Generation Of Supercomputers Research Articles

Associative memory based on the principles of optical data processing for the new generation of supercomputers

The factors involved in implementing optical associative memory for the new generation of supercomputers are considered. The problem of constructing rapid-access high-capacity associative memory is solved by combining several associative memory modules in a single system.

Simulation of two‐ and three‐dimensional internal subsonic flows using a finite element method

AbstractIn this paper a finite element method is presented to predict internal subsonic flows. Using a low‐Mach‐number approximation, the pressure is decomposed into a mean thermodynamic contribution and a dynamic fluctuation to deal with the complex role of the pressure in internal aerodynamics. A semi‐implicit time integration and a finite element method with a moving mesh are described to take into account complex geometries and moving boundaries. An Uzawa algorithm accelerated by a preconditioned residual method is introduced to solve the coupled non‐symmetric linear system for the velocity components and the pressure. An efficient conjugate gradient method combined with an incomplete LU preconditioning is used to solve the non‐symmetric linear systems arising from the discretization. The implementation of the numerical scheme on parallel supercomputers is also discussed. Efficient algorithms for the finite element assembly phase and for the solution of linear systems are described which take advantage of the parallel architecture of the new generation of supercomputers. With this technique a global speed‐up of 10 is achieved on a supercomputer with eight processors. To illustrate the capabilities of the numerical method, 2D and 3D simulations of flows in the combustion chamber of a reciprocating engine and around the combustor dome of a gas turbine engine are presented.

Practical numerical considerations for the Alekseev–Mikhailenko method

Programs that utilize the Alekseev–Mikhailenko method are becoming viable seismic interpretation aids because of the availability of a new generation of supercomputers. This method is highly numerically accurate, employing a combination of finite integral transforms and finite difference methods, for the solution of hyperbolic partial differential equations, to yield the total seismic wave field.In this paper two questions of a numerical nature are addressed. For coupled P–Sv wave propagation with radial symmetry, Hankel transforms of order 0 and 1 are required to cast the problem in a form suitable for solution by finite difference methods. The inverse series summations would normally require that the two sets of roots of the transcendental equations be employed, corresponding to the zeroes of the Bessel functions of order 0 and 1. This matter is clarified, and it is shown that both inverse series summations may be performed by considering only one set of roots.The second topic involves providing practical means of determining the lower and upper bounds of a truncated series that suitably approximates the infinite inverse series summation of the finite Hankel transform. It is shown that the number of terms in the truncated series generally decreases with increasing duration of the source pulse and that the truncated series may be further reduced if near-vertical-incidence seismic traces are avoided.

Galaxy Interactions and the Stimulation of Nuclear Activity

The idea that interactions between galaxies can lead to enhanced galactic activity has a long and noble lineage. Since the early days of Baade and Minkowski (1954), through the seminal paper of Toomre and Toomre (1977) with its “stoking the furnace” imagery, to the similarly colorful and influential “feeding the monster” paper of Gunn (1979), and the landmark work by Larson and Tinsley (1978), this idea was elaborated upon and generalized. The pace of development of this hypothesis has accelerated greatly on both the observational and theoretical fronts during the 1980’s, due in part to a growing perception that the AGN phenomenon could only be understood when the role of the environment in triggering or nurturing the activity was understood (cf. Balick and Heckman 1982). In recent years, the impact of the extragalactic IRAS database (on the observational side) and of the new generation of supercomputers and innovative software (on the numerical side) have led to an almost explosive growth in the number of papers written about this subject.

Scientific computing challenges arising from space-borne observations

Five supercomputing problems are described: (1) determining the motions of the Earth's inner core boundary from outer-space measurements of the magnetic field; (2) simulating the global dynamics of the Earth's ocean currents, sea-ice cycles, land surface biology, and interannual atmospheric climates; (3) upper atmospheric chemistry; (4) modeling solar flare eruptions; and (5) the role of jet processes in the birth of stars. In addition, two examples indicating the growing importance of the use of supercomputers for scientific visualization are presented. The first deals with remote mapping from space of the Earth's topography or other planets. The second problem from astrophysics deals with mapping the cosmic background radiation in the presence of other astronomical signatures. A brief overview of interdisciplinary modeling plans under development for the next generation of supercomputers is given. >

