We had had a good year in 2004 in terms of High Performance Computing and Grid Computing development in NUS. Some of the milestones reached include the setting up of the Grid Innovation Zone at NUS (GIZ@NUS), the launch of TCG@NUS, one of the largest PC Grids in the region, and achieving a record of more than 1000 registered users at SVU. GIZ@NUS, a collaboration with IBM and Intel, is an initiative aimed at providing expertise and resources to support Grid computing development in NUS and the region. TCG@NUS, a collaboration with Singapore Computer Systems, is another initiative to create a new research computing capability at NUS by harnessing the idle compute cycles of PCs across the campus. Our focus in the coming year will be on delivering more resources, new technologies and services to our users. We will also explore more collaboration with researchers, other HPC resource owners and industry partners to further enhance NUS leadership in High Performance and Grid Computing. In this issue of HPC@NUS, we will share with you the HPC usage profile at SVU and some projected trends (Review 2004). We have also included tips on using the Abaqus software (Using Abaqus Environmental File) and text editing tools (Simple Ways to Edit Files on SVU Hosts), and updates on the Grid middleware (Highlights from GlobusWORLD 2005). There is also a preview of a new visualisation technology (New Version of the Remote Visualisation Software).

We hope you will enjoy reading this issue of our newsletter!

Have you ever wondered how the HPC resources are used at SVU? Do you like to know who else are doing similar type of research or using the same type of application software so that experience can be shared? Would it be nice to know the peak period so that you can schedule your work to avoid it? Do you like to know whether you are using the right resources or making full use of the services provided by SVU? How about having a preview of the new technologies coming your way? If your answer is yes, please read on ...

New Abaqus users have often approached SVU for help on problems related to their job runs which can easily be solved if they know how to manipulate the requirement parameters for their Abaqus analysis jobs. A check with advanced users will tell you that there is an environment file that you can create which is then read by Abaqus before your job run to set the appropriate parameters for your job, for example, pre_memory, standard_memory, scratch directory or even simple tasks like disabling the prompt to ask you to delete old data files. The confusing part is that the instructions for using the environment file are found in the Abaqus Installation and Licensing Guide, something which normal users would skip. Click on our Abaqus online help (http://panther1.stf.nus.edu.sg:2080/v6.4/), then click on "ABAQUS Installation and Licensing Guide" followed by "Customizing the ABAQUS environment". Below are some of the usual parameters that users would use in their environment file (abaqus_v6.env):

• pre_memory=256 (default value is 256MB)
• scratch=/scratch (scratch files are stored in /scratch during runtime)
• standard_memory=256 (default value is 256MB)
• ask_delete=OFF (do not prompt to delete older data files, caution: cannot undelete)

Users are encouraged to read and try out the usage of the abaqus_v6.env file for their job runs, especially when prompted to do so (i.e. to modify certain parameters) by the job's .dat file. More tips on Abaqus or other applications in the next issue of HPC@NUS.

For many, grids and grid computing is synonymous with the Globus Toolkit, the middleware that make grids possible. Grids first emerged in academia/research, as scientists and engineers across the US sought to access scarce high-performance computing resources concentrated at a few locations. The Globus Project to develop the Globus Toolkit, initiated in 1996 by researchers at Argonne National Laboratory and University of Chicago, now has developers from all over the world contributing to its evolution. Since its release in 1998, the Toolkit has become (arguably) the most widely used grid middleware. GlobusWORLD, the premier grid computing conference featuring the Globus Toolkit, has been held annually since 2003. Some highlights from this year's conference are presented.

Whether you are developing your own High-Performance Computing (HPC) application or just simply running a commercial off-the-shelf application software, there is no running away from using some form of command-line commands. This is because the HPC systems in use today are pre-dominantly Unix/Linux based (as shown in the top500 list at www.top500.org), and access to most Unix/Linux systems are non-GUI (Graphical User Interface) based. For Windows users who are used to the point-and-click GUI, accessing the Unix/Linux systems for the first time through command-line can be frustrating. But not to worry, you can find most of the basic commands needed to get started on the SVU's systems at http://www.nus.edu.sg/comcen/svu/techinfo/unixcom.pdf. In this article, we will provide more tips on file editing/management, a basic and essential task for most users. Click here to view the full article.

The remote visualization system software will be upgraded from its original beta version (aka Golden Gate) to its latest version 2.1, with a new name called HP Remote Graphics Software. The Remote Graphics Software is a combination of the remote 3D graphics (the Golden Gate) and another HP product called HP Remote Workstation. Apart from the improvements in the performance and image quality, another major change in the new product is related to user authentication. With the new single sign-on feature, pre-access booking of time slot will no longer be necessary. In addition, the new SSL encryption support for authentication and keyboard input transactions will ensure secure transaction between the client and the server systems. Click here to get a more detailed preview of the new software.