CC BY
26
THE HARD-end-SOFT AUTOMATIC DESIGN OF SELF-ORGANIZING ADAPTIVE SYSTEMS
POLYAKOV GENNADY
Tel.: (0572) 47-49-26. E-mail: tda_ua@pochtamt.ru
1. The Self-organizing adaptive systems as new class of computing systems and new stage of digital techniques development
The analysis of a number of projects at the end of XX and at the beginning of XXI centuries allows one to consider that the new stage of development in computer field has begun. There are the following examples: multiprocessing systems with the programmed architecture [1]; computing systems with the programmed structure architecture [2]; project “Adaptive selforganizing computers, ASC” [3.. .7]; program ESPRIT project P1085, evolving configurable architectures based on transputers [8]; IBM, Hewlett Packard, Intel elaboration of other flexible dynamically configurable architectures [9], Microsoft company project of self-programming computer creation by 2010 year; the production of FPGA, CPLD (by nXilinx, Altera, Actel, Cypress, Atmel etc.) [10.13].
According to authors point of view, the new stage’s contents is the elaboration and application of self-organizing adaptive systems (SAS) that are capable, by system means, for automatically goal-seeking change (for efficiency increase) of parallel hard/soft ware architecture, characteristics and functioning under range of application change or according to raised demands and tasks.
2. The Automatic design technolo gy of the multiparallel hard&soft of the self-organizing adaptive systems
SAS are going to have the following features:
Multiparallelism - it means application of all or just couple of known data parallel processing methods with automatic methods optimization according to specific application fields, raised tasks and users demands.
Automatic hard/soft compromise - task of hard/soft ware optimum relationship choice.
Self-organizing element base - based on dynamically programmed, configurable and self-organizing VLIC.
Parallel hard/soft ware self-organization - hard/soft ware new architectures automated layout.
Adaptability - control of high quality under the change of application field, raised demands and tasks and also under take-over and failure.
Automatic evaluation and visualization of efficiency index and their degree of conformity to the demands and limitations.
It’s obvious that present Digital Units Automated Design (EDA) and Parallel Program Design technologies can’t meet the demands for elaboration of system with such features.
The purpose of SAS technology is the support of Selforganizing Adaptive System creation in the next 5-10 years.
The basic components of SAS technology are:
- Database containing tasks C-programs, demands and restrictions, multiprocessor architecture characteristics, hardware environment libraries, parallel programming libraries.
- C-program to C-graph translation subsystem.
- Sub sy stem of e stimatio n of parallel data pro cessing optimal methods set used for appropriate tasks under proper demands and restrictions.
- “Separation” subsystem - separates hardware and software according to tasks, demands and restrictions.
- Multiparallel component base design subsystem - controls the determination of mono- and multifunctional modules set for SAS hardware, and modules parallel scheme with the givenfeatures synthesis, configuration ofprogrammed VLIC.
- Parallel SAS hardware design subsystem for the task solution with the hardware generated methods (Intelligent Designer). This subsystem provides the automatic layout of tasks parallel models for synthesis of SAS hardware, SAS operation subsystem, SAS control unit, SAS operation subsystem and control unit bonds “closure”.
- Parallel SAS software design subsystem - task solution with the software generated methods (Intelligent Compiler). This subsystem provides the tasks parallel models design for SAS software synthesis, thread programs of the join parallel program formation, input of the utilities of input/output threads and timing synchronization of thread programs, thread programs and data distribution on SAS processors.
- Test design subsystem and also SAS soft/hard ware design results verification.
- Operational visualization subsystem of self-organization main stages results and estimation of SAS efficiency.
The SAS generalized architecture is showed and the basic computing steps are described.
In the given paper there is the information about the elaborated prototypes of SAS technology main compounds and demonstration examples of their application.
Expectation useful applied effects:
- considerable diminution of: design time (problem Time-to-Market), time of parallel programming, laboriousness, design cost;
- essential increase of: complexity of design tasks (problem System-on-Chip), complexity of programming tasks, parallel programs quality;
- essential expansion of application fields.
References: 1. Kalyaew A. W. The Multiprocessing Systems with programmed architecture. M.: Radio end Communication, 1984. P .240. 2. Kornyeew W. W. The Architecture of the computing systems with programmed structure. Novosibirsc: Science, 1985. P.168.3. Polyakov G.A. The construction principles and automatic design of the selforganizing systems with parallel data processing end raising the efficiency of the radio engineering complexes / Edit by prof. W.K.Strelnicov. Kharkov: WIRTA PWO,1986. P.571.4. Polyakov G.A., Umrihin Yu.D. The Design automation of the communication and control digital systems. M.: Radio and Communication, 1988. P. 304. 5. Polyakov G.A. The Adaptive self-organizing computing complexses(ASCC). M.: Scientific-Technical Committee on informatization end computer science USSR.Section N°1"Computing machines end complexses”, 1989. 6. Polyakov G.A. The Adaptive self-organizing computer complexes.Methods of representation and processing ofcasual signals and fields: 11 All-Union Scientific-Technical Conference.: Kharkov, 1991. 7. Polyakov G.A.The Adaptive selforganizing systems with multiparallel data processing-the computer science development strategy in XX1 century. Kharkov. Applied radioelectronics, 2002. Vol.1, N°1. P.57. 8. Transputers Databook.INMOS Ltd.1985. Architecture and software: Trans. engl. / Edit. by G.Harp. M.: Radio and Communication,1993. P.303. 9. Kornyeew W.W., KiselewA.W. The modern microcomputers. M.: NOLIG, 2000. P. 312. 10. Ugrumov E.P. The digital techics.-SPb.:BHW-Sanct-Petersburg, 2000. P.528. 11. Altera:The DSP ltader for programmable logic/News and Views. Altera. 2000. Feb. P.31-32. 12. Amtel Corporation Programmable Logic and System-Level ICs Data Book. Amtel Corporation. 2000. 13. Mercury programmable logic device family. Altera data sheet. 2001. P.83.
R&I, 2003, Ns 3
149
