Official Launch of National Grid Pilot Platform

1 November 2003

The National Grid Pilot Platform (NGPP) is officially launched today by BG (NS) George Yeo, Minister for Trade & Industry. The NGPP is the first phase of a national cyber-infrastructure that links up compute resources in Singapore. The National Grid has the vision of transforming Singapore into a nation where computer resources can be connected together via a high-speed network. Sharing of resources in a secure, reliable and efficient manner by authenticated users for education, commercial, entertainment, R&D, national security and other purposes will improve the economic and technological competitiveness of Singapore.

The NGPP is co-funded by the two research councils of the Agency for Science, Technology and Research (A*STAR) - Science & Engineering Research Council and Biomedical Research Council, the Defence Science & Technology Agency (DSTA), the Economic Development Board (EDB), the Infocomm Development Authority of Singapore (IDA), the National University of Singapore (NUS), the Nanyang Technological University (NTU) and the Singapore-MIT Alliance (SMA).

Three A*STAR’s research institutes together with NUS, NTU and SMA have committed to share their technical computing resources on the NGPP. Today, the 1 Gbps high-speed network connects A*STAR’s research institutes and the two main universities, making available over 200 CPUs that provide about 750 Giga FLOPS (750 billion FLoating point Operations per Second) of heterogeneous computing resource for sharing.

Mr. Peter Ho, Chairman of the National Grid Steering Committee and Permanent Secretary for the Ministry of Defence said: “There is a growing demand for advanced computing resources, as our technology advances and more complex computational systems are developed. This Pilot Platform will provide the infrastructure that allows companies and researchers to solve problems beyond the capacities of their in-house computing resources. The National Grid will enhance the technological edge of Singapore and attract more high value R&D investments.”

Nobel Laureate Dr Sydney Brenner, Deputy Chairman of National Grid Steering Committee and Chairman of the Biomedical Research Council of A*STAR commented: “We need to continuously look ahead to be ready for next wave to maintain Singapore’s position as a technologically progressive connected hub
The National Grid recognises the potential of networked distributed computing and resources. It is the solution that would enable efficient resource utilisation and sharing to fully exploit the connected virtual community”

Several Grid Computing applications are in use today. They include:

Collaborations with industry partners have been established to extend the grid resources to them. For example, IHPC is collaborating with BAE Systems, Rolls- Royce and several UK research institutions to perform engineering simulation of complex systems and engines. Local interest in the grid is also displayed by the partnership between ST Engineering, IBM and IHPC to develop a testbed to demonstrate the feasibility of Grid Computing to virtualize technical computing resources to geographically distributed design and engineering units. For biomedical arena, Bioinformatics Institute (BII) has also formed a partnership with University of California, San Diego and San Diego Supercomputing Centre, to develop an “Encyclopedia of Life” (EOL). The research collaboration is conducted using grid computing. EOL will catalog the database of complete proteome of every living species, allowing calculation of 3D models and assignment of biological functions for all recognisable proteins. This system will enable efficient data management and exchange and allow novel queries from users.

Professor Peter Cowley, Chief Scientist of Research and Technology, Rolls- Royce plc., said “IHPC has World Class Computing facilities and they have very capable and helpful staff. Rolls-Royce is interested in research where there is a clear route to industrial implementation and Singapore has an excellent reputation for linking research activities to business needs. In the case of our collaborative work on intelligent agents to support diagnostics and repair scheduling, we hope eventually to link the technology with 'real world' repair and overhaul activities. Across all these activities, the Grid will help us to work more closely with global partners starting with research on the use of Grid technology itself, and eventually looking at more general applications.”

The NGPP has also received strong support from the ICT vendors. Cisco, Starhub, Singapore Computer Systems, Dell, IBM, HP, and Sun Microsystems have contributed equipment and services worth several million dollars.

The National Grid operates under a consultative and participative model. It welcomes the participation of stakeholders and partners to realize a Grid-enabled economy in Singapore.

The next phase of the National Grid will include improving the security, quality of service and Grid services on the NGPP as well as extending connectivity to other institutes of higher learning (such as the polytechnics), industry, schools, and hospitals.

Annex 1: About National Grid Office
Annex 2: National Grid Steering Committee members
Annex 3: Information on existing grid projects

For more information, please contact:

Dr. Lee Hing Yan
Project Director, NGPP
& Deputy Director, National Grid Office
21 Heng Mui Keng Terrace
Singapore 119613
Tel: (65) 6874-7863
Fax: (65) 6872-1361

Ms Ng Hwee Lin
Senior Officer
Corporate Communications, A*STAR
Tel: (65) 6478 9593
Fax: (65) 63375360

---

Annex 1: About the National Grid Office

The NGO was established in January 2003 to promote Grid Computing and to develop a cyber-infrastructure that steers Singapore towards a Grid-enabled economy where computing resources, services and intellectual property can be provisioned securely on a high-speed network.