Mapping of neural networks on the honeycomb architecture

The use of the Honeycomb architecture for the implementation of neural networks in the next generation of supercomputers is discussed. The major point is that the relative value of various neural network models depends heavily on the chosen implementation technology. Models that are superior when it comes to conventional implementation architectures and related technologies may be inferior if the Honeycomb architecture and related technologies are used, and vice versa.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Performance analysis of direct N-body calculations

A theoretical framework for analyzing the computational cost of gravitational N-body codes is introduced and applied to three different types of direct-summation codes, including the type of Aarseth code which has found most general use. The method of analysis, based on the probability distribution of nearest-neighbor distances, is described. The number of time steps required for a variety of different versions of the Aarseth scheme and a variety of physical models of spherical star clusters is estimated in order to measure the effects of different degrees of central concentration. Analytical estimates of computer time required are compared with actual measurements, and the validity of the scaling outside the range actually tested is discussed. A practical result for planning star cluster simulations on the next generation of supercomputers is derived. It is found that the consumption of computer time can be very centrally concentrated. 15 references.

Strategies for obtaining the maximum performance from current supercomputers

The exploitation of special features of the current generation of supercomputers is discussed with particular reference to computational quantum chemistry. Modification of algorithms in the light of the very large memory available on the CRAY 2 is described, and multi-processing of programs is considered for both the CRAY XMP and CRAY 2. The overhead in multiprocessing can be reduced to a small percentage even for steps that perform extensive IO. The division of work among CPU's at different program levels is considered for various examples, and optimal levels for the division are discussed.

The Japanese national project for new generation supercomputing systems

This paper review the current status of the R&D of the Japanese national project for new generation supercomputing systems. The project identifies the basic technology for ultra-high speed supercomputers based on parallel processing with new devices. The target system is expected to operate at more than 10 Gflops. In this paper, the R&D on computer architecture is emphasized from the viewpoint of parallel processing.

Comparison of the CRAY X-MP-4, Fujitsu VP-200, and Hitachi S-810/20

A set of programs, gathered from major Argonne computer users, was run on the current generation of supercomputers: the CRAY X-MP-4 is a consistently strong performer over a wide range of problems. The Fujitsu and Hitachi computers excel on highly vectorized programs and offer an attractive opportunity to sites with IBM-compatible computers.

Optical Interconnections Between Integrated Circuit Chips

A major limitation to achieving significant speed increases in VLSI lies in the metallic interconnects. They are costly not only from the charge transport standpoint but also from capacitive loading effects. The Defense Advanced Research Projects Agency, in pursuit of fifth generation supercomputing, is investigating alternatives to the VLSI metallic interconnects, especially the use of optical techniques to transport the information between chips or between boards. As the on‐chip performance of VLSI continues to improve via the scale‐down of the logic elements, the problems associated with transferring data off and onto the chip become more severe. The use of optical carriers to transfer the information within the computer is very appealing from several viewpoints. Besides the potential for gigabit propagation rates, the conversion from electronics to optics conveniently provides a decoupling of the various circuits from one another. Significant gains will also be realised in reducing cross‐talk between the metallic routings, and the interconnects need no longer be constrained to the plane of the VLSI chip. In addition, optics can offer an increased programming flexibility for restructuring the interconnect network.

