Next: Computer Networks Laboratory Up: Software Department Previous: Foundation of Computer Science Lab

Distributed Parallel Processing Laboratory


/ N. N. Mirenkov / Professor
/ S. G. Sedukhin / Professor
/ Oleg Monakhov / Visiting Professor
/ Evgueny V. Tetenov / Visiting Professor

Our general program is high-performance computing and advanced information infrastructure. We are also thinking about a planetary machine and active knowledge being developed by humanity. We are undertaking research efforts in parallel and distributed algorithms/architectures, visual (multimedia) languages and tools. The research project ``The new paradigms for clustered computing" is focused on modeling, developing software packages, investigating fault-tolerance and performance analysis for the clustered computing and Web servers.

We are studying the problem of parallel program portability, skeleton-based techniques and developing the supporting tools. Our theoretical work is related to efficient algorithms for highly-parallel numerical computation including linear algebra, Fourier transformation, and image processing. Information about our activities can be found in: http://www.u-aizu.ac.jp/labs/sw-dpp/, which contains links to the experimental laboratory WWW servers. These servers hold the High-Performance Computing and Communication topics, on-line CD-ROM Digital Library and serve as mirror sites of, for instance, the ``Computational Science Educational Project". Web server with clustered architecture was also installed to support students activity in the area of high-performance clusters (http://typhoon.u-aizu.ac.jp).

We are working on the design and development of the system for interactive multi-resolution dynamic terrain rendering using OpenGL technology and advanced graphical 3D accelerators. Based on parallel streaming data exchange, algorithms and data structures for real-time navigation over unlimited terrain data were developed. The system was tested on the available data sets of Fukushima Prefecture.

As one of our results we would like to mention the S4CAD system. S4CAD (System of Systolic Structures Synthesis) is a high level interactive software tool for designing, analysis and modeling of systolic VLSI-oriented array processors. Starting with a localized data dependence graph of the algorithm S4CAD automatically produces all permissible array processors, among which the designer may chose the optimal one. Web page of the S4CAD is available at http://www-dpp.u-aizu.ac.jp/HPCC/S4CAD/.

The important question of our research is parallel program transparency and accessibility of massively parallel computers to large user populations. We are developing a multimedia technology for the interactive specification of application algorithms. This technology is based on self-explanatory components in a film format. A film is a series of color stills supported, if necessary, by text and voice/sound. Each still is to represent a view (some features) of objects or processes. Each film is to represent a multiple view (an extended set of dynamic and/or static features) of objects or processes. Different views can be related to different moments of time, positions in space, levels of hierarchy, values of data attached to space points, etc. Different media can be used for different views. A self-explanatory film means that the associated stills are organized and presented in such a way that the semantic richness is clearly brought out. The investment of meaning in the film is reduced to developing a series of views watching (and hearing) in non-linear order. Usually, a still is self-evident and a film is a result of special gathering of clues or hints. This result is a piece of knowledge. So, self-explanatory adequacy depends on this knowledge. The more accurate and relevant views are used, the greater adequacy is reached.

A computational method film is a series of stills displaying one or more parameterized sets of nodes and/or moving objects in multi-dimensional space-time. Each still highlights a subset of these nodes and/or moving objects. Each film defines a partial order of scanning of the nodes or objects.

As a rule, computation specified on different nodes (objects) of a still is considered to be performed in parallel. Computation specified in different stills is performed sequentially. So, it is possible to say: the shorter film the better.

The user defines the specification by creating his new film. The corresponding program (sequential or parallel) is generated automatically. In fact, we are developing film machines where data, knowledge and algorithms (as well as results) are specified by films (self-explanatory components).

We lead two projects related to filmification of methods and data: Active Knowledge Studio and F-mail System for children, elderly and handicapped people. (http://dpp6.u-aizu.ac.jp/).


Refereed Journal Papers

  1. Mirenkov, N., Vazhenin, A., Yoshioka, R., Ebihara, T., Hirotomi, T., and Mirenkova, T., Self-explanatory software components: a new programming paradigm. International Journal of Software Engineering and Knowledge Engineering, vol. 11, No. 1, pp. 5-36, 2001.

