Lapas attēli
PDF
ePub

I am very optimistic about the future of this center. The planners have emphasized that they are NOT interested in building Taiwan's own supercomputer or engaging in advanced computer science software research, but instead want to concentrate on the research associated with applications. NCHC staff are already "on the road," providing training via courses and seminars on techniques related to high-performance computing. They are funding university faculty to port and tune their existing programs to make better use of whatever supercomputer is going to be installed. (Typically, existing programs will not take maximum advantage of vector or parallel architectures.) I am optimistic also because in recruiting staff for the new center, double talents are being emphasized, application fields first and then good knowledge about computer architecture, algorithms, etc. The hope is to get advanced computing to the end user quickly. User groups based on application area are being formed. Finally, the vendor that is finally selected will be required to provide not only systems support but application research support, researcher exchange, seminars, and so forth.

David K. Kahaner joined the staff of the Office of Naval Research Asian Office as a specialist in scientific computing in November 1989. He obtained his Ph.D. in applied mathematics from Stevens Institute of Technology in 1968. From 1978 until 1989 Dr. Kahaner was a group leader in the Center for Computing and Applied Mathematics at the National Institute of Standards and Technology, formerly the National Bureau of Standards. He was responsible for scientific software development on both large and small computers. From 1968 until 1979 he was in the Computing Division at Los Alamos National Laboratory. Dr. Kahaner is the author of two books and more than 50 research papers. He also edits a column on scientific applications of computers for the Society of Industrial and Applied Mathematics. His major research interests are in the development of algorithms and associated software. His programs for solution of differential equations, evaluation of integrals, random numbers, and others are used worldwide in many scientific computing laboratories. Dr. Kahaner's electronic mail address is: kahaner@xroads.cc.utokyo.ac.jp.

JAPAN ATOMIC ENERGY RESEARCH INSTITUTE (JAERI)

INTRODUCTION AND SUMMARY

Computer-related research at the Japan Atomic Energy
Research Institute (JAERI) is summarized.

The Japan Atomic Energy Research Institute (JAERI) was established in 1956 to perform the research and development (R&D) associated with implementing the country's nuclear program. It is funded under the Science and Technology Agency (STA), which is an arm of the Prime Minister's Office. Currently, the major projects are as follows.

• Nuclear energy production system including high temperature gascooled reactor and fusion reactor

• Nuclear safety

• Radiation applications

• Nuclear ship

JAERI has an administrative headquarters and a radioisotope school in Tokyo. The main research center and a fusion research center are in Tokai and Naka, small towns on Japan's Pacific coast about 70 miles north of Tokyo. Radiation chemistry research occurs both near Osaka and at Takasaki, halfway between Tokai and the Japan Sea side of the main (Honshu) island. Nuclear ship research occurs at the northern tip of Honshu. (Japan's nuclear waste disposal facility is also near the northern end of Honshu, but this is not run by JAERI.)

by David K. Kahaner

JAERI's budget and staffing grew rapidly through the mid-1980s but have essentially been constant since 1985, nearly $880M (about 90% from STA) and about 2,500 staff (one-third each research, technician, and administrative). JAERI is currently building a high temperature test reactor, operating fusion experimental equipment, building a synchrotron orbital radiation (SOR) facility, etc.; plant and equipment are expensive and account for the very large budget.

Japan generates approximately 30% of its electrical power by nuclear means, although this figure would reduce to 17% if other electrical generating plants were used to capacity. Sentiment in the country is modestly antinuclear, but the major reason that JAERI budgets have not increased is that nuclear technology is now mature. The nuclear ship, MUTSU, has been behind schedule and immersed in controversy for years; it is now scheduled to be decommissioned in the spring of 1992.

