Article

Article title IMPROVING THE STRUCTURE OF COMPUTATION CHANNELS RNS-BASED DEVICE FOR ENHANCING THE RELIABILITY OF THE CALCULATION
Authors E.S. Balaka, A.N. Schelokov
Section SECTION VI. COMPUTING SYSTEMS OF NEW GENERATION AND NEUROCOMPUTERS
Month, Year 06, 2015 @en
Index UDC 004.272.2
DOI
Abstract Opportunity to improve the reliability of digital devices through the use of codes that can detect and correct calculations involved in making mistakes is an alternative to the majority of redundancy having their characteristic high redundancy. Best basis for constructing such codes can serve as modular arithmetic. Corrective opportunities modular codes are known. However, in practice, they have a wide application not received. Primarily, this is due to the hardware costs of implementing the encoding and decoding algorithms (non-modular), which are series-parallel in nature. Thus, the actual problem arises of reducing the hardware costs of the non-modular operations. In this paper, we propose a fundamentally new approach to the construction of computing channels modular device based on the bimodular arithmetic with internal redundancy in the representation of the operands. The proposed method have the same type of encoding components of the representation deductions allowed to reduce the computation of the module p to module (p-1). Thus, it was possible to realize the parallel structure of the modular channel and organize the control calculations for each mod p modular device. The experimental results showed that the developed set of methods to improve the protection of the modular unit from failure reduces hard- ware costs compared to redundant - the cost of monitoring equipment amounted to 40 % compared to an unprotected circuit, the loss of productivity by 7%.