    A new multimedia programming paradigm is presented. It is ba sed on a system of micro- and macro-icons representing self-explanatory software components in a film format. A film is a series of color stills supported, if necessary, by text and sound. Each still is to represent a view of objects or processes. Each film is to represent a multiple view (an extended set of dynamic and/or static features) of objects or processes. A self-explanatory film means that the associated stills are organized and presented in such a way that the semantic richness of a computational scheme is clearly brought out. Icons and films are acquired in a net-accessible database. The user should not study them in advance. The film management system provides simple access to database items and modes to manipulate films. In this paper we explain where the database items are taken from and how the self-explanatory features of items are reached. We also describe how these items can be used for multimedia representation of methods and data and for programming users' algorithmic ideas. In addition, some technical details related to the film management system, rendering engines used for displaying various features of the software components, and the icon language are presented. A special attention is paid to how computational formulas can be attached to a film.

  2. Ebihara, T. and Mirenkov, N., Self-explanatory software components for computation on pyramids. The Journal of Tree Dimentional Images, vol. 14, No. 4, pp. 158-163, 2000.

    A new type of multimedia programming environment based on filmification of methods is considered. The environment is based on libraries of self-explanatory software components, multiple-view films, and a management system for them. In this paper, we consider self-explanatory features by showing examples of multiple-view films related to computation on pyramids. Implementation techniques of a 3-D graphics engine to show 3-D graphics objects and film stills are also presented.

  3. Yoshioka, R. and Mirenkov, N., Rendering techniques for self-explanatory software components. The Journal of Tree Dimentional Images, vol. 15, No. 1, pp. 160-165, 2001.

    Features and implementation techniques of a specialized rend ering engine to display self-explanatory software components are presented. These self-explanatory components employ images and animations to assist the programmer in understanding, modifying and specifying behavior of components. All behavior and functions of a component, including methods to modify and specify its various properties, should be intuitively understandable by just browsing. To reach this extent of self-explanatory level, a film format is used to construct the component's interface. We have implemented a rendering engine specialized in rendering 2-D contents of our components. It is can render static and animated images on to any part of the screen, scaled to any size. A single rendering engine can be used to render a tiled view of several stills on to a panel. In addition, the structure of a special data format used to specify what to render to the engine is described.

  4. Saber, M. and Mirenkov, N., Filmification of methods: cellular programming approach. The Journal of Tree Dimentional Images, vol. 15, No. 1, pp. 110-115, 2001.

    This research is a step into developing a new visual user-oriented programming system to support more understandable user interface and more reusable software components. This system should be used for specifying and programming cellular automation-like algorithms and computation on 2-D grids including image processing. In this paper we present a modified sets of micro-icons to be used for creating and editing software components in a film format. Also, examples of cellular computations prepared in this format are explained. We pay special attention to the structure and functions of template programs supporting the generation of executable codes.

  5. Peng, S. and Sedukhin, S., Design of Multi-dimensional DCT Array Processors for Video Applications. Lecture Notes in Computer Science, vol. 1900, pp. 1086--1094, 2000.

    In this paper, we propose array processors for computing the 2-D and 3-D DCTs. We first introduce a new method, called dimensional splitting method, for the design of array processors for multi-dimensional image transforms. The method can be applied to any multi-dimensional image transforms with separable kernels such as DFT and DCT. Then, we propose a new coding scheme for the 1-D DCT in which the need for generating the kernel matrix in advance is eliminated. Finally, we show array processors for computing r-DCT (r>=2) which are scalable, regular, locally-connected, and fully-pipelined.

Refereed Proceeding Papers

  1. Vazhenin, A., Mirenkov, N., Vazhenin, D., Multimedia representation of matrix computations and data. Proceedings of the Fifth Joint Conference on Information Sciences, Atlantic City, USA, pp. 592-595, 2000.

    We describe techniques for a multimedia representation of matrix computations based on filmification of application methods and data. The multimedia representation is related to special-purpose pictures and animations rendering intermediate/final results of computation and schemes of corresponding computational methods. To support rendering data, a multimedia interface and matrix filtration and matrix scaling techniques are used. To support rendering computational schemes, a film technology is used. Within the framework of this technology, self-explained series of frames, an interface for formula attachment, a program management subsystem as well as tools for data manipulating are discussed.

  2. Mirenkov, N., Yoshioka, R., Ebihara, T., Hirotomi, T., Vazhenin, A., and Mirenkova, T., Multimedia components for methods and data representation. Proc. of SCI 2000, Conference on Information Systems Analysis and Synthesis (ISAS-2000), Orlando, USA, Vo.VII, pp. 342 - 349, 2000.

