This talk will introduce the massively parallel computer research and development of the Japanese Real World computing Project (RWC). RWC is a ten-year project whose budget is expected to be 600 million to 700 million...
详细信息
This talk will introduce the massively parallel computer research and development of the Japanese Real World computing Project (RWC). RWC is a ten-year project whose budget is expected to be 600 million to 700 million dollars (US). We plan to develop an original 16 K PE stand-alone computer with a parallel C++ language, a more abstract object-oriented language, and a massively parallel operating system.
This talk gives an overview of the progress in research on Concurrent object-orientedcomputing both in theory and practice. From the theoretic standpoint, the developments of semantics bases for concurrent object-ori...
详细信息
Control and data parallelism are two complementary but often mutually exclusive paradigms used to program massively parallel systems. We propose to encapsulate both control and data parallelism in regular classes of a...
详细信息
ISBN:
(纸本)0818656026
Control and data parallelism are two complementary but often mutually exclusive paradigms used to program massively parallel systems. We propose to encapsulate both control and data parallelism in regular classes of a sequential objectoriented language: a SPMD programming model is used and thus no language extensions are needed, provided a Shared Virtual Memory is available. We show how these ideas are implemented in EPEE, our Eiffel parallel Execution environment. As an example, we present the implementation of both paradigms on a toy linear of both paradigms on a toy linear algebra example and show how they can interoperate. We conclude with some performance results and prospective remarks.
The objective of this paper is the description of the implementation of the WAVE model and language for highly parallel processing in open computer networks as well as an application of the distributed WAVE interprete...
详细信息
The objective of this paper is the description of the implementation of the WAVE model and language for highly parallel processing in open computer networks as well as an application of the distributed WAVE interpreter for modelling complex systems within an object-oriented philosophy. WAVE represents a new distributed programming paradigm enabling the dynamic creation and processing of arbitrary network-structured knowledge. The model operates in a spatial pattern-matching, logic-flow, mode, without any centralized control.
This paper describes a powerful environment for the development and execution of object-orientedparallel applications on a multicomputer or a cluster of workstations. The XENOOPS environment implements a global objec...
详细信息
ISBN:
(纸本)0818650605
This paper describes a powerful environment for the development and execution of object-orientedparallel applications on a multicomputer or a cluster of workstations. The XENOOPS environment implements a global object space and offers support for dynamic load balancing and fault tolerance. Workload redistribution at run time is realized applying object migration and granularity control. Fault tolerance is realized by object replication. A advanced debugging tool is presented. This tool adds support for the observation of interacting objects in a transparent, dynamic and selective way by extending the object invocation subsystem of the execution environment. A meta-object can be created for every object to intercept the invocations on that particular object and to observe the state changes. Results from an advanced parallel application in the area of partial differential equations illustrate the applicability of our observation software.
In this paper, we report the design of a parallelobject-oriented implementation of a polymer dynamics application. This application is currently used for the simulation of very large and complex molecules such as hig...
详细信息
Heterogeneously distributed and parallelcomputingenvironments are highly dependent on hardware, data migration, and protocols. The result is significant difficulty in software reuse, portability across platforms, an...
详细信息
The proceedings contain 161 papers. The special focus in this conference is on High-Performance computing and Networking. The topics include: High performance computing and networking for numerical weather prediction;...
ISBN:
(纸本)9783540579816
The proceedings contain 161 papers. The special focus in this conference is on High-Performance computing and Networking. The topics include: High performance computing and networking for numerical weather prediction;high performance computing and networking as base element of the environmental information system UIS of Baden-Württemberg;high speed networking for the research community in Europe;parallel knowledge processing in open networks;distributed file caching in parallel architectures utilizing high speed networks;application fundamentals of distributed, high-performance computing and networking systems;networking perspectives for collaborative supercomputing in European aerospace research and industry;future aerospace working scenarios using high speed networks and supercomputers applied to flow simulation for complete aircraft;a software environment for cooperative simulation and visualization in the aerospace field;implementation of visualization tools in a distributed environment;European HPC competitiveness centres for industry;an open high-performance computing infrastructure in industry;a catalog of classifying characteristics for massively parallel computers;building a parallelcomputingenvironment from heterogenous resources;a distributed computing center software for the efficient use of parallel computer systems;cluster management software for open objectoriented systems;transmission rates and performance of a network of computers;evaluation of parallel operating systems;recent developments in hybrid cg methods;the preconditioned conjugate gradient method on distributed memory systems;communication cost reduction for krylov methods on parallel computers and optimisations for the memory hierarchy of a singular value decomposition algorithm implemented on the MIMD architecture.
In this paper, we report on the design and implementation of a parallelobject-oriented implementation of a protein simulation application, which is the heart of a drug design software package, to be exploited at Novo...
详细信息
Control and data parallelism are two complementary but often mutually exclusive paradigms used to program massively parallel systems. We propose to encapsulate both control and data parallelism in regular classes of a...
详细信息
Control and data parallelism are two complementary but often mutually exclusive paradigms used to program massively parallel systems. We propose to encapsulate both control and data parallelism in regular classes of a sequential objectoriented language: a SPMD programming model is used and thus no language extensions are needed, provided a shared virtual memory is available. We show how these ideas are implemented in EPEE, our Eiffel parallel Execution environment. As an example, we present the implementation of both paradigms on a toy linear algebra example and show how they can interoperate. We conclude with some performance results and prospective remarks.< >
暂无评论