I spent 1 day at Tokai visiting with scientists in the Computing and Information Systems Center (CISC), which is essentially the computer support organization. They also perform some of their own research. CISC is run by

Dr. Masayuki Akimoto
General Manager
Computing and Information
Systems Center

Japan Atomic Energy Research
Institute

Tokai-mura, Naka-gun,
Ibaraki-ken 319-11, Japan

Tel: +81-292-82-5611
Fax: +81-292-82-6070
E-mail: j2304@jpnjaeri.bitnet

I also had an opportunity to meet with three members of his research staff,

Dr. Mitsuo Yokokawa
Dr. Hideo Kaburaki
Mr. Hiroo Harada

(same address as above)
Tel: 0292-82-5976
Fax: 0292-82-6070

E-mail: yokokawa@catalyst.tokai.

jaeri.go.jp

and also with

Dr. Masashi lizumi
Deputy Director General
Tokai Research Establishment
Japan Atomic Energy Research
Institute

Tokai-mura, Naka-gun,

Ibaraki-ken 319-11, Japan
Tel: +81-292-82-5014
Fax: +81-292-82-6111

Interestingly, Dr. Iizumi and I had some common background, as he has visited the reactor facility at the National Institute of Standards and Technology (NIST) in Maryland several times.

During the time of my visit, Prof. Eugene Wachspress (University of Tennessee) was also visiting JAERI in order to give a series of lectures on

various topics in the area of numerical linear algebra related to reactor problems. As is well known, the nuclear industry provided some of the earliest problems in numerical analysis and was a significant motivator for the development of large-scale supercomputers. There are still major problems associated with all aspects of nuclear energy that require extensive computer analysis. While new civilian power plants are not being built, there are many applications of nuclear energy. Also there are many research reactors and medical applications. Finally, there are the knotty questions about disposal, leakage, etc.

In his lectures, Wachspress focused on sparse matrix methods for linear equations and also for eigenvalue problems. These techniques are often at the heart of nuclear application codes. Another purpose of his visit was to assist JAERI in thinking about how best to use new computers. He remarked to me that there is a great deal of time and money spent getting old programs operational on new machines. This includes fitting old algorithms into parallel and vector architectures when new algorithms especially designed for these computers would be much more efficient. This is a common problem at any laboratory that has a significant development effort in place using large, existing program packages.

CISC has two Fujitsu VP2600 supercomputers, used mostly for running large nuclear codes. One is at Tokai, the other at nearby Naka where the fusion research is located. There are also three other Fujitsu M780 series mainframes and many workstations. In Tokai the computers are served by a 157.5-GB disk, 88-GB optical disk, tapes, etc., and there is similar equipment at Naka. There is a 6-km on-site local area network (LAN) (205 Mbps) and a 9.2-km optical fiber connecting Tokai and the fusion laboratory. The Tokai laboratory has been connected to Bitnet

(56 Kbps) for several years and is currently connecting itself to Internet. There are also two DECnet links, one serving the fusion research community (I commented in an earlier report that worldwide this group is heavily DEC and VAX oriented), another that supports TCP/IP. (Interestingly, JAERI's earliest computer was an IBM 650, the same model that I began computing on at the Watson Laboratory, then at Columbia University.)

CISC has a total staff of about 55, 10 involved in R&D (including 2 senior scientists) and 20 in systems, networks, etc. The remaining 25 are contractors, etc. The remaining 25 are contractors, mostly for operations. Some interesting statistics for last fiscal year follow.

[blocks in formation]

intelligent nuclear power plants; and peripherally to create and transfer artificial intelligence (AI) techniques in the nuclear field. Akimoto pointed out that a paper on HASP has been submitted to "Expert Systems and Computer Simulation in Energy Engineering," 17-21 March 1992 at Erlangen, Germany.

Computer integration of nuclear design programs. Building a nuclear reactor is a complex process involving engineering from diverse areas. Currently, JAERI scientists are constantly running large programs, taking the resulting output of one and using parts of it for yet another calculation. One CISC project is to integrate the programs (which are frequently produced by different commercial vendors) and use expert systems to ease the transition between different programs.