The roles of the National Grid are:

  
The NGO strives to achieve the vision for National Grid by the following means:

• Formulate the framework & policies
• Plan & develop a secure platform
• Adopt common open standards
• Encourage the adaptation of Grid Computing
• Demonstrate the commercial viability of compute-resource-on-tap

Lay the foundation for a vibrant Grid Computing economy
The NGO reports to the National Grid Steering Committee (see Annex 2).

Advising the committee on technical matters are the Working Groups (WGs) and Virtual Grid Communities (VGCs). The WGs comprise industry practitioners, academics and researchers who volunteer their time and expertise to provide technical advice. The current WGs focus on Grid Networks, Grid Middleware, Grid Security, Grid Applications, and Governance. The VGCs are distributed collaborations centered on a Grid-enabled environment. Such communities of researchers and users have been empowered to innovate and eventually revolutionize what they do, how they do it, and who participate. The current VGCs are Life Sciences and Physical Sciences.

Plans are underway to form VGCs in Digital Media and Manufacturing.

Annex 2: National Grid Steering Committee members

Chairman Mr Peter Ho
Permanent Secretary, Ministry of Defence

Deputy Chairman Nobel Laureate Sydney Brenner
Chairman, Biomedical Research Council, A*STAR

Members

• Mr. Boon Swan Foo, Managing Director, Agency for Science, Technology & Research (A*STAR)
• Mrs. Tan Ching Yee, CEO, Infocomm Development Authority
• Mr. Quek Kim Pew, Deputy CE (Technology), Defence Science & Technology Agency
• Prof. Tan Chorh Chuan, Director (Medical Services), Ministry of Health
• Mr. Tan Chek Ming, Assistant Managing Director, Economic Development Board
• Mr. Alex Siow, Vice President, StarHub Pte Ltd
• Prof. Lam Khin Yong, Director MDO, Agency for Science, Technology & Research (A*STAR)
• Mr. Timothy Cooley, IT Director, Lilly System Biology
• Prof. Lee Tong Heng, Vice President (Research/Science, Engineering & Humanities), National University of Singapore
• Prof. Er Meng Hwa, Deputy President, Nanyang Technological University
• A/Prof. Kong Hwai Loong, Executive Director, Biomedical Research Council , A*STAR
• Prof. Chong Tow Chong, Acting Executive Director, Science & Engineering Research Council, A*STAR
• Prof. Lawrence Wong, Executive Director, Institute for Infocomm Research (I2R)
• Dr. Lim Khiang Wee, Director, Science & Engineering Research Council, A*STAR
• Dr. Gunaretnam Rajagopal, Acting Director, Bioinformatics Institute (BII)
• Dr. Kurichi Kumar, Deputy Executive Director (Industry), Institute of High Performance Computing (IHPC)
  

Annex 3: Information of Grid projects

1.Geo-rectification of Satellite Images for Environmental Monitoring

Raw satellite images of the earth require the correction of skew caused by the earth's curvature. This is an essential step for visualization in environmental monitoring such as detecting the distribution and variability of phytoplankton to better understand ocean primary production and global biogeochemistry, concentration of plant pigment (including chlorophyll) in the water, monitoring oil spills, disaster mapping & response (flood, earthquakes & red tides) and forest fire response.

Industrial applications that can tap the utilization of remote sensing imagery over a Grid environment include urban planning (such as airborne or satellite survey systems and mapping) as well as for manging earth resources (such as geologic mapping & exploration and vegetation mapping).

Geo-rectification is a compute-intensive process. Grid Computing enables independent tasks to be separated and executed concurrently - tapping unused resources over the Grid to provide the speed-up in image geo-rectification and data visualization for analyzing constant changes in environment. The georectification process involves several steps including image sampling, resolution conversion and image matching before the final image is produced. The execution of the geo-rectification was performed over the National Grid Pilot Platform using NUS Grid resources which consists of 8 Intel Xeon 3.06 GHz using Globus 2.4 and LSF job manager.

Contact: A/Prof. Teo Yong Meng, NUS (teoym@comp.nus.edu.sg)

2.Computer-Assisted Cel Animation (CACAni)

The objective of CACAni is to design and apply computer graphics and imaging technology to develop an advanced 2D animation system for the digital media industry. The thrust is to increase the creativity and productivity of artists by significantly cutting down time and labour cost, especially on frame drawing and painting.

Given two key frames drawn by artists, the CACAni is able to automatically generate a user-specified number of in-betweens. It can also propagate color information from a color frame to its uncoloured successors in a sequence. The CACAni is applicable to various types of input frames, from line drawings to grayscale, from the black & white to the true color.

