In this paper, we present an integrated system architecture for metacomputing on top of distributed scientific resources via semantic-driven information management. Our approach is to consider both data and programs as objects of an application-specific ontology, describing them via corresponding metadata schemata that can be exploited to search for and to generate information in an efficient and automated way. Metacomputations are expressed in the form of workflows thereby enabling a flexible combination of data and program objects that can reside on different servers or grid resources. Workflow descriptions are associated with ontology concepts and can be activated as a side-effect of searching for information to produce new data on demand. We also report on the current status of an implementation that provides support for several aspects of this architecture.

Often scientists need to locate appropriate software for their problems and then select from among many alternatives. We have previously proposed an approach for dealing with this task by processing performance data of the targeted software. This approach has been tested using a customized implementation referred to as PYTHIA. This experience made us realize the complexity of the algorithmic discovery of knowledge from performance data and of the management of these data together with the discovered knowledge. To address this issue, we created PYTHIA-II—a modular framework and system which combines a general knowledge discovery in databases (KDD) methodology and recommender system technologies to provide advice about scientific software/hardware artifacts. The functionality and effectiveness of the system is demonstrated for two existing performance studies using sets of software for solving partial differential equations. From the end-user perspective, PYTHIA-II allows users to specify the problem to be solved and their computational objectives. In turn, PYTHIA-II (i) selects the software available for the user's problem (ii) suggests parameter values, and (iii) assesses the recommendation provided. PYTHIA-II provides all the necessary facilities to set up database schemas for testing suites and associated performance data in order to test sets of software. Moreover, it allows easy interfacing of alternative data mining and recommendation facilities. PYTHIA-II is an open-ended system implemented on public domain software and has been used for performance evaluation in several different problem domains.

Problem-solving e*nvironments (PSEs) interact with the user in a language “natural” to the associated discipline, and they provide a high-level abstraction of the underlying, computationally complex model. The knowledge-based system PYTHIA addresses the problem of (parameter, algorithm) pair selection within a scientific computing domain assuming some minimum user-specified computational objectives and some characteristics of the given problem. PYTHIA's framework and methodology are general and applicable to any class of scientific problems and solvers. PYTHIA is applied in the context of Parallel ELLPACK where there are many alternatives for the numerical solution of elliptic partial differential equations (PDEs). PYTHIA matches the characteristics of the given problem with those of PDEs in an existing problem population and then uses performance profiles of the various solvers to select the appropriate method given user-specified error and solution time bounds. The profiles are automatically generated for each solver of the Parallel ELLPACK library.— Authors' Abstract

In this paper we formulate new mapping strategies for partial differential equations

An allocation model for mapping a real-time application to certain k-processor multiprocessor systems is developed and analyzed. Its objective is minimizing the total processing time of the application by exploiting the parallelism of the application-architecture pair. The model is formulated in terms of the performance characteristics of the system and the resource requirements of the computation involved. Experience with the model suggests that it can be used effectively for the performance evaluation of application-distributed system pairs.>

In this paper we study the partitioning and allocation of computations associated with the numerical solution of partial differential equations (PDEs). Strategies for the mapping of such computations to parallel MIMD architectures can be applied to different levels of the solution process. We introduce and study heuristic approaches defined on the associated geometric data structures (meshes). Specifically, we study methods for decomposing finite element and finite difference meshes into balanced, nonoverlapping subdomains which guarantee minimum communication and synchronization among the underlying associated subcomputations. Two types of algorithms are considered: clustering techniques based on sequential orderings of the discrete geometric data and optimization based techniques involving geometric or graphical metric criteria. These algorithms support the automatic mode of a geometry decomposition tool developed in the parallel ELLPACK environment which is implemented under X11-window systems. A brief description of this tool is presented.

The authors concentrate on the simulation study of two distributed task allocation procedures: the load balancing and the LOCO procedure. The first is widely used in general-purpose processing. The second was recently introduced and analyzed in the processing environment corresponding to the complex multitask jobs typical of some supercomputing and artificial-intelligence-oriented systems. Both procedures are simulated and compared in realistic situations, where both processing resources and interconnection network may represent the system bottleneck. Both were tested using carrier-sense multiple-access communications protocols. The CSMA/CD protocol performed better than TDMA with load balancing, but no difference was found with the LOCO procedure. For a large number of special-purpose processing resources and a large number of jobs in the system, LOCO produced better results than load balancing.>

This paper presents closed form expressions for average delay in a class of work conserving single server queueing systems suggested by, but not limited to, the behavior of heavily I/O oriented computing systems. Jobs arrive at the system according to a Poisson process, and each job consists of a set of tasks which must be accomplished in a prescribed order. The tasks with indexihave a general service time distribution Giand a priorityPiwhich.is arbitrary but prescribed. Tasks are serviced within composite priority groups in a preemptive resume fashion and in the order in which, the tasks reached a ready-for-service status. Such queue behavior would be characteristic, for example, of a front end processor or a terminal controller in a communications environment. A design example is presented which illustrates the usefulness of the model to compare alternatives in communications systems design and other areas as well.

This paper describes the performance of a store-store-and-forward communications network with an adaptive routing strategy. The network chosen is a simplification of the AUTODIN network (described below) and the strategy is a modification of the routing of ARPANET. The objective is to make a network such as AUTODIN, with its wide variation in message length, effectively responsive for computercomputer and terminal-computer messages as well as record traffic. A second objective is to increase survivability following isolated equipment failures. A simulation study is used to show that improved performance is achieved.