Parallel programming with message passing and directives

Bova, Steve; Breshears, Clay P.; Gabb, Henry A.; Kuhn, Bob; Magro, Bill; Eigenmann, Rudolf; Gaertner, Greg; Salvini, Stefano; Scott, Heidi

doi:10.1109/5992.947105

Cited by 24 publications

(8 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…There are no strict rules on the type of architectures to be employed. A combination of parallel architectures, hybrids, can be used to enhance areas like message passing and load balancing [13][14][15]. More research can be done on these areas.…”

Section: Discussionmentioning

confidence: 99%

Progressive Learning Paradigms using the Parallel K-Iterations Fast Learning Artificial Neural Network

Tay

Zhang

2007

2007 International Joint Conference on Neural Networks

View full text Add to dashboard Cite

This paper illustrates how progressive network learning can be performed using concepts from the K-Iterations Fast Learning Artificial Neural Network (KFLANN). It is common in certain application domains to require knowledge updating of existing neural networks and many existing learning paradigms require extensive re-training. The KFLANN is an efficient clustering algorithm that provides consistent clusters within a short number of epochs and the paper explains how its capabilities in efficient and consistent clustering aid progressive learning. Progressive learning is useful in domains that handle data samples that arrive at periodic stages over stages of time. The bioinformatics industry is one such domain where data often arrives as segmented blocks during the various stages of bio-molecular experimentation. This work provides an option for progressive learning of partial data that subsequently updates as more complete information is made available. The paper discusses how FLANN, a subset of the KFLANN, can be configured as a progressive learning algorithm to merge segmented data. It further compares the results with those obtained from the typical non-segmented processing. In the process of introducing the progressive learning paradigm in using FLANN, we also present a feasible parallelization algorithm known as the Parallel-KFLANN (P-KFLANN).

show abstract

Section: Discussionmentioning

confidence: 99%

Progressive Learning Paradigms using the Parallel K-Iterations Fast Learning Artificial Neural Network

Tay

Zhang

2007

2007 International Joint Conference on Neural Networks

View full text Add to dashboard Cite

show abstract

“…The case of mixed mode parallel programming has also attracted the attention of some researchers in the last few years (Smith and Bull, 2001;Sannino et al, 2001;Bova et al, 2001;Cappello and Etiemble, 2000), however, the effective application of this paradigm still remains an open question.…”

Section: The Parallel Algorithmmentioning

confidence: 99%

IDEWEP: Web service for astronomical parallel image deconvolution

Almeida

Blanco

Mendez

et al. 2009

Journal of Network and Computer Applications

View full text Add to dashboard Cite

“…Parallel computers are typically programmed by using a message passing programming paradigm (MPI [5,15] PVM [10], etc.) or a shared memory programming paradigm (OpenMP [9,14,18] Multi Threading, etc.).…”

Section: Hybrid Programming Paradigms: Mpi + Openmpmentioning

confidence: 99%

Performance Optimization for Large Scale Computing: The Scalable VAMPIR Approach

Brunst

Hoppe

Nagel

et al. 2001

Computational Science - ICCS 2001

View full text Add to dashboard Cite

Abstract. Performance optimization remains one of the key issues in parallel computing. Many parallel applications do not benefit from the continually increasing peak performance of todays massively parallel computers, mainly because they have not been designed to operate efficiently on the 1000s of processors of todays top of the range systems. Conventional performance analysis is typically restricted to accumulated data on such large systems, severely limiting its use when dealing with real-world performance bottlenecks. Event based performance analysis can give the detailed insight required, but has to deal with extreme amounts of data, severely limiting its scalability. In this paper, we present an approach for scalable event-driven performance analysis that combines proven tool technology with novel concepts for hierarchical data layout and visualization. This evolutionary approach is being validated by implementing extensions to the performance analysis tool Vampir.

show abstract

Parallel programming with message passing and directives

Cited by 24 publications

References 7 publications

Progressive Learning Paradigms using the Parallel K-Iterations Fast Learning Artificial Neural Network

Progressive Learning Paradigms using the Parallel K-Iterations Fast Learning Artificial Neural Network

IDEWEP: Web service for astronomical parallel image deconvolution

Performance Optimization for Large Scale Computing: The Scalable VAMPIR Approach

Contact Info

Product

Resources

About