The potential of this application is targeted at the media industry and may include animation studios, game developers, 2D and 3D software designers, and individual animation fans.

A reasonable smooth effect of motion requires 25 frames per second. For a feature film of one and half hour, it needs 135,000 frames! In the case of inbetween frames for a pair of key frames, up to 27,000 keys need to be drawn by animators, manually, and the rest of 108,000 frames will be automatically generated. Under this circumstance a significant amount of computing power is required to handle all interpolation in parallel.

Therefore a cluster of computers networked and synchronized by a Grid engine is the ideal platform for this project. language. Pairs of key frames are submitted to all Grid machines available across the National Grid Pilot Platform before the interpolated results are collated afterwards.

Contact: A/Prof. Seah Hock Soon, NTU (ashsseah@ntu.edu.sg)

3. Distributed Dissipative Particle Dynamics (DPD) Simulation

DPD is a relative new mesoscale simulation technique; its mean quantities satisfy the conservation laws exactly. This project aims (a) to create DPD models for DNA molecules, blood cells and micro vessels and (b) to explore in particular bio-fluid flows in micro channels and simulate micro cell trapping and micro filtration.

DPD as a tool helps exploring the micro-rheological properties of complex fluids and the behaviour of micro flow and obtaining detailed information, such as macromolecular motion in micro channel flow, for better design of BioMEMS devices. DPD also helps researchers to understand the relation between red blood cell coagulation, aggregation and flow behaviour in micro vessels.

DPD simulation is a compute-intensive process. Grid computing enables each flow domain to be divided into several subdomains which is then assigned to compute resources in the Grid environment where the simulations of particle flows are executed concurrently. One to three millions particles is required to model a micro vessel filled with fluid. The simulation has to run for one million time steps to be statistically significant.

In the Grid environment, the simulation runs were executed concurrently on SMA, NUS Supercomputing Visualization Unit and IHPC Grid resources comprising HP rx5670 (60 CPU 1GHz Itanium-2), 8 Intel Xeon 3.06 GHz and IBM Regatta (16x1.3 GHz Power4 CPU) using Globus 2.4.

Contact: Prof. Phan-Tien Nhan, NUS & SMA (nhan@nus.edu.sg)

4. Distributed Simulations of Flows over Dimpled Surfaces

Flows on dimple depressions have attracted increasing attention in recent years.

The flow structures induced by dimpled walls have been recognized to have various industrial applications. For example, dimples on a golf ball help to improve the ball's flying range. In aerospace, turbine blades with dimpled surfaces can be cooled down much easier.

Dimpled wall structures on the blades may also reduce total drag, therefore, greatly saving energy consumptions and increase air or marine vehicle's performance. Russian researchers have shown that surface dimples may alter near wall turbulence and lead to drag reductions.

At current stage, we have successfully performed simulations of laminar flow and turbulent flow over multiple dimples. Reduction of total drag can be obtained for laminar flow under certain flow conditions.

Grid-based computing, involving clusters within NGPP, has been exploited to cope with the intensive computation required. For our execution, our codes have been successfully tested across Singapore MIT Alliance (SMA) PC Clusters Hydra I (Pentium) and Hydra III (Itanium) using Globus.

Contact: A/Prof. Boo-Cheong Khoo, NUS & SMA (mpekbc@nus.edu.sg)

5.GridBLAST for Applications in Life Sciences

Bioinformatics has made significant contributions by offering easy-to-use tools to compare genes with each other. One particular tool called BLAST has been especially successful. The throughput of currently available BLAST tools is limited because they can usually analyse only one gene at a time. Our solution would scale with rapidly increasing demand for very compute and storage intensive applications of BLAST, including a user interface that is optimized for the analysis of large sets of BLAST search results.

This kind of high throughput similarity search can be useful to understand a large set of sequences, including (a) the level of redundancy in the set; (b) the identification of clusters of very similar sequences; (c) to determine which sequences are novel and which ones are already known; (d) discovery of interesting patterns in large sets of sequences, e.g. in comparative genomics analysis.

Implementing a high throughput version of BLAST on a cluster is fairly easy, given its embarrassingly parallel nature. This is an SPMD (Single Program Multiple Data) problem where the same sequence of instructions is carried out on different sets of data. Scaling the application to run on a Computational Grid needs to address challenges such as different geographically distributed compute nodes not sharing a file system. The databases and executables required for carrying out a BLAST run may not be available on the remote nodes.

The solution is to effectively and efficiently staging the necessary files (including the executables, databases and query files), running the BLAST searches on the remote machines and then gathering the results generated back to the local machine.