Computer models and automata theory in biology and medicine

The applications of computers to biological and biomedical problem solving go back to the very beginnings of computer science, automata theory [1], and mathematical biology [2]. With the advent of more versatile and powerful computers, biological and biomedical applications of computers have proliferated so rapidly that it would be virtually impossible to compile a comprehensive review of all developments in this field. Limitations of computer simulations in biology have also come under close scrutiny, and claims have been made that biological systems have limited information processing power [3]. Such general conjectures do not, however, deter biologists and biomedical researchers from developing new computer applications in biology and medicine. Microprocessors are being widely employed in biological laboratories both for automatic data acquisition/processing and modeling; one particular area, which is of great biomedical interest, involves fast digital image processing and is already established for routine clinical examinations in radiological and nuclear medicine centers, Powerful techniques for biological research are routinely employing dedicated, on-line microprocessors or array processors; among such techniques are: Fourier-transform nuclear magnetic resonance (NMR), NMR imaging (or tomography), x-ray tomography, x-ray diffraction, high performance liquid chromatography, differential scanning calorimetry and mass spectrometry. Networking of laboratory microprocessors linked to a central, large memory computer is the next logical step in laboratory automation. Previously unapproachable problems, such as molecular dynamics of solutions, many-body interaction calculations and statistical mechanics of biological processes are all likely to benefit from the increasing access to the new generation of supercomputers. In view of the large number, diversity and complexity of computer applications in biology and medicine, we could not review in any degree of detail all computer applications in these fields; instead, we shall be selective and focus our discussion on suggestive computer models of biological systems and those fundamental aspects of computer applications that are likely to continue to make an impact on biological and biomedical research. Thus, we shall consider unifying trends in mathematics, mathematical logics and computer science that are relevant to computer modeling of biological and biomedical systems. The latter are pitched at a more formal, abstract level than the applications and, therefore, encompass a number of concepts drawn from the abstract theory of sets and relations, network theory, automata theory, Boolean and n-valued logics, abstract algebra, topology and category theory. The present analysis of relational theories in biology and computer simulation has also inspired a number of new results which are presented here as Conjectures since their proofs are too lengthy and too technical to be included in this review. In order to maintain a self-contained presentation-the definitions of the main concepts are given, with the exception of a minimum of simple mathematical concepts. The purpose of these theoretical sections is to provide the basis for approaching a number of basic biological questions: (1) What are the essential characteristics of a biological organism as opposed to an automaton? (2) Are biological systems recursively computable? (3) What is the structure of the simplest (primordial) organism? (4) What are the basic structures of neural and genetic networks? (5) What are the common properties of classes of biological organisms? (6) Which system representations are adequate for biodynamics? (7) What is the optimal strategy for modifying an organism through genetic engineering? (8) What is the optimal simulation of a biological system with a digital or analog computer? (9) What is life?

Applications of microcomputers in engineering geology: Introductory Note

1985 marks the tenth anniversary of the release of the first microcomputer in kit form in the USA. It is hard to believe that just over 40 years ago, the first computers were invented to crack secret message codes during the 2nd World War. These huge computers consisted of thousands of valves which essentially provided switches for on/off binary logic. The development of the transistor in 1948 led to the valve being replaced and started the miniaturization of computers. This miniaturization trend has continued since the 1950s, leading to the development of minicomputers in the 1970s and resulting today in a computer technology which allows over 100000 components to be etched onto a single chip of silicon less than l cm 2 . This is the microprocessor or computer on a chip and contains the equivalent of the many thousands of valves used in the earlier computers which occupied entire rooms. In a similar manner, the storage of digital information has undergone a revolution and it is common today to be able to store up to 1 Mbyte of information on one floppy disk. The advent of laser disks will, in the future, probably increase storage capacity by a factor of one million and the power of the new generation of super computers, such as the Cray, has led to problems being solved in microseconds rather than hours. These trends have resulted in faster precise processing of data and ready access to more reliable, smaller and cheaper computers with a greater capacity

Parallel Processor Programs in the Federal Government

In 1982, a report dealing with the nation's research needs in high-speed computing called for increased access to supercomputing resources for the research community, research in computational mathematics, and increased research in the technology base needed for the next generation of supercomputers. Since that time a number of programs addressing future generations of computers, particularly parallel processors, have been started by U.S. government agencies. The present paper provides a description of the largest government programs in parallel processing. Established in fiscal year 1985 by the Institute for Defense Analyses for the National Security Agency, the Supercomputing Research Center will pursue research to advance the state of the art in supercomputing. Attention is also given to the DOE applied mathematical sciences research program, the NYU Ultracomputer project, the DARPA multiprocessor system architectures program, NSF research on multiprocessor systems, ONR activities in parallel computing, and NASA parallel processor projects.

Users' experience with the ILLIAC IV system and its programming languages

The ILLIAC IV is a unique machine which has led the research and development of lockstep parallel processing. The machine has been operational since 1973, in experimental mode, and since 1975 in full production mode. There has been on the order of a hundred users of the machine and these users and their codes have been well documented. Four languages are available on the machine ranging from high level to machine code. A survey has been conducted of the users in order to determine how the ILLIAC IV has been employed and how the high level programming languages have facilitated the use of this machine. This paper presents the results of that survey.The survey attempts to confirm or eliminate some of the folklore that has grown up around the ILLIAC IV facility. It can be helpful in the design of the next generation of supercomputers and their languages and in the improvement of the present generation of languages. The responses to the survey indicate: 1) that the ILLIAC IV has been accepted by the scientific community; 2) that a wide range of different application areas have used the machine; 3) that users have had to construct their programs so as to minimise the effects of the serious bottleneck created by the movement of data between the main and backing stores; and 4) that the high level programming languages available have insufficient or inefficient structures which at times require the use of machine code.