In this issue, we would like to share with you a number of major developments which took place at SVU in the last few months. We would like to inform users that the SGI Origin2000, which had served us very well for many years, has been replaced with Itanium2 based servers with a 16 fold increase in computing capacity. Some performance figures of the new servers are highlighted in the first article. We have also doubled the size of the ACAD-HPC Linux cluster from 32 to 64 processors. Besides providing more resources for general academic computing, we have also web-enabled the StarOffice 7 Office Suite to run from the ACAD-HPC Portal. Details can be found in the second article. In the area of Grid computing, we had recently launched the Grid Innovation Zone (GIZ). We had also completed some applications testing and portal development work. Some user experiences in the testing of an application on the Campus Grid and the PC Compute Farm (PC Grid) are presented in the third and fourth articles. And, in the last article, we invite you to join us in taking a peep into the trends and directions of HPC developments taking place around the world.

Happy reading!

Together with the expansion of the ACAD-HPC Linux cluster from 32 to 64 processors, we have brought you the Sun StarOffice 7 Office Suite, an open source alternative to the Microsoft Office. It offers word processing, spreadsheet, presentation, drawing, and database capabilities.

Some highlights of the useful features for the Sun StarOffice software:

• Can read, write and edit documents from other office suite, including Microsoft Office.
• Save files to PDF with a single click, which enables convenient document-sharing.
• Export presentation to Micromedia Flash format, which can be displayed in almost any web browser.

Go to this website (http://wwws.sun.com/software/star/staroffice) to learn more about StarOffice 7 Office Suite. The software is web-enabled on the ACAD-HPC Portal, which means that you can run it anywhere anytime from a web browser. With this implementation, you can now run your research simulation, process data, write report and prepare presentation under the same Linux platform. To start this office suite, register/login to our ACAD-HPC portal (https://acad-hpc.nus.edu.sg/login), select StarOffice 7 from "My Unix Apps" to customize "My ACAD-HPC" and then click on "Launch StarOffice 7". New users of StarOffice 7 can click File--> Autopilot to get your fingers wet on how to make use of the powerful office suite.

In this article, the authors explore the use of computational fluid dynamics software in a grid computing environment. The problem domain being modeled is the left ventricle of the human heart. The generation of the computational meshes were performed over two clusters which are part of a prototype grid computing environment being set up at SVU. Please read on for more details of this exploration and direct any queries to Grace Foo at .

Shortening the time-to-solution is always an aspiration for researchers in their everyday simulation works. Here are two recent success stories we like to share with you. In both cases, postgraduate students Ma Shengwei and Phung Minh Hoang from the Faculty of Engineering are able to shorten the time-to-solution of their simulations significantly using the PC compute farm (or PC Grid). Please click here to read the details of the article and learn about their experiences. Feel free to contact us at should you have any enquiries.

The rapid development and adoption of the cluster computing technology in the past few years has a great impact on both the resource provider and users in the area of High Performance Computing. This trend is well reflected at the TOP500 Supercomputers site (http://www.top500.org/), a project started in 1993 to provide a reliable basis for tracking and detecting trends in high-performance computing. Twice a year, a list of the sites operating the 500 most powerful computer systems in the World is assembled and released. It is interesting to note that the first cluster system made its appearance on the June 1997 list. Today, there are 291 cluster systems listed in the TOP500 list released recently. The number is expected to increase in the future release.

How does the development affect the resource provider and users? For the resource provider, larger number of processors with much more affordable price come with a higher administrative overhead. For the users, with access to more distributed memory resources now, there is a need to consider a different programming model in order to make effective use of the parallel processing capability of the system. The most widely used programming model for cluster computing at the moment is the one based on MPI or Message Passing Interface. You should be able to find many sites providing online guide on MPI programming by doing an Internet search on "MPI programming guide"

At SVU, we provide three different types of cluster to cater for different types of computational load. We have an Intel Xeon based cluster with fast Ethernet interconnect to run serial and throughput jobs. We have a PC Grid that harnesses desktops idle cycles for parallel processing of jobs. Both resources are accessible through the ACAD-HPC Portal (http://acad-hpc.nus.edu.sg/acaddocs/). We are also providing a cluster with high-performance and low-latency Myrinet interconnect to run parallel applications (http://www.nus.edu.sg/Major/SVU/about_highend.html). A total of 216 processors are made available for research and academic computing through the three cluster systems. Today, more than 50% of usage (in term of CPU hours) at SVU are being generated on the cluster systems. Please contact us at if you like to explore any new research computing possibility using the cluster technology.