on pro

shop brought together researchers whose primary work has been developing parallel hardware and software systems. The discussion of grain size in parallel machines sharply divided the audience into "coarse grain" and"fine grain" advocates. Those in favor of coarse grained processors based their arguments grammability, while the fine grained spokesmen want to exploit the concurrency and cost efficiency available infine grained machines. However, the difference between the coarse grained and the fine grained designs in question differ by less than one order of magnitude in memory capacity per node.

Stephen W. Keckler is a Ph.D student at the MIT AI Laboratory. He worked as a circuit designer at Intel on the 386SL microprocessor. His

research interests include comparallel puter architecture, processing, and VLSI and system design.

Keckler received a BS degree in electrical engineering from Stanford University and an MS in electrical engineering and computer science from MIT. He is a National Defense Science and Engineering Graduate Fellow, and is a member of the IEEE, Sigma Xi, and Phi Beta Kappa.

Dr. William J. Dally received a B.S. degree in electrical engineering from Virginia Polytechnic Institute in 1980, an M.S. degree in electrical engineering from Stanford University in 1981, and a Ph.D. degree in computer science from Caltech in 1986. From 1980 to 1982 Dr. Dally worked at Bell Telephone Laboratories where he contributed to the design of the BELLMAC-32 microprocessor. From 1982 to 1983 he worked as a consultant in the area of digital systems design. From 1983 to 1986 he was a research assistant and then a research fellow at Caltech. He is currently an associate professor of computer science at the Massachusetts Institute of Technology. Dr. Dally's research interests include concurrent computing, computer architecture, computer-aided design, and VLSI design.

by David K. Kahaner

INTRODUCTION

Earlier reports have discussed the computing and factory automation situation in Singapore,

["singapor", 19 Jun 1991;
"dattab", 16 Sept 1991;
"iccim.91", 4 Nov 1991]
(E-mail:

Kahaner@cs.titech.ac.jp)

as well as the general science structure of this small country.

Since the 1950s, Singapore's annual per capita income growth has averaged more than 6%. The country began building a manufacturing base in the late 1950s based on textiles, moved upstream to simple electronics, and then to advanced electronics. It has made major gains as a financial center and is currently looking at biotechnology as a new growth industry. Actually, there is some indication that Singapore's income growth has been as much a result of the generous subsidies to foreign corporations that invested there as they have to productivity gains in manufacturing. Foreign investments have been huge. In 1992, U.S. firms will invest almost US$1B, an increase of over 30% from a year

earlier. Estimates of U.S. investments in other southeast Asian countries show that only Indonesia ($1.9B) and Malaysia ($1B) will have a greater infusion of U.S. capital.

Computer technology is very heavily infused into Singapore's infrastructure. A new project began this year is the development of a manufacturing information network (Manunet). Starting first with metal and plastics fabrication, it will comprise updated information on automation products, suppliers, technologies, and patent details. Manunet will be expanded to include manufacturing services for

engineering design. There are almost 3000 manufacturing companies in Singapore, most small-to-medium sized. These companies will invest more than $1B U.S. on automation within the next 2-3 years. The development of Manunet is to be done at the new Institute of Manufacturing Technology (IMT).

INSTITUTE OF MANUFACTURING TECHNOLOGY

In my earlier report [iccim.91], I described GINTIC (GINTIC Institute of CIM), a new computerintegrated manufacturing training

and research facility on the campus of the Nanyang Technological University (NTU). Now NTU has established IMT as another new institute, focusing on manufacturing processes and manufacturing automation. I did not go to IMT (it is probably too early to see much) but was briefed about it by Dr. Beng Siong Lim, of GINTIC[GBSLIM@NTUVAX.BITNET]. Like GINTIC, IMT will provide (graduate) training as well as research and development in an industry-driven way. For example, one immediate task is to work with computer disk drive manufacturers to improve that industry's manufacturing, testing, and maintenance infrastructure. IMT has also identified these ten projects of interest.

Disk drive head mounting, alignment and fabrication Multilayer ceramics Electron microscopy of large plate weldments Near net shape precision casting and molding Waterjet cutting Laser machining

Robotic vision-assist welding Crane robots

Advanced surface technology,

including SMT technology Garment material optimization.

The philosophy of research is interesting. I quote from their brochure. "Postgraduate research experience at IMT need not necessarily involve hi-tech, in the sense of the latest, glamour-type research (e.g. robots, AGV (autonomous guided vehicles) superconductivity, etc.) but enabling technologies. We have to bear in mind that new markets taken in by the multinational companies require substantive support from home grown small- and medium-sized enterprises, particularly in well developed and proven techniques of established manufacturing, such as casting, heat treatment, automation interfaces, part feeders-we have not yet reached the critical mass in this know-how. Some of them have evolved from hundreds of years into reliable, proven techniques of mass production."

A building is being constructed for IMT on the NTU campus, and the first part, to be opened in 1994, is budgeted at about $6M U.S. Eventually IMT will cover about 12,000 m2 of lab space and workshops and will have a research staff of 200. IMT's model is the German Fraunhofer Institute (IPA), using its operational policies such as industrial-oriented research and active interchange of staff with industry in Singapore. There is also a plan for significant interchanges with Germany, particularly with the State of Baden-Wurttenburg.

I expect to visit IMT during my next trip to Singapore. In the meantime, further information can be obtained from the following:

Director, Institute of Manufacturing Technology Nanyang Avenue

SISIR has missions somewhat similar to the U.S. National Institute

of Standards and Technology (NIST), the former National Bureau of Standards (my home organization). In fact, in the Measurement Standards Center, Dr. Sze Wey Chua has has been been working with scientists at NIST on voltage standards through sponsorship from the United Nations. He also explained to me that SISIR has MOUS (memoranda of understanding) with China (PRC), Taiwan, and Australia on related work. SISIR also has an MOU with the British Standards Institute (BSI), based on the International Standards Organization (ISO) 9000 standards.

