SOFTWARE TOOLS FOR SCALABLE MULTILEVEL APPLICATION ENGINEERING

Citation
Km. Decker et Bjn. Wylie, SOFTWARE TOOLS FOR SCALABLE MULTILEVEL APPLICATION ENGINEERING, The international journal of supercomputer applications and high performance computing, 11(3), 1997, pp. 236-250
Citations number
14
Categorie Soggetti
Computer Application, Chemistry & Engineering","Computer Sciences, Special Topics","Computer Science Hardware & Architecture","Computer Science Interdisciplinary Applications
ISSN journal
10783482
Volume
11
Issue
3
Year of publication
1997
Pages
236 - 250
Database
ISI
SICI code
1078-3482(1997)11:3<236:STFSMA>2.0.ZU;2-G
Abstract
Contemporary activities at CSCS/SCSC have resulted in two complementar y software systems for practical parallel programming. Both developmen ts are user oriented and application driven, efficiently exploiting an d reusing demonstrated portable technologies at multiple levels that w e have proven to scale to systems with large numbers of processors. Th e Annai parallel application engineering environment supports existing standards for portable program development (HPF, Fortran, C, MPI) and offers convenient program browsing and navigation, execution control, and interaction mechanisms. Specific functionality for parallel progr amming includes high-level language support for unstructured computati ons, interactive source-level symbolic debugging with deadlock detecti on and deterministic execution replay, SPMD/data-parallel debugging wi th distributed breakpoints and array visualization, scalable profile s ummary displays of execution statistics accumulated at runtime, and de tailed program evolution anti processor interaction charts. The intell igent program development environment (PDE) is typified by four charac teristics supporting programming at a very abstract level, closer to t he scientist's perspective: application-oriented problem description f ormalisms, the use of design skeletons and templates, an interactive u ser guidance mechanism, and automatic program synthesis techniques. On going work aims at the integration and further development of sofar de monstrated results with additional advanced technologies into comprehe nsive application engineering and problem-solving environments for pro ductive parallel computing with distributed resources.