V.N. Ruchkin - Dr. Sc. (Eng.), Professor, Department of Informatics and Computer Engineering, Ryazan State Uni-versity named for S.Yesenin. E-mail: v.ruchkin@rsu.edu.ru B.V. Kostrov - Dr. Sc. (Eng.), Head of Department of Computer, Ryazan State University named for S.Yesenin. E-mail: kostrov.b.v@evm.rsreu.ru V.A. Romanchuk - Ph. D. (Eng.), Associate Professor, Department of Informatics and Computer Engineering, Ryazan State University named for S.Yesenin. E-mail: v.romanchuk@rsu.edu.ru V.A. Fulin - Head of Centre, Ryazan State University named for S.Yesenin. E-mail: v.fulin@rsu.edu.ru

The paper suggests a generic model for compilation and definition of explicit parallelism in embedded computing systems (ECS), based on clustering. It also describes a rule-oriented production model of an expert system for analyzing implicit parallelism that employs a criteria-based fuzzy-set approach, the number of such criteria equal to the number of ECS modules, and also on the microprogram-s memory capacity, the system-s performance, etc. The structure of the paper is as follows. In first section discusses the architecture of Very Long Instruction Word (VLIW) goes back to the parallel microcode, developed at the very dawn of the computer era. In second section discusses the purposes of the research. Section III focuses on the target model of explicit parallelism in the computing system. The fourth section defines a production model of the expert system based on ECS. Section V employs a criteria-based fuzzy-set approach to analyze implicit parallelism. Here we provide qualitative evaluation of criteria that are important for various data-processing structures. Section VI describes an expert system of smart compilation. The results of the research are summarized in Section VII. The main results of the article are: generic model of explicit parallelism in a computing system with compilation, explicit and implicit parallelism, production model of an expert system for choosing ECS, analysis of implicit parallelism via a criteria-based fuzzy-set approach, algorithm of analyzing implicit parallelism via criteria-based fuzzy-set choice, expert System of intellectual compiling, parameters of qualitative estimates of the criteria-s importance, graphic interface has been developed in the Visual Prolog language. The suggested generic model of design for embedded computing systems enables initially compilation of explicit parallelism, and then, based on fuzzy-set clustering, analysis of implicit parallelism inside these fuzzy structures, across their technical specifications. The resulting product is an intellectual compiler in the form of an expert system for decision-making, where the expert intervenes during the stage of experimental research to select the preferable option of a vector-pipeline implementation of the ECS structure-s parallelization in terms of commands and data

 

1. Baranovskaja T.P., Lojjko V.I., Semenov M.I., Trubilin A.I. Arkhitektura kompjuternykh sistem i setejj: Ucheb. posobie / Pod. red. V.I. Lojjko. M.: Finansy i statistika. 2003. 256 s.
2. Korjachko V.P., Skvorcov S.V., Telkov I.A. Arkhitektury mnogoprocessornykh sistem i parallelnye vychislenija. M.: Vysshaja shkola. 1999.
3. Korjachko V.P., Skvorcov S.V., Taganov A.I., SHibanov A.P. EHvoljucija avtomatizirovannogo proektirovanija ehlektronno-vychislitelnykh sredstv // Radiotekhnika. 2012. № 3. S. 97−103.
4. Korjachko V.P., Skvorcov S.V., Telkov I.A. Model planirovanija parallelnykh processov v superskaljarnykh processorakh // Informacionnye tekhnologii. 1997. № 1. S. 8−12.
5. Zlobin V.K., Ruchkin V.N., Nejjroseti i nejjrokompjutery. SPb.: BKHV-Peterburg. 2011. 256 s.
6. Jordan R., Jozwiak L., Corporaal H. Instruction-set Architecture Exploitation of VLIW ASIPs Using a Genetic Algorithm // Proceding 3nd Mediterranean Conference on Embedded Computing MECO-2014. Montenegro. Budva. P. 32−35.
7. Diken E., Jozwiak L., Jordan R., Corporaal H. Construction and Exploitation of VLIW ASIPs with Multiple Vector-Widths // Proceding 3nd Mediterranean Conference on Embedded Computing MECO-2014. Montenegro. Budva. P. 244−247.
8. Grigorenko D.V., Ruchkin V.N. Povyshenie vosstanavlivaemosti klasternykh nejjroprocessornykh sistem obrabotki dannykh // Cifrovaja obrabotka informacii. № 1. 2013.
9. Viksne P.E., Fomin D.V., CHernikov V.M. Odnokristalnyjj cifrovojj nejjroprocessor s peremennojj razrjadnostju operandov // Izvestija VUZov. Priborostroenie. 1996. № 7. S. 13−21.
10. Romanchuk V.A., Ruchkin V.N. Algoritmy analiza vychislitelnykh struktur na baze nejjroprocessorov // Vestnik RGRTU. 2012. № 2. Vyp. 40.S. 61−67.
11. Ruchkin V., Romanchuk V., Sulitsa R. Clustering, Restorability and Designing Of Embedded Computer Systems Based On Neuroprocessors // 2nd Mediterranean Conference on Embedded Computing MECO-2013. Budva. Montenegro. P. 58−61.
12. Akimenko T.A., Arshakjan A.A., Larkin E.V. Upravlenie informacionnymi processami v robototekhnicheskikh kompleksakh // Tula: Izd-vo TulGU. 2012.
13. Karpov V.S., Ivutin A.N., Suslin A.A. Podkhod k realizacii metodiki ocenki nadezhnosti PO na osnove kompleksnykh metrik // Izvestija Tulskogo gosudarstvennogo universiteta. Tekhnicheskie nauki. 2009. № 4.
14. Ruchkin V.N., Kostrov B.V., Kolesenkov A.A., Ruchkina E.V. Anthropogenic Situation Express Monitoring on the Basis of Fuzzy Neural Networks // Proceedings of the 3nd Mediterranean Conference on Embedded Computing MECO. 2014. Budva. Montenegro. P. 166−169.