The DCABES 2009 presentation ppt can be downloaded as follows:
Hai Xiang Lin: Trends and Challenges in High Performance Computing
Stefan Vandewalle: Analysis of a Time-Parallel Time-Integration Method
Shiming Xu: Utilizing CUDA for Preconditioned GMRES Solvers
Hyoungseok Chu: Parallel ADI Method for Parabolic Problems on GP-GPU
Mingxi Yang: Fast Signature Scheme for Network Coding
Nicholas Christakis: The Application of Artificial Neural Networks in Engineering and Finance
PENG Hai: A FRACTAL ENCRYPTION ALGORITHM
Fahad Fazal Elahi Guraya: People Tracking via a Modified CAMSHIFT Algorithm
Miloslav Hub: Method of Password Security Evaluation
Keynote Speakers of the DCABES 2009
Parallel time domain methods
Professor Stefan Vandewalle
Computer Science Department Faculty of Engineering, Katholieke Universiteit Leuven, Celestijnenlaan Leuven, Belgium
Integrated service and desktop grids for scientific computing
Professor Kacsuk Peter
Coordinator of EDGeS
EDGeS - Enabling Desktop Grids for e-Science
( European project with the aim of creating an integrated Grid infrastructure that seamlessly integrates a variety of Desktop Grids with EGEE type of service Grids.)
A Prototype for Detecting Defective Pills During Manufacturing
Prof Craig C. Douglas
Distinguished Professor of Mathematics Department, School of Energy Resources University of Wyoming and Senior Research Scientist of Computer Science, Yale University
ABSTRACT Contaminants and impurity formation in pharmaceutical products during manufacturing cause inconvenience and costly recalls. Determining when to adjust a process to control impurity formation, or halt production and destroy end products should be an automatic task using experts’ knowledge, data (audio and chemical spectral footprints), adaptive models and algorithms, and different scales (data and models). Integrated sensing and processing (ISP) optimizes sensing systems that integrate the traditionally independent units of sensing, signal processing, communication and targeting. By employing ISP, computational complexity within traditional sensing system is substantially reduced through determining efficient low-dimensional representations of those sensing problems that were originally posed in high-dimensional settings by traditional sensing architecture. An ISP-based imaging spectrometer produces detector signals directly correlated to desired sample information, obviating the need for post-collection chemometrics and converting data directly to knowledge. We apply ISP techniques to novel acoustic spectral sensors designed and created as part of this project. The ISP is reprogrammable on demand. It uses data libraries that are created offline. The creation is expensive computationally, but it makes the use of the sensor easy. We have to use high performance computing facilities to quickly generate new or modified libraries.
Keywords:Defect detection, manufacturing, integrated sensing and processing, dynamic data-driven application systems, DDDAS, cyber physical systems, CPS, parallel procesing, and algorithms.
Trends in High Performance Computing
Prof hai xiang Lin
Faculty of Electrical Engineering, Mathematics and Computer Science, Delft University of Technology
ABSTRACT In the past 3 decades, high performance computing has undergone a tremendous fast development. The achievement in high performance computing is obtained through the synergy in the development of integrated circuit technology, parallel computing systems and efficient algorithms. Performance of a single processor has been increased by a factor of 106 since the introduction of the first electronic computer in the late 1940s, the application of massively parallel processing adds an increase by another factor of 105, and moreover the development of fast and efficient algorithms further accelerate the solution speed by factor of 106 or more for many large and complex computer applications. The current trends in parallel architectures moving towards multi-cores, much larger number of processors or cores, and heterogeneity pose great challenges to software and tool developers, to algorithm and application programmers.
Keywords: Parallel and distributed computing, Multi-core, GPU, HPC systems and algorithms, technological development in HPC.