Download PDF

Keywords Redundant RNS codes; bimodular arithmetic; architectural failure protection.
References 1. Kalyaev I.A., Levin I.I., Semernikov E.A., Shmoylov V.I. Rekonfiguriruemye mul'tikonveyernye vychislitel'nye struktury [Multiconference reconfigurable computing structures]. Rostov-on-Don: Izd-vo: YuNTs RAN, 2008, 320 p.
2. Osipenko P. Odinochnye sboi – vyzov dlya sovremennykh mikroprotsessorov [Single failures – a challenge for modern microprocessors], Elektronnye komponenty [Electronic Components], 2009, No. 7, pp 12-15.
3. Iyudu K.A. Nadezhnost', kontrol' i diagnostika vychislitel'nykh mashin i system [Reliability, control and diagnostics of computers and computer systems]. Moscow: Vysshaya shkola, 1989, 216 p.
4. Dadaev Yu.G. Teoriya arifmeticheskikh kodov [The theory of arithmetic codes]. Moscow: Radio i svyaz', 1981, 272 p.
5. Torgashev V.A. Sistema ostatochnykh klassov i nadezhnost' TsVM [The system of residual classes and the reliability of the CVM]. Moscow: Sov. Radio, 1973, 118 p.
6. Amerbaev V.M., Balaka E.S., Solov'ev R.A., Tel'pukhov D.V. Postroenie obratnykh preobrazovateley modulyarnoy arifmetiki s korrektsiey oshibok na baze poliadicheskogo koda [The construction of inverter modular arithmetic with the error correction code on the basis politicheskogo], Neyrokomp'yutery: razrabotka i primenenie [Neurocomputers: development
and application], 2014, No. 9, pp. 30-36.
7. Akushskiy I.Ya., Yuditskiy D.I. Mashinnaya arifmetika v ostatochnykh klassakh [Machine arithmetic in residual classes]. Moscow: Sovetskoe radio, 1968, 440 p.
8. Stempkovskiy A.L., Amerbaev V.M. Printsip faktorizatsii v probleme proektirovaniya modulyarnykh protsessorov [The principle of factorization in the problem of designing modular processors], VI Vserossiyskaya nauchno-tekhnicheskaya konferentsiya «Problemy razrabotki perspektivnykh mikro- i nanoelektronnykh sistem – 2014»: Sb. trudov [VI all-Russian scientific-technical conference "Problems of development of perspective micro- and
nanoelectronic systems – 2014": the collected works], Under the General ed. ak. RAS Stempkovskogo A.L. Moscow: IPPM RAN, 2014. Part IV, pp. 183-186.
9. Amerbaev V.M., Malashevich D.B. Analiz effektivnosti realizatsii modul'nykh operatsiy indeksnoy modulyarnoy arifmetiki [Analysis of the effectiveness of the implementation of modular operations, modular arithmetic index], Izvestiya vuzov. Elektronika [Izvestiya vuzov. Electronics], 2009, No. 80, pp. 54-57.
10. Stempkovskiy A.L., Amerbaev V.M., Kornilov A.I. Modulyarnaya logarifmetika – novye vozmozhnosti dlya proektirovaniya modulyarnykh vychisliteley i preobrazovateley [Modular logaritmica – new opportunities for the design of modular calculators and converters], IV Vserossiyskaya nauchnotekhnicheskaya konferentsiya «Problemy razrabotki perspektivnykh mikro- i nanoelektronnykh sistem – 2010»: Sb. nauchn. tr. [IV all-Russian scientific-technical conference "Problems of development of perspective micro- and nanoelectronic systems – 2010": Collection of scientific papers], Under the General ed. ak. RAS Stempkovskogo A.L. Moscow: IPPM RAN, 2010.
11. Amerbaev V.M., Balaka E.S. Bimodul'nye vychisleniya nad polem Galua GF(p) [Bimodule computation over Galois field GF(p)], Vestnik Moskovskoy gosudarstvennoy akademii delovogo administrirovaniya. Seriya: Ekonomika [Bulletin of the Moscow state Academy of business administration. Series: Economics], 2013, No. 1 (20), pp. 36-42.
12. Stempkovskiy A.L., Amerbaev V.M., Solov'ev R.A. Printsipy rekursivnykh modulyarnykh vychisleniy [The recursive principles of modular computing], Informatsionnye tekhnologii [Information Technologies], 2013, No. 2, pp. 22-27. 13. Amerbaev V.M., Balaka E.S., Solov'ev R.A., Tel'pukhov D.V. Analiz i sintez arifmetiche-skogo uzla prof. Pospelova D.A. polya Galua [Analysis and synthesis of arithmetic node Prof. Pospelov D. A. Galois field], VI Vserossiyskaya nauchno-tekhnicheskaya konferentsiya «Problemy razrabotki perspektivnykh mikro- i nanoelektronnykh sistem – 2014»: Sb. Trudov [VI all-Russian scientific-technical conference "Problems of development of perspective micro- and nanoelectronic systems – 2014": the collected works], Under the General ed. ak. RAS Stempkovskogo A.L. Moscow: IPPM RAN, 2014, Part IV, pp. 179-182.
14. Pospelov D.A. Arifmeticheskie osnovy vychislitel'nykh mashin diskretnogo deystviya [Arithmetic basics of computers discrete action]. Moscow: Vysshaya shkola, 1970, 308 p.
15. Stempkovskiy A.L., Amerbaev V.M., Kornilov A.I. Modulyarnaya logarifmetika – novye vozmozhnosti dlya proektirovaniya modulyarnykh vychisliteley i preobrazovateley (kratkiy obzor) [Modular logaritmica – new opportunities for the design of modular calculators and converters (a brief overview)], IV Vserossiyskaya nauchno-tekhnicheskaya konferentsiya
«Problemy razrabotki perspektivnykh mikro- i nanoelektronnykh sistem – 2010»: Sb. nauchn. tr. [IV all-Russian scientific-technical conference "Problems of development of perspective micro- and nanoelectronic systems – 2010": Collection of scientific papers], Under the General ed. ak. RAS Stempkovskogo A.L. Moscow: IPPM RAN, 2010.
16. Amerbaev V.M., Balaka E.S., Konstantinov A.V., Tel'pukhov D.V. Metody uskoreniya vychisleniy skalyarnykh proizvedeniy vektorov v bazise modulyarnoy logarifmetiki [Methods to accelerate the computation of scalar products of vectors in the basis of modular logarithmitic], IV Vserossiyskaya nauchno-tekhnicheskaya konferentsiya «Problemy razrabotki perspektivnykh mikro- i nanoelektronnykh sistem – 2010»: Sb. trudov [IV all-Russian scientific-technical conference "problems of development of perspective micro - and
nanoelectronic systems – 2010": proceedings of the], Under the General ed. ak. RAS Stempkovskogo A.L. Moscow: IPPM RAN, 2010, pp. 378-381.
17. Amerbaev V.M., Balaka E.S., Shchelokov A.N. Primenenie strukturnoy izbytochnosti dlya povysheniya nadezhnosti arifmeticheskogo uzla vychislitel'nogo elementa bimodul'noy arifmetiki [Use of structural redundancy for increase of reliability of arithmetic unit of the computing element of bimodular arithmetic], Izvestiya YuFU. Tekhnicheskie nauki [Izvestiya SFedU. Engineering Sciences], 2014, No. 7, pp. 248-254.
18. Kornilov A.I., Isaeva T.Yu., Semenov M.Yu. Metody logicheskogo sinteza summatorov s uskorennym perenosom po modulyu (2n-1) na osnove BDD-tekhnologii [Methods of logic synthesis of adders with accelerated migration modulo (2n-1) based on BDD technologies], Izvestiya vuzov. Elektronika [News of higher educational institutions. Electronics], 2004, No. 3, pp. 54-60.
19. Amerbaev V.M. Teoreticheskie osnovy mashinnoy arifmetiki [Theoretical foundations of computer arithmetic]. Alma-Ata: Nauka, 1976, 324 p.
20. Kalashnikov V.S. Issledovanie i razrabotka metodov proektirovaniya bystrodeystvuyushchikh vychislitel'nykh uzlov dlya realizatsii otkazoustoychivykh sistem na osnove modulyarnoy arifmetiki: Dis. … kand. tekh. nauk: 05.13.05 [Research and development of design techniques
for high-performance computing nodes to implement failover systems based on modular arithmetic. Cand. eng. sc. diss.]. Moscow, 2007.
21. Jaberipur and S. Nejati. Balanced minimal latency RNS addition for moduli set {2n-1,2n,2n+1}, in Proc. 18th Int. Conf. Systems, Signals and Image Processing (IWSSIP), 2011, pp. 1-7.

Comments are closed.