[Projects][Publications][Misc][Search SSD]

      Supercomputer's Software and Architecture


Preable:
Scientific seminar of SSD started its functioning the same
week, as SSD was established. Sessions of the seminar are organized in the conference hall of the Institut every Wednesday at 15:00 during a semester. Sessions of the seminar are open for attendance for every body, information on the next session of the seminar is widely disseminated in Novosibirsk.
The seminar plays the great role in scientific life of SSD. This seminar provides the opportunity to all our colleagues for exchange and dissemination of scientific information. The following topics are submitted for the seminar sessions:
SSD stuff members and colleagues from the related Institutions are invited as speakers.
Seminar is the part of the educational programs of the Universities. The students of the State University and the State Technical Universities of Novosibirsk that are working on their diploma, Master and Ph.D. theses in SSD are obliged to attend the sessions of the seminar.

1999 - 2000 academic year.

Date
Oct 6, 1999
Authors
Victor E. Malyshkin
Title
"Problems of parallel realization of big size numerical models
on multicomputers"
Abstract
The steps of numerical models realization on multicomputers and the problems arisen on each step are considered. Fine and coarse grain approachs to parallelzation of application problems are compared. Closeness of continuous and discrete solutions.

Date
Oct 13, 1999
Authors
Anton V. Selikhov
Title
"A Cellular Neural Network approach to modeling of
reaction-diffusion processes: a formal background for a choice
of cell parameters"
Abstract
A Cellular Neural Network is considered as a fine-grain parallel approach to modeling of reaction-diffusion processes. To generate basic type autowave processes, a formal background for a choice of cell parameters is described on the base of investigation results of cell phase plane properties. Some new experimental results are presented. A possibility to realize the CNN approach in a corse-grain parallel system is also discussed. (for more information [RUSSIAN]...).

Date
Oct 20, 1999
Authors
Sergey G. Pudov
Title
"Learning of Homogeneous Cellular Neural Networks"
Abstract
In this talk we present an approach to first order Cellular Neural Networks (CNN) learning. It is based on the ideas of Perceptron learning rule. The proposed method allows to find parameters of a CNN connection template which provides the formation of patterns with preset properties. The method was applied to learning of 1-D and 2-D CNN with symmetric templates and was verified by simulations. We present some experimental results which were compared with the known ones theoretically obtained in [3] for CNN pattern properties (for more information...).

Date
Oct 27, 1999
Authors
Ann S. Nepomniaschaya,
M.A. Dvoskina
Title
"A simple implementation of Dijkstra's shortest path algorithm
on associative parallel processors"
Abstract
Finding shortest paths in networks is a fundamental problem in combinatorial optimization. Efficient sequential algorithms are known for a variety of shortest path problems both for directed and undirected graphs. An important variant of the shortest path problem is the single-source problem. Given a directed n-vertex and m-edge graph G with the real edge weights and with a distinguished vertex s, the single source shortest path (SSSP) problem is to find for each vertex v the length of the shortest path from s to v. When all edge weights are nonnegative, the most efficient solution gives Dijkstra's sequential shortest path algorithm.

In this talk we propose a natural straight-forward implementation of Dijkstra's shortest path algorithm for directed weighted graphs represented as a weight matrix on a model of associative parallel processors of the SIMD type with bit-serial (or vertical) processing (the STAR-machine). We show that the use of such an architecture allows us to obtain a simple and natural parallelization of all parts of Dijkstra's algorithm. We also show how to extend this implementation in a natural and robust way for restoring the shortest path from the source vertex to any given vertex. These algorithms are represented as the corresponding STAR procedures whose correctness is proved. We have also shown that every of these procedures takes O(rn) time, where r is the number of bits required for coding the maximal weight of shortest paths from the source vertex.


Date
Nov 10, 1999
Authors
Olga L.Bandman, Ann S. Nepomnischaya, Sergey G. Pudov
Title
"The impressions about two conferences:
"Automata-99" (Lyons, France),
"Concurrency, specification and programming" (Warsaw, Poland)."
Abstract

Date
Nov 24, 1999
Authors
V.N.Martynov
Title
"Some algorithms of solution of the direct dynamic problems of
seismic and seismology based on
integral transforms."
Abstract

Date
Dec 1, 1999
Authors
Michael B. Ostapkevich
Title
"Event-driven approach to open system design"
Abstract
A comparison between the traditional and event-driven approaches for open system design is drawn. A brief overview of event-driven paradigm evolution in computer science is presented. A sketchy description is given for a library that is intended to support a developer of a system with open architecture. The vague ideas on which features a language for the considered approach ought to have are to be discussed.

Date
Dec 8, 1999
Authors
Yakov I. Fet
Title
"Ya.I.Fet shares impressions about three conferences"
Abstract
"Electronic Imaging and the Visual Arts (EVA'99)" (Moscow, Russia), "PROMETEUS Working Weekend" (Helsinki, Finland), "Information Society Technologies (IST-99)" (Helsinki, Finland).

