.comment-link {margin-left:.6em;}


Wednesday, July 26, 2006

Windows Compute Cluster Server 2003

Microsoft has released its server application for supercomputing clusters, Windows Compute Cluster Server 2003 based on the new 64-bit operating system. Trial versions of Compute Cluster Server can be downloaded from Microsoft site. The company expects to start distributing volume licenses in August.

The system will allow users to exploit high performance computing power from their desktops using the familiar Windows desktop environment. Microsoft expects the new system to appeal to researchers in areas as diverse as biological and medical research, advanced engineering, processor-heavy visualisations, and even financial market predictions.

The Server 2003 has already been deployed at several sites for a variety of computationally intensive applications. The Computational Biology Service Unit in Ithaca, New York, attached to Cornell University, has been using the system for research activities in bioinformatics, including sequence-based data mining, population genetics and protein structure prediction. Northrop Grumman is using the new system for work in its Space Technology sector, supporting work for its contracts from the United States government for military and civil space systems. Other users include National Center for Atmospheric Research, Tokyo Institute of Technology's Global Scientific Information and Computing Center, French firm AREVA-Challenge, BAE Systems, CASPUR of Italy, Petrobras (the Brazilain Government's oil company), Queen's University of Belfast in Northern Ireland.

Windows® Compute Cluster Server 2003 is also the underlying operating system for a new High Performance Computing cluster at the National Center for Supercomputing Applications(NCSA) that recently achieved 4.1 teraflops on 896 processors.

Wednesday, July 19, 2006

International Math Olympiad

The 47th International Mathematics Olympiad was held between July 10 and July 16, 2006, in the Slovenian city of Ljubljana.

Zhiyu Liu of China, Iurie Boreico of Moldova and Alexander Magazinov of Russia got perfect score of 42 (for solving 6 problems) to become Top Gold medal winners. Caili Shen of China was at 2nd rank scoring 37 followed by Przemysław Mazur of Poland who scored 36.

Here are 2 relevant links: Problems, Results.

Wednesday, July 12, 2006

Maple Toolbox for Matlab

Maplesoft has come up with a great product that users of both Maple and MATLAB were trying to have for a long time to receive the benefit of both by combining all advantages of symbolic and numeric computation offered by these two softwares.

Maplesoft has announced Maple™ Toolbox for MATLAB® to extend their helping hands to the mathematics community. It consists of two components, Maple-10 and the Maple-MATLAB Connector. With this toolbox, Maplesoft offers a technical computing solution that is tightly integrated with MATLAB, providing direct access to all the commands, variables, and functions of each product while working in either environment. MATLAB-users get direct access to Maple from within MATLAB. It can access to over 50 interactive assistants and tutors to aid rapid development of solutions fully compatible with all MATLAB applications that use the Symbolic Math Toolbox (Standard and Extended) without any reworking needed.

The following are the key Features of the toolbox: Access to full Maple; Interactive technical document interface with intuitive 2-D equation editor for typeset-quality mathematical presentation; Up to 50 times faster than the Extended Symbolic Math Toolbox. Additional features include: Statistics, Units, Optimization, Tolerances, Enhanced graphics, Code generation to 5 languages (C, FORTRAN, Visual Basic, Java and MATLAB).

A public beta version of the Maple-MATLAB Connector is currently available at: http://www.maplesoft.com/products/MapleMATLAB. We used it and we just fell in love with it. Hats off to Maplesoft for addressing this long-time request of the mathematics community and bringing this out just at the beginning of summer. We'll find enough time to play with it before schools start and then apply it in our classes.

Thursday, July 06, 2006

New Version of Star-P

Star-P developed by Interactive Supercomputing is an excellent interactive parallel computing platform . It allows scientists, engineers and other researchers to code algorithms and models on their desktops using familiar mathematical software packages such as The MathWorks’ Matlab, and run them instantly and interactively on parallel servers. This eliminates the need to re-program the applications in C, Fortran or MPI languages to run on parallel computers. Such process of re-programing may sometimes take months depending on complexity of the algorithm and how computationally intensive those problems are.

The new version of Star-P 2.3 (announced on June 20) is designed to make high performance computing more economical and accessible for a wider range of technical computing users. It runs on lower-cost AMD Opteron-based systems. Users who previously didn’t have the resources to tap the power of parallel computing clusters can now tackle much larger problems on their desktops, while arriving at a solution in a fraction of the time.

The new 2.3 version will be available from SGI’s Professional Services team and distribution partners. For high-end parallel processing needs, Star-P also runs on Itanium-based SGI Altix servers supporting from one to 512 processors, 24 terabytes of memory, and running 64-bit Linux. Star-P automatically connects desktop applications to Opteron-based servers and parallelises the application code on the fly, enabling users to scale their applications across any multi-processor system or parallel cluster in real time.

Star-P is a great tool that computationally-intensive industry applications may utilize to improve their overall efficiency, even without having resources for parallel programming.