    A new approach for the preparation of educational materials is considered. It is based on self-explanatory components being created in a film format. The components provide and support well- and intuitively-defined interpretive procedures for multimedia representation of methods and data. A few levels of visual (multimedia) expression compaction are used to keep materials rather small and easily learnable. In this paper we describe our idea of self-explanatory components for the representation of computational methods and data. Three rendering engines used for displaying various features of the components are also considered.

  3. Mirenkov, N., Yoshioka, R., Hirotomi, T., Ebihara, T., Vazhenin, A., and Mirenkova, T., Self-explanatory components for methods and data representation. Proceedings of the International Conference AIIC 2000, Taipei, pp. 14-20, 2000.

    A new multimedia programming paradigm is considered. It is based on a system of micro- and macro-icons (composite pictures) representing self-explanatory software components in a film format. Icons and films are collected in a special database. The user should not study them in advance. In this paper we explain where the database items are taken from and how the self-explanatory features of items are reached.

  4. Sedukhin, S., and Takigahira, T., Performance Evaluation of Clustered Web Server. Proceedings of the International Conference on Parallel and Distributed Techniques and Applications, editor: H.R. Arabnia, pp. 30--40, CSREA Press, June 2000.

    We have made an experiment to measure the performance of clustered Web server. Our Web server is formed by Convoy cluster software and Microsoft Internet Information Server 4.0. The performance is measured by Web Bench 3.0. The purpose of our experiments is to investigate the scalability of our clustered system and performance characteristics for downloading different size files. The result of experiment shows the good scalability of Convoy cluster Web server for downloading at least 50 KB size files. For less than 50 KB file, Convoy cluster Web server does not have good performance scalability because the overhead for load balancing inside clustering system is bigger than the performance improvement by clustering.

Chapters in Books

  1. Sedukhin, S. and Peng, S., Architectural Design of Array Processors for Multi-dimensional Discrete Fourier Transform. Chapter 4, accepted. Highly Parallel Computations: Algorithms and Applications, WIT Press, 2001, Southampton, SO40 7AA, UK, M.P. Bekakos, Editor.

    In this chapter, the design of systolic array processors (SAPs) for computing multi-dimensional discrete Fourier transform ($r$-D DFT) is considered. We introduce two approaches of design: one uses the data dependency graph and a systematic method; the other uses a dimensional splitting method. For the first approach, we investigated three different computational schemes for designing SAPs for 2-D DFT. The systematic method guarantees to find optimal SAPs from a large solution space in terms of the number of processing elements and I/O channels, the processing time, topology, pipeline period, etc. This approach is difficult to extend to the 3-D case since the 6-D data dependency graph for the 3-D DFT cannot be reduced easily. Therefore, for the 3-D DFT we propose another approach: the dimensional splitting method. Using this method, computing of the $r$-D DFT with $r\geq 2$ is done iteratively with each iteration handling the 1-D DFTs of different dimensions. Finally, an application of the proposed SAPs to the prime-factor DFT is presented.

Grants

  1. Nikolay N. Mirenkov., Cooperative research with F-COM Co., Grant-in-Aid, 2000. Research and development of the multimedia interface and the system to support the aged and disabled for the Internet communication.

  2. Nikolay N. Mirenkov., F-Mail sytem. Grant-in-Aid for the Advancement of Scientific Research from Fukushima Foundation, C, 2000.

  3. Stanislav G. Sedukhin., Interactive High-performance Terrain Visualization for Fukushima Prefecture. Grant-in-Aid for the Advancement of Scientific Research from Fukushima Foundation, 2000.

Academic Activities

  1. Nikolay N. Mirenkov., Member of the Program Committee of the PERT-2000, Australia, 2000.

  2. Nikolay N. Mirenkov., Member of the Program Committee of the PaCT2000, Russia, 2000.

  3. Nikolay N. Mirenkov., Member of the Program Committee of the HC-2000, Japan, 2000.

  4. Nikolay N. Mirenkov., Member of the IFIP Working Group 10.3 (Concurrent Systems), 2000.

  5. Nikolay N. Mirenkov., Associate Editor of the Tamkang Journal of Science and Engineering, International Journal, 2000.

