PARA 2008: Invited/Panel Speakers

Workshop on State-of-the-Art in Scientific and Parallel Computing

May 13-16, NTNU, Trondheim, Norway

Keynote and Invited Speakers:

	Dr. Chris R. Johnson Current positions at Univ. of Utah, USA: Director, Scientific Computing and Imaging Institute (SCI). Co-Director, Center for Integrative Biomedical Computing (NIH NCRR). Distinguished Professor of Computer Science Dr. Johnson is the PARA'08 Keynote Speaker " Large-Scale Scientific Visualization"
	Dr. Jack Dongarra Current positions include: University Distinguished Professor University of Tennessee Department of Electrical Engineering and Computer Science Distinguished Research Staff Oak Ridge National Laboratory Computer Science and Mathematics Division Turing Fellow University of Manchester School of Mathematics School of Computer Science Adjunct Professor Rice University Computer Science Department "Four Important Concepts to Consider When Using High Performance Computers"
	Dr. Kimmo Koski Current positions include: Managing Director CSC, The Finnish IT Center for Science, Finland since 2004 Chair HET (HPC in Europe Task Force) Previous positions include: Management positions at Nokia Reserach (4.5 years) 10 years at CSC "Case CSC Finland in building the European HPC Ecosystem">
	Prof. Bo Kågström Current positions include: Professor in Numerical Analysis and Parallel Computing, Department of Computing Science, Umeå University, Sweden Director Hign CPerformance Computing Center North (HPC2N) Björn Adlerborn, Robert Granat, Bo Kågström and Daniel Kressner: "New Parallel Variants of Dense Eigenvalue Solvers and Reordered Real Schur Forms"
	Prof. Syvert P. Nørsett Professor of Numerical Analysis, Department of Mathematical Sciences, Norwegian University of Science and Technology "Furiously Fast Fourier Transforms"
	Prof. Brian Vinter Professor of Computer Science at University of Copenhagen Previous Director of Nordic Grid Facilities "Not All Fun and Games"

Additional Invited Panel Speakers:

	Prof. Rene Belsø, Current positions include: Director Danish Center for Scientific Computing, DCSC, Denmark
	Dr. Sverker Holmgren, Current positions include: Director Swedish National Infrastructure for Computing, SNIC Associate Professor, Division of Scientific Computing, Uppsala University, Sweden
	Prof. Kristjan Jonassen Chair, Department of Computer Science, University of Island Previous positions include: Dept. of Mathematics, University of Island
	Dr. Jacko Koster Current positions include: Managing Director Uninett Sigma, Trondheim, Norway, Organization coordinating NOTUR II (Norway's Metacente for Computational Science, NorGrid (Norwegian Grid Infrastructure) and NorStore (Norwegian Storage Infrastructure)
	Prof. Dana Petcu Professor and Director of Computer Science at Western University of Timisoara, Romania Director of Insitute e-Austria Timisoara, Romania
	Dr. Olaf O. Storaasli Distinguished Research Scientist Computer Science and Mathematics Division Oak Ridge Bational Laboratories, Tennessee,USA

Abstracts

Chris Johnson: "Large-Scale Scientific Visualization"

Modern high performance computers have speeds measured in teraflops and handle data set sizes measured in terabytes and petabytes. Although these machines offer enormous potential for solving very large-scale realistic computational problems, their effectiveness will hinge upon the ability of human experts to interact with their simulation results and extract useful information. One of the greatest scientific challenges of the 21st century is to effectively understand and make use of the vast amount of information being produced. Visualization will be among our most most important tools in helping to understand such large-scale information.

Our research at the Scientific Computing and Imaging (SCI) Institute at the University of Utah has focused on innovative, scalable techniques for large-scale 3D visualization. In this talk, I will present state- of-the-art visualization techniques, including scalable visualization algorithms and software, cluster-based visualization methods and innovate visualization techniques applied to problems in computational science, engineering, and medicine. I will conclude with an outline for a future high performance visualization research challenges and opportunities.

Jack Dongarra: "Four Important Concepts to Consider When Using High Performance Computers "

In this talk we examine how high performance computing has changed over the last 10-year and look toward the future in terms of trends. These changes have had and will continue to have a major impact on our software. Some of the software and algorithm challenges have already been encountered, such as management of communication and memory hierarchies through a combination of compile--time and run--time techniques, but the increased scale of computation, depth of memory hierarchies, range of latencies, and increased run--time environment variability will make these problems much harder. We will look at four areas of research that will have an importance impact in the development of software.

We will focus on following themes:

Redesign of software to fit multicore architectures

Automatically tuned application software

Exploiting mixed precision for performance

The importance of fault tolerance

Björn Adlerborn, Robert Granat, Bo Kågström and Daniel Kressner: " New Parallel Variants of Dense Eigenvalue Solvers and Reordered Real Schur Form" Recent improvements to the QZ algorithm for solving generalized eigenvalue problems are reviewed. Among the major modifications are novel multishift QZ iterations based on chasing chains of tiny bulges and an extension of the so called aggressive early deflation strategy aiming at reducing the overall computation time on modern computing systems. A novel parallel implementation of a multishift QZ algorithm is presented. In addition, some inherent bottlenecks and possible remedies and improvements are discussed. The second part of the presentation will discuss parallel blocked algorithms for eigenvalue reordering in the standard and generalized Schur forms, filling another gap of the ScaLAPACK eigenproblem functionality. The algorithms are applied in parallel computation of stable invariant subspaces arising from algebraic Ricacati equations.

Kimmo Koski: "Case CSC Finland in building the European HPC Ecosystem "

During the last few years Europe has put together a number of grid infrastructure projects (DEISA, EGEE, ...) which have paved the way for pan-European ICT collaboration for science. One recent project example is a joint initiative with a target to provide European research access to top end computing facilities exceeding a petaflop/s performance. This Partnership for Advanced Computing in Europe (PRACE) collaboration has started a preparatory phase project, partly funded by European Union, and targets to launch the first petaflop center in 2010. To be able to serve the user community with varying needs efficiently, it is important to link the future top-of-the-pyramid resources to the full European HPC Ecosystem - including all elements for supporting a research project from actual hardware resources to competence in providing services for scientific computing.

The talk describes the HPC Ecosystem and gives an update to the European petaflop initiative. The new research infrastructures and their computing needs are discussed. CSC - the Finnish IT Center for Science is a partner in many of the EU projects building the European infrastructure collaboration and is used as a case example.

Brian Vinter: Not All Fun and Games"

High Performance Computing groups around the world has started looking into the use of game-consoles and high end graphics cards as inexpensive means for obtaining very high computing power. The desire for still more physical realism in computer games have created processors for game consoles, such as the CELL-BE and highly capable GPU such as the NVIDIA G92. The vendors of these systems have decided to open the platforms enough towards the user community that it is possible to program these chips for other purposes than "Fun and Games". This talk introduces some of the platforms that we to some extend today and most certainly tomorrow, may use to obtain a very attractive price/performance ratio for HPC applications. The talk will cover both the promises of these architectures and the rather rocky path to get to those promises. A number of real application experiences will drive the motivation of the talk.

Contact Information:

Please mail comments re. this page to:

E-mail: elster-at-idi.ntnu.no (replace -at- with @)

It was last updated on Jan. 31, 2008. Comments welcome.