APPLICATIONS

The National University of Singapore (NUS) and the Singapore Computer Systems (SCS) signed a deal today towards building a supercomputing Grid, aiming to reduce the time spent on performing complex academic calculations. The Tera-Scale Campus Grid at NUS initiative will initially comprise 800 university-owned computers as nodes. By tapping the computer power from its 800 desktops, NUS' grid is expected to reach a theoretical peak performance of between 2 to 6 teraflops, the university claimed. "This new computing capability will enable NUS researchers to explore new areas of research, to widen their research scope, or just simply to generate results faster and more accurately," said NUS provost professor Tan Chorh Chuan. He added, "You can see it as a major resource for NUS, part of our research infrastructure, because it allows our academics access to very big computing power in order to solve problems. At the same time you can also see it as our way of developing it as a test bed for more applications in order to advance the technology." The Grid will be used in areas such as bioinformatics, life sciences, financial simulation and any study that requires multiple runs of the same programs using different variables, he added. Last November, Singapore's National Grid Office (NGO) unveiled the National Pilot Grid Platform, an initiative for local universities and research to share resources. In July, Oracle and the Infocomm Development Authority of Singapore agreed to invest US$15.1 million to help local IT firms develop enterprise grid solutions. "We are seeing the increasing use of grid technology in areas such as digital media, life sciences and engineering research and development," said Pek Yew Chai, president and CEO of SCS. "We already have customers who are exploring with us the ability to link their in-house Linux clusters to a central grid resource for additional computing power when they exceed their peak capacity." The initiative entails a US$151,000 investment each from NUS and SCS. As part of the deal, the latter will also receive 100,000 CPU hours worth of processing power over two years. According to Tommy Hor, director of NUS' Computing Center, this software expenditure is estimated to save US$1.2 million to US$1.8 million in equivalent hardware for the university. He added that NUS plans to expand the number of nodes to over 1,000 within two years.