Abstract not found

Massively parallel computers are necessarily distributed memory architectures (DMA), as only DMAs offer unlimited upward scalability. However, the usual programming model for DMAs, the message-passing paradigm, is too difficult to make such machines acceptable for a larger community of users. Therefore, easier high-level programming models are needed, in conjunction with parallelizing compilers. This will allow the user to formulate a parallel program in terms of application-specific concepts, while low-level issues such as optimal data distribution and coordination of the parallel threads is handled by the compiler. Specifically, the model must provide expressive means for defining explicitly all kinds of applicationspecific data structures, array-like or hierarchical, static or dynamic, regular or irregular. The paper describes such a model and programming system, called PROMOTER, and its underlying concepts. Experiences with a first implementation of PROMOTER are communicated. 1. IN...

Mapping is regarded as one of the core components of parallel computing environments on various platforms, and influences to a large extent the runtime performance of a parallel application. To be efficient, mapping algorithms must be able to exploit as much as possible high-level information about spatial and temporal structures that occur in an application. Even in so-called irregular applications, data and communication structures are often not entirely irregular, but possess certain regularities and symmetries.

For problems with highly dynamic behavior, our experiments showed that there are specific characteristics for different applications. We therefore propose a mapping environment providing several strategies for both granularity control and dynamic load balancing. For appropriate selection and parameterization of strategies, we extract the application characteristics using trace-based profiling and appropriate evaluations. Our profiling approach differs from others in that more information (like argument sizes or the branching factor) is collected and more advanced evaluations are performed. Large real-life applications have been successfully measured, and examples demonstrate differences in characteristics and which the corresponding strategies being appropriate are. 1 Introduction Although functional programs, by definition, provide a rich source of parallelism because of the absence of side-effects, they nevertheless pose similar problems as encountered with procedural programs for c...

: The PROMOTER programming language is designated for data parallel applications that are to run on massively parallel computers with distributed memory. This paper presents an object-oriented implementation model for the PROMOTER language. An objectoriented approach to compile data-parallel programs to message passing programs can reduce design complexity, facilitate reuse of components, and ease specialization of code, result in an efficient and portable implementation. Keywords: Massively Parallel Programming Model, Data Parallelism, SPMD, Data Topology. This work is supported by the Real World Computing Partnership (RWCP), Japan. 2 PROMOTER: Implementation Model Contents 1 Introduction 3 2 An Abstract Interface for Data-parallel Operations 3 2.1 Processes and Domains : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 3 2.2 Topologies : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 4 2.3 Distributed Objects : : : : : : : : ...