D. Szafron and J. Schaeffer. An experiment to measure the usability of parallel programming systems. Concurrency: Practice and Experience, 8(2):147--166, 1996.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....by design pattern is not as efficient as the hand crafted PVM version. It is understandable since DPnDP is built on top of PVM. Such performance penalty is expected in high level systems which provide more robust code. Other design pattern based systems report similar or worse loss of performance[24]. 5.3 Application Specific Performance The PQSRS application described in section 3 was implemented and experiments were conducted to ascertain its performance on a cluster of Sun Sparc workstations connected by ethernet. The application was used to sort randomly generated integers. The Table 3 ....

D. Szafron and J. Schaeffer. "An Experiment to Measure the Usability of Parallel Programming Systems". Concurrency: Practice and Experience, 8(2):146--166, 1996.

....in the Proceedings of the 1997 International Scientific Computing in Object Oriented Parallel Environments Conference (ISCOPE 97) 24] of compiler and run time technology. In this progression, we can see more emphasis placed on the usability of the tools rather than raw performance gains [44]. Each successive system reduced the probability of introducing programmer errors. However, since performance considerations cannot be ignored, each successive system also supported incremental application tuning. A related performance issue is openness [40] where a user is able to access ....

D. Szafron and J. Schaeffer. An experiment to measure the usability of parallel programming systems. Concurrency: Practice and Experience, 8(2):147--166, 1996.

No context found.

D. Szafron and J. Schaeffer. An experiment to measure the usability of parallel programming systems. Concurrency: Practice and Experience, 8(2):147--166, 1996.

.... that cannot be represented by parallel design patterns How does this approach compare to other parallel programming approaches Fortunately, questions like these have been asked before and there is a compilation of thirteen characteristics that define the ideal parallel programming system [28, 32]. These characteristics are grouped into three categories as shown in Figure 1 and are explained in more detail later in the paper. Careful consideration of these thirteen characteristics led us to add two more features to our new process, design pattern templates and multiple programming layers. ....

....C tools, but Orca, as a new language, does not supply an extended tool set. 12. Usability The usability characteristic of parallel programming systems is the most important aspect that seems to be ignored in the literature. The two notable exceptions are studies undertaken for Enterprise [32] and Orca [33] This lack of attention is partially due to the difficulty in conducting usability studies. We have undertaken a preliminary usability study for the CO 2 P 3 S environment. Some results are presented in the next section. These early results concentrate on the amount of code that a ....

[Article contains additional citation context not shown here]

D. Szafron and J. Schaeffer. An experiment to measure the usability of parallel programming systems. Concurrency: Practice and Experience, 8(2):147-166, 1996.

.... that cannot be represented by parallel design patterns How does this approach compare to other parallel programming approaches Fortunately, questions like these have been asked before and there is a compilation of thirteen characteristics that define the ideal parallel programming system [27, 30]. These characteristics are grouped into three categories as shown in Figure 1 and are explained in more detail later in the paper. Careful consideration of these thirteen characteristics led us to add two more features to our new process, design pattern templates and multiple programming layers. ....

....C tools, but Orca, as a new language, does not supply an extended tool set. 12. Usability The usability characteristic of parallel programming systems is the most important aspect that seems to be ignored in the literature. The two notable exceptions are studies undertaken for Enterprise [30] and Orca [31] This lack of attention is partially due to the difficulty in conducting usability studies. Since PDP is a new process and the CO 2 P 3 S environment is under construction, no user studies have been conducted. However, since the design pattern templates of PDP are an extension of ....

[Article contains additional citation context not shown here]

D. Szafron and J. Schaeffer. An experiment to measure the usability of parallel programming systems. Concurrency: Practice and Experience, 8(2):147-166, 1996.

....model in Enterprise was at a much higher level, with many of the low level details handled by a combination of compiler and run time technology. In this progression, we can see more emphasis placed on the usability of the tools rather than raw performance gains, as shown by Szafron and Schaeffer [10]. Each successive system reduced the probability of introducing programmer errors. Performance considerations cannot be ignored, though, so successive systems also supported incremental application tuning. A related performance issue identified by Singh et al. 9] is openness, where a user is ....