Date
Dec 15, 1999
Authors
Vladimir D. Korneev
Title
"Portable Parallel Programming with the Message-Passing Interface"
Abstract
Message passing is a programming paradigm used widely on parallel computers, especially Scalable Parallel Computers (SPCs) with distributed memory, and on Networks of Workstations (NOWs). Although there are many variations, the basic concept of processes communicating through messages is well understood. Over the last ten years, substantial progress has been made in casting significant applications into this paradigm. Each vendor has implemented its own variant. More recently, several public-domain systems have demonstrated that a message-passing system can be efficiently and portably implemented. It is thus an appropriate time to define both the syntax and semantics of a standard core of library routines that will be useful to a wide range of users and efficiently implementable on a wide range of computers. This effort has been undertaken over the last three years by the Message Passing Interface (MPI) Forum, a group of more than 80 people from 40 organizations, representing vendors of parallel systems, industrial users, industrial and national research laboratories, and universities.

Date
March 1, 2000
Authors
Alexander P. Vazhenin
Title
"VIM-technology: Main Elements and Operations."
Abstract
The main elements of multimedia programming technology or VIM-technology based on ^”filmification^‘ of methods and data will be presented. Within the framework of this technology, software modules are created in a ''film'' format. This format is to support a new type of abstraction combining both mathematical and physical notions. Special operations on the modules to create new ``films'' are allowed. Films can be considered not only as building blocks, but also as multimedia signs with self-explanatory features, as well as large-grained units of data/knowledge representing some properties of real world objects/processes. So, databases of such films should become sources of ''active'' knowledge. The goal of this talk is in providing an overview of VIM-approach, presenting examples of the film frames and describing modes to manipulate films. A special attention is paid to how computational formulas can be attached to a film (a multimedia algorithmic skeleton).

Keywords: multimedia, human-computer interfaces, computer software, information systems, communication via films.


Date
March 22, 2000
Authors
Sergey A. Simonov
Title
"Attempts to commercialize intellectual products."
Abstract
The idea of commercializing intellectual/scientific results is discussed. The more perspective ways now in current condition are: propose existing small programming widgets, and implementing the problem of foreign companies offers. As an example describe the newest approach to using informational resources with the help of wireless access to Internet with the help of cellular phones. The approach based on new exchange protocols (WAP - wireless application protocol) and new makeup language (WML wireless makeup language).

Date
Apr 5, 2000
Authors
Vitaly A. Morozov
Title
"Implementation of a software package for studying problems of
linear instability"
Abstract
In developing a software package of automatic studying the problems of linear instability, oriented to using in physical modelling of vibrational, termocapillary or mixed convection, the important issue is an implementation of accurate, universal and fast kernel. The basic, repeated and the longest in duration operation of the package is solving of special structure system of linear algebraic equations. The matters of implementing the solver with mentioned requirements are discussed. The high reliability and accuracy is achieved by applying changeable in length operands. The natural slowdown of kernel processing is compensated by effective parallel implementation of the algorithm in MPI-media.

Date
Apr 12, 2000
Authors
Yakov I. Fet
Title
"Architects and Architectures"
Abstract
This report does not intend to be a systematic exposition of the history of computer architecture. Instead, the main purpose of it is to show, by means of several examples, the relationship between the architect's personality and the greatness of his creations, their significance to the society. Brief essays are given on the lives and discoveries of Charles Babbadge, Alan Turing, John von Neumann, Seymour Cray, Daniel Hillis, Alexey Lebedev, Leonid Kantorovich. The similarity is emphasized between the creative work of computer architect and the designer of buildings. An opinion is claimed that the work of this kind is related to the junction of science, technology, and art. Some problems are touched of the physiology of creative work as well as of the possibility to influence the development of corresponding inclinations in course of school and university education.

Date
Apr 19, 2000
Authors
Nikolay V. Kuchin
Title
"High Performance Cluster Computing"
Abstract
The recent advances in high speed networks and improved microprocessor performance are making clusters or networks of workstations an appealing vehicle for cost effective parallel computing. Clusters built using commodity hardware and software components are playing a major role in redefining the concept of supercomputing. The trend in parallel computing is to move away from specialized platforms, such as the Cray/SGI T3E to cheaper, general purpose systems consisting of loosely coupled components built up from single or multi-processor workstations or PCs. This approach has a number of advantages including that of being able to build a platform for a given budget, which is suitable for a large class of applications and workloads.

IEEE Task Force on Cluster Computing
http://www.dgs.monash.edu.au/~rajkumar/tfcc/


Date
Apr 26, 2000
Authors
V.I. Elokhin (his collaborators are E.I. Latkin and V.V.Gorodetskii)
Title
"DIFFERENT MECHANISMS OF OSCILLATORY AND AUTOWAVE BEHAVIOUR
DURING CARBON MONOXIDE OXIDATION OVER Pt AND Pd:
CELLULAR-AUTOMATA MODELS"
Abstract
The aim of thie study is to compare the specific features of the statistical lattice models for imitating the oscillatory and autowave dynamics in the adsorbed layer during CO oxidation  over Pt and Pd  monocrystals. The statistic cellular automata model is used  to simulate the process.The states of the cells are determined according to the rules prescribed by the detailed reaction mechanisms:
The synchronous oscillations of the reaction rate, coverages and surface phases are exhibited within the range of the suggested model parameters (under the conditions close to the experimental observations). These oscillations are accompanied by the autowave behaviour of surface phases and adsorbate coverages.

The page last updated on Jul 11, 2000