SISIR has a staff of about 450 and is 80% self-financed. Its programs are divided into two areas: hard technology-including contract R&D, process or product design/development, consulting/training, testing, evaluation and failure analysis; and soft technology-including standards/certification, quality assessment issues, accreditation, technology transfer, and incubation.

In addition to the usual R&D associated with measurement standards, SISIR is building a food biotechnology center and an electromagnetic compatibility test center. There are already centers for chemistry, polymers, metals and advanced materials, surface and particle technology, mechanical technology, information technology, electronics testing, and design and development. Singapore will soon join Japan as the only Asian countries with accreditation from the Electrotechnical Commission, meaning that Sinaporean electronic products will be much more readily accepted in the West (especially Europe).

Although I received a briefing on SISIR's overall activities, I was most interested to meet scientists with computer interests. I found two. Dr. Henry Hechneng Sun (Mechanical Technology Center) has been doing finite element analysis on scaffolding in shipyards and also analysis of blast deflectors inside jet engines. This involves using various engineering analysis packages; Dr. Sun is not developing his own software. Like many Singaporean scientists, Sun spent time in the United States, at NASA and the University of Cincinnati.

Dr. Tao Zhuang, originally from the PRC, worked on nonlinear fracture mechanics in China and at Syracuse University. His main interests are in crack propagation perpendicular to interfaces of laminated materials, and his work has been applied to studies of armor plates in the PRC. This type of analysis is highly nonlinear and can benefit from large-scale computer modeling, as well as serious analysis. Unfortunately, SISIR is heavily contract-oriented, and there is not much call for this kind of research. (I feel that there are good opportunities for collaboration with Western

researchers who can provide Zhuang with the facilities he needs.) Lately, he has been working on laser welding for the aerospace industry and on various crack testing (lowtemperature environment) strategies. One of his projects is for the Chicago Bridge and Iron Company.

I was surprised to see the significant number of papers on industrial automation that were presented at IA'92 by SISIR researchers (see below).

Finally, I should note that like other Singaporean government organizations, SISIR has an astonishingly complete and coherent set of English language documents. Perhaps it is not surprising, but English is the language of business here, and it is probably easier for English speakers to operate in Singapore than anywhere in this part of Asia. Of course, many people also speak Malay, Tamil, or one of Chinese languages.

2ND ASIA-PACIFIC INDUSTRIAL AUTOMATION CONFERENCE (IA'92)

Singapore has established itself as the international conference center of southeast Asia. Although the first IA Conference was held in 1990, there have been many computer-related meetings and expositions since then, including a Computer Integrated Manufacturing (CIM) conference that I reported on last fall. The conference was held in Singapore's World Trade Center. There was a large exhibition at which a variety of vendors displayed flexible manufacturing systems, systems, robots, storage and retrieval systems for factories, pneumatic and hydraulic components and systems, CAD/CAM packages, and test measurement and control equipment. There was also an impressive technical program of about 75 papers in two parallel tracks extending over three days.

Not unexpectedly, the featured speakers were all from the West-U.S., U.K., Sweden, and Australia. It was also true that the most fascinating work described was from the West. Singapore has made remarkable progress in its manufacturing capabilities, but it is now trying to develop its own expertise. As mentioned above, there has been a tremendous amount of foreign investment but not always transfer of technology. Most of the manufacturing activity in Singapore is from small- to medium-sized companies, with little incentive to conduct basic research. The National University of Singapore (NUS) has a manufacturing research component, but many of the papers presented here from NUS reflected a lack of real, hands-on experience and were often about simulations. There were, however, some very interesting ideas described, and it is clear that Singaporean scientists have a good sense that the battle for flexible assembly in the 1990s will be in software. (Also, some very interesting papers were developed jointly with Western scientists.) There were excellent contributions from the (still largely unknown) Nanyang Technological University.

As mentioned in the preceding section, this is where the government has decided to set up two major manufacturing the

manufacturing institutes, GINTIC Institute of CIM and the Institute of Manufacturing Technology. Even here, the emphasis is on developing home-grown expertise, learning, and more costeffective solutions rather than completely breaking new ground. I expect this trend to continue. A very good example was given by Ng Kok Loon, former deputy director of SISIR, on free-ranging AGVs. Loon now has left SISIR to set up a company to develop such vehicles. Unfortunately, his paper was not

included in the Proceedings, but Loon described described the the history of trackless AGVs (dead reckoning, laser beacons, inertial guidance, optical or ultrasonic imaging, optical stereoscopic, and others) and then presented a product prototype (available as a real product by the end of 1992), using ultrasonics and transponders, which he claimed had a positioning accuracy of 10 mm. The point here is that this work is not really groundbreaking but that Loon and colleagues had what they felt was a very economical system (cheaper than lasers) that would be of Singaporean origin.

There were a few interesting papers that presented plans or proposals. The Singaporean government has developed an automation master plan, calling for stepped up work in five major areas-CAD/CAM/CAE, robotics,

automated material-handling, computing technologies, and manufacturing production and control systems. The latter is the weakest area in Singapore, and there were several papers devoted to this topic. In large companies, (over $50M/year), production planning is done mainly with material requirements and manufacturing resource-planning systems, but they are not regarded as very successful. Smaller companies (US$10 50M/year) almost all use manual production planning and control. It is estimated that about 600 or more manufacturing companies in Singapore are in the second category, and the institutes, universities, and government labs are seeking solutions for them. This involves capacity planning, material planning, and activity planning. The latter seemed most interesting to me, as the techniques to deal with it are more computer-based, such as Analytical models

(branch and bound,