D. Szafron and J. Schaeffer. An experiment to measure the usability of parallel programming systems. Concurrency: Practice and Experience, 8(2):147-- 166, 1996.

....information as to the expected behavior of the program, and forces it to meet those constraints. Blossom is a C class library. Unlike C Linda and Fortran Linda, it does not add new syntax, alter the semantics, or extend the host language, each of which tends to result in programmer confusion [17]. C is a modern programming language with a rich set of features. This enables Blossom to achieve many (a) b) c) Fig. 1. Differences between tuple spaces; a) Original Linda tuple space: It is a single collection of tuples of arbitrary structure. This implies that the user can try to insert ....

....Blossom. There is no easy metric, such as speedup, to compare against. A fair evaluation would require users to implement solutions using C Linda and Blossom, and then compare the time it takes to correctly implement their solutions. Although we have experience performing these type of experiments [17], it is too early in the life cycle of Blossom to go to this effort. Once we have completed the full Blossom implementation, we will be very interested in quantifying the influence of safer tuple spaces on the program design and implementation time. Blossom is an evolving system. The following ....

Szafron, D., and Schaeffer, J. An Experiment to Measure the Usability of Parallel Programming Systems. Concurrency: Practice and Experience 8, 2 (1996), 147--166.

....evolve the system beyond a certain point. Consequently, an entirely new project, called Enterprise, was initiated. Enterprise is a template based parallel programming environment which offers a much wider range of related tools for parallel program design, coding, debugging and performance tuning [16, 17, 18, 19, 20, 21]. It has been publicly available since 1993 (http: web.cs.ualberta.ca enter) Several other parallel programming systems have relied on techniques that are similar to the approach used by us (for example [22, 11, 23, 24, 25, 26, 10] Many of our results and experiences with FrameWorks and ....

....the template based model. These tools must support the same level of abstraction as the programming model. 3. Tool Usability (Usability) The ideal tool should have a high degree of usability. It should be easy to learn and easy to use. Usability assessments have been neglected in the literature [20]. 4. Application Portability (Portability) The tool should allow the user to port applications to a number of different architectures. Some performance losses may be expected for a poorly chosen architecture, but the program should still run. 5 Outline of FrameWorks This section provides a brief ....

[Article contains additional citation context not shown here]

D. Szafron and J. Schaeffer. "An Experiment to Measure the Usability of Parallel Programming Systems ". Concurrency: Practice and Experience, 8(2):146--166, 1996.

....and evaluating the performance of a parallel application. That is, the implementation of the parallelism is handled by the PPS. The user chooses the computational and I O templates that give the best performance. A recent study examined the usability of several parallel programming systems [33]. The study found that using computational templates to create parallel applications is beneficial. The user code complexity is significantly reduced and the application is up and running much sooner since the templates are correctly implemented by the PPS for the selected parallel behaviour. The ....

D. Szafron and J. Schaeffer, "An Experiment to Measure the Usability of Parallel Programming Systems," Concurrency: Practice and Experience, 8(2), pp. 147-166, 1996.

....in its uniformity with the underlying model and the system into which it is integrated. It is easy to learn, simple to manage, and effective. Context switches between the Design, Animation and Replay Views are completely seamless. This uniformity increases the productivity of parallel programmers [Szafron and Schaeffer, 1996]. Figure 5. Breakpoint Browser and Replay View. 3.4 THE PERFORMANCE VIEWS Performance debugging is quite different from logic debugging and there are different issues that must be addressed [Woloschuk et al. 1995] Performance tuning can be divided into three operations. The first operation ....

D. Szafron and J. Schaeffer, An Experiment to Measure the Usability of Parallel Programming Systems. To appear in Concurrency Practice and Experience, 1996.

....model in Enterprise was at a much higher level, with many of the low level details handled by a combination of compiler and run time technology. In this progression, we can see more emphasis placed on the usability of the tools rather than raw performance gains, as shown by Szafron and Schaeffer [12]. Each successive system reduced the probability of introducing programmer errors. However, since performance considerations cannot be ignored, each successive systems also supported incremental application tuning. A related performance issue identified by Singh et al. 10] is openness, where a ....

D. Szafron and J. Schaeffer. An experiment to measure the usability of parallel programming systems. Concurrency: Practice and Experience, 8(2):147--166, 1996.

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