugging mode as during the original program execution. Methods for deterministic replay of nondeterministic programs can be found in [11, 19, 17] 1. 2 Related Systems Phred is similar to several other visual parallel programming environments, namely Code [9, 20] HeNCE [4, 5], Paralex [1, 2] and Schedule [12] However, Phred is unique in its graph structures and its emphasis on determinacy. The intent is to develop a parallel programming environment which encourages deterministic programs without requiring programs to be deterministic. In this paper, we describe the ....

....parallel languages. Previous papers [3, 6] illustrate Phred s utility by presenting a variety of parallel programs expressed in Phred. The Phred system was built as a prototype, the software was not developed to the state where it could be distributed to users. A similar follow on system, HeNCE [4, 5] is freely available. However, the HeNCE system does not yet support determinacy checking features or the data flow graph provided by Phred. This paper describes the Phred programming language, determinacy analysis techniques, and the realization of these concepts in the Phred programming ....

A. Beguelin, J. Dongarra, G. A. Geist, and V. S. Sunderam. Visualization and debugging in a heterogeneous environment. IEEE Computer, June 1993. To appear.

....travelling backwards in time. In order to determine if causality violations occurred, we ran a small test program on approximately ten machines (including Sun4 s and Pmax s, all running Mach) connected by Ethernet here at Carnegie Mellon. The program was written using PVM version 3. 1 (see [2, 3, 6, 10]) and consisted of a master process on one machine and a slave process on each of the others. Every five minutes, the master process would awaken and do the following for each slave: 1. Get the current time M1. 2. Send a message to the slave. 3. Wait for a reply message. 4. Get the time M2 ....

A. Beguelin, J. Dongarra, G. A. Geist, and V. S. Sunderam. Visualization and debugging in a heterogeneous environment. IEEE Computer, June 1993. To appear.

....of supercomputers can provide an even more powerful computing resource than any single machine. These ideas are not new; parallel computing has long been an active area of research. The fact that networks of computers are commonly being used in this fashion is new. Software tools like PVM [1, 7], P4 [11] Linda [4] Isis [2] and Express [6] allow a programmer to treat a heterogeneous network of computers as a parallel machine. These tools allow the programmer to partition a program into pieces which may then execute in parallel, occasionally synchronizing and exchanging data. The tools ....

A. Beguelin, J. Dongarra, A. Geist, and V. Sunderam. Visualization and debugging in a heterogeneous environment. IEEE Computer, 26(6):88--95, June 1993.

....is a recurring benefit of all the modifications used in our system integrated approach: a user decides at execution time whether or not the visualization support is active. This is a clear advantage over relinking to special libraries or adding extra header files as necessary in systems such as Xab[1]. Every effort is made to insure that support needed for visualization does not hinder or delay the application programmer. It simply should be there on demand. 4.3 Synchronization Modifications The Conch system provides common synchronization primitives such as barriers, Ada style rendezvous, ....

Adam Beguelin, Jack Dongarra, Al Geist, and Vaidy Sunderam. Visualization and debugging in a heterogeneous environment. Computer, 26(6):88--95, June 1993.

No context found.

A. Beguelin, J. Dongarra, A. Geist, R. Manchek, and V. Sunderam. Visualization and Debugging in a Heterogeneous Environment. IEEE Computer, 26(6):88--95, June 1993.

No context found.

A, Beguelin, J.J. Dongarra, G.A. Geist, V.S. Sunderam. Visualization and Debugging in a Heterogeneous Environment. IEEE Computer 26, 6 (1993).

No context found.

A, Beguelin, J.J. Dongarra, G.A. Geist, V.S. Sunderam. Visualization and Debugging in a Heterogeneous Environment. IEEE Computer, Vol 26, No. 6, 1993.

No context found.

A. Beguelin, J. Dongarra, A. Geist, and V. Sunderam, "Visualization and Debugging in a Heterogeneous Environment," IEEE Computer Magazine, June 1993.

No context found.

A. Beguelin, J. Dongarra, A. Geist, and V.S. Sunderam, "Visualization and Debugging in a Heterogeneous Environment", IEEE Computer 26, 6, June 1993.

No context found.

A. Beguelin, J. Dongarra, A. Geist, R. Manchek, and V. Sunderam, Visualization and debugging in a heterogeneous environment. Computer, 26 (6): 88-95, 1993.

No context found.

A. Beguelin, J. Dongarra, A. Geist, and V. Sunderam, "Visualization and Debugging in a Heterogeneous Environment," Computer, 26(6), pp. 88-95, 1993.

No context found.

Beguelin, A., J. Dongarra, A. Geist, and V. Sunderam. Visualization and Debugging in a Heterogeneous Environment. IEEE Computer 26(6):88-95, June, 1993.

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