  6. Nikolay N. Mirenkov., Guest editor of International Journal of Software Engineering and Knowledge Engineering, 2001, Vol.11, No.1.

  7. Nikolay N. Mirenkov., Member of ACM and IEEE, 2000.

  8. Nikolay N. Mirenkov., Referee of the Journal of Computer Physics Communications, 2000.

  9. Stanislav G. Sedukhin., Member of the Editorial Board of the International Journal Parallel Processing Letters, World Scientific Publ., (1991.9 - ), 2000.

  10. Stanislav G. Sedukhin., Member of the Editorial Board of the International Journal Neural, Parallel & Scientific Computations, Dynamic Publ., (2001.2 - ), 2000.

  11. Stanislav G. Sedukhin., Member of the ACM and IEEE Computer Society, 2000.

  12. Stanislav G. Sedukhin., Program Committee Member of the 1st International Conference on Parallel Processing and Distributed Computing, Applications and Technologies (PDCAT'2000), May 2000, Hong Kong.

  13. Stanislav G. Sedukhin., Program Committee Member of the Australian Computer Systems Architecture Conference (ACSAC'2001), Feb. 2001, Gold Cost, Australia.

  14. Stanislav G. Sedukhin., Program Committee Member of the International Symposium on Parallel Architectures, Algorithms, and Networks (I-SPAN'2000), Dec. 2000, Dallas, USA.

  15. Stanislav G. Sedukhin., Program Committee Member of the 2nd International Conference on Parallel Processing and Distributed Computing, Applications and Technologies (PDCAT'2001), July 2001, Taipei, Taiwan.

  16. Stanislav G. Sedukhin., Program Committee Member of the International Workshop on Optical Networks, International Conference on Parallel Processing (ICPP-01), Sep. 2001, Valencia, Spain.

  17. Stanislav G. Sedukhin., Program Committee Member of the Asia Pacific Web Conference (APWEB'01), Nov. 2001, Changsha, China.

  18. Stanislav G. Sedukhin., Referee of the IEEE Signal Processing Magazine, March 2001.

  19. Stanislav G. Sedukhin., Member of the IASTED Technical Committee.

  20. Nikolay N. Mirenkov., Invited examiner of PhD thesis at University of Victoria, Australia, Apr. 2000.

Others

  1. Fujita, K., Graduation Thesis: Human-Human/Human-Computer Interface in F-Mail System. Univ. of Aizu, 2000, Thesis Advisor: Nikolay N. Mirenkov.

  2. Sakurama, S., Graduation Thesis: Design and Analysis of On-line Database for F-Mail System. Univ. of Aizu, 2000, Thesis Advisor: Nikolay N. Mirenkov.

  3. Matsuda, S., Graduation Thesis: Help system design to make F-mail system be self-explanatory. Univ. of Aizu, 2000, Thesis Advisor: Nikolay N. Mirenkov.

  4. Otani, N., Graduation Thesis: Multiple Views for Self-explanatory Software Components. Univ. of Aizu, 2000, Thesis Advisor: Nikolay N. Mirenkov.

  5. Okayasu, A., Graduation Thesis: Filmification of Methods: Template Programs for 2D Grids. Univ. of Aizu, 2000, Thesis Advisor: Nikolay N. Mirenkov.

  6. Kokubun, K., Graduation Thesis: Multimedia Hieroglyphs: Self-Explanatory Language for Multimedia Communication. Univ. of Aizu, 2000, Thesis Advisor: Nikolay N. Mirenkov.

  7. Ishimatsu, Y., Graduation Thesis: Graphical User Interface for Convoy Cluster. Univ. of Aizu, 2001, Thesis Advisor: Stanislav G. Sedukhin.

  8. Niitsuma, K., Graduation Thesis: Streaming SIMD Implementation of Matrix Inversion by Division-free Algorithm. Univ. of Aizu, 2001, Thesis Advisor: Stanislav G. Sedukhin.

  9. Yuchi, H., Graduation Thesis: Parallel Implementation of 4x4 Matrix Inversion on Pentium III Processor. Univ. of Aizu, 2001, Thesis Advisor: Stanislav G. Sedukhin.



Next: Computer Networks Laboratory Up: Software Department Previous: Foundation of Computer Science Lab


www@u-aizu.ac.jp
July 2001