Vectorizing. Much of the work at CISC centers around getting the best performance from existing computers on standard workhorse programs (hence Wachspress' remark above). Since the mid-1970s more than 50 nuclear programs have been vectorized. To assist them in assessing what might be expected from the Fujitsu VP machines, CISC has assembled a suite of benchmark programs for testing vectorization. The list below includes programs of direct interest to JAERI and also some standard benchmark problems that readers will recognize.

SLWALF

AEOLUS

STREAM

Robots at work inside nuclear power plants. (Parts of this environment must be among the most hostile imaginable.) Working in Lisp, CISC scientists have been simulating how human-acts could be realized by a human-shaped intelli- VDIRECT gent robot (HASP: Human Acts Simulation Program). This is part of a 10-year program begun in 1987 to develop basic and underlying design technologies for intelligent robots; to develop the basic technologies for very advanced,

SRAC

CITATION

Simulation Code for Alpha-
Particle Heating
Three-Dimensional (3D) Mag-
netohydrodynamic Code
Thermal Hydraulics Analysis
Code

High Temperature Test Reac-
tor Code

Thermal Reactor Standard
Neutronics Code System

3D Neutron Diffusion Code TVVOTRAN Two-Dimensional (2D) Neu

tron Transport Code

[blocks in formation]
[blocks in formation]

>90

2582

14-15

15-16

>16

23

Thus most programs result in only modest speedups. This is fairly typical And the following vectorization speed- for programs that were originally writups on the VP2600. ten for older scalar machines.

Two CISC scientists I met (Yokokawa and Kaburaki) have experience with parallel machines as well as traditional supercomputers. Dr. Yokokawa is a Ph.D. graduate of Tsukuba University and worked under Professor Y. Oyanagi (now at the University of Tokyo) and Professor M. Mori (University of Tokyo), two well-known Japanese numerical analysts. Dr. Kaburaki was coauthor of a paper on the use of Fujitsu's AP1000 parallel processor for Monte Carlo simulation of gas dynamics. This paper was presented at the recent joint FujitsuAustralian National University workshop. Yokokawa and Kaburaki gave me another paper coauthored by JAERI and Fujitsu scientists on parallelizing the Monte Carlo program MCACE for the AP1000. This had also been presented at the aforementioned workshop. At the moment, Kaburaki is also working with N. Ito, another recent graduate from the University of Tokyo's

No. of Programs

[merged small][merged small][merged small][merged small][merged small][merged small][merged small][merged small][merged small][ocr errors][merged small][merged small][merged small][merged small][merged small]

physics department on the development of a special purpose parallel computer (m-TIS) for doing Ising spin computations. Ito's primary thesis work was on using a supercomputer for similar calculations. He was not at JAERI during my visit, so I will report on this work later.

I also had an opportunity to learn about JAERI's scientific subroutine library, JSSL, from Mr. Harada. This was of special interest to me as I have done research in development of similar packages. I was mildly disappointed as this library is not really a current product; its last revision was in 1982. There is only a very small staff to manage and work on the library, and it is a big job to do so. JSSL does contain some excellent programs, such as MA21A originally from the Harwell library, but also some very old routines, including Crout reduction of matrices and Romberg quadrature. It did not appear to have any of the modern, standard programs such as those from Linpack, Eispack, Quadpack, etc. In fact, some JAERI scientists were surprised to hear that in the West, large quantities of high quality mathematical software are in the public domain and can be obtained easily and rapidly by electronic mail. Of course, some scientists are using this approach for their own personal research, but it is far from common, and in any case the routines have not made their way into JSSL. These observations were consistent with comments made to me by Wachspress, who noticed that among the Japanese scientists he met, awareness of developments in the West was uneven. Essentially all open research from any country is known here--and at a high level of absorption--but its distribution appears to be less uniform than in the United States or European Community. Electronic communication will go a long way, in my opinion, to assist scientists in accessing modern information from anywhere in the world. This is a very important trend.

« iepriekšējāTurpināt »