Power sharing makes light work of data-heavy tasks
5 Nov 2003

The promise of efficient data management and the quick calculation of complex computational problems has various organi-sations flocking to the National Grid for spare computing resources.

The National Grid Pilot Platform has had an enthusiastic response from various organisations and companies that want to use its cyber infrastructure for research. There are 10 pilot projects, but not all are live yet. Some of the institutions are now writing software to grid-enable their applications, so that they can run on the Pilot Platform.

Here is a brief snapshot of some of the projects.

• Monitoring the weather

Satellite images, beamed back to earth several times every day, help scientists either prevent or quickly detect environmental disasters such as forest fires, earthquakes and oil spills.

Scientists in Singapore are using satellite images to monitor an aquatic pheno-menon known as the 'Red Tide' which can kill millions of fish. Every day, hundreds of images are beamed from orbiting satellites to receivers at the National University of Singapore's Centre for Remote Imaging, Sensing and Processing (Crisp).

The raw images, however, are difficult to read as they appear 'curved' due to the spherical shape of the earth. Computational power is used to correct this curvature through a process known as geo-rectification. It takes up to 20 minutes for a single Pentium 4 processor to rectify each image.

But scientists at the School of Computing (SOC), NUS, have developed a grid application that can break up each image into segments and give each segment to a processor on the grid. 'As a result what took 20 minutes can be done in a fraction of the time,' said Associate Professor Teo Yong Meng of the SOC, who heads the project.

• Drawing faster

The Distributed Computer-Assisted Cel Animation ven-ture by Nanyang Technology University seeks to increase the creativity and productivity of artists by cutting down time and labour costs associated especially with frame drawing and painting. This is achieved by sending pairs of key frames drawn by artists to different compute resources to automatically generate user-specified scenes.

• Boosting research on strokes

A project called distributed Dissipative Particle Dyna-mics (DPD) aims to create software models for DNA molecules and blood cells to explore the flow of fluids in these channels.

As DPD simulation is a computer-intensive process, grid computing enables each flow domain to be divided up and then assigned to compute resources where the simulations of particle flows are executed concurrently. The study can help boost research on strokes. The Singapore MIT Alliance, NUS and the Institute of High Performance Computing (IHPC) are carrying out the task.

• Flows over dimpled surfaces

Dimples on a golf ball help to improve the distance it can fly when it is hit. Industrial applications have since recognised that dimpled surfaces can influence flow structures. For example, in aerospace, turbine blades with dimpled surfaces can be cooled down more easily. This is an assignment by Singapore MIT Alliance and the IHPC to simulate flow structures.

• Finding cures for diseases

Undertaken by the BioInformatics Institute (BII), this project seeks to speed up gene analysis using a technique called Blast. A software application, Blast helps scientists match genes they are studying against a database of known genes to find similarities, which can help in drug research or help find cures for diseases. It takes a single computer a long time to run blast searches, as the software must sift through databases containing millions of genes. Grid-enabling blast applications now allow the same task to be com-pleted in much less time.

• Discovering new drugs

A collaboration between Singapore Polytechnic and Keio University of Japan, this project aims to develop an e-Cell gene simulation system on a multi-processor platform. The collaboration is to jointly develop a simulation that will run on the National Grid although it is currently running on a single processor platform.

According to Dr Ng Yan Hong at the polytechnic, an e-Cell is an electronic simulation of a cell with only 127 genes, which was deve-loped by scientists at Keio University, to study cell behaviour on computers rather than under microscopes in the labs. The aim of the e-Cell simulation is to help scientists discover new drugs. - Radha Basu with inputs from the National Grid Office

[Source: Computer Times, 5 Nov 03]