Approach to planning the load of processors of critical multiprocessor systems

DOI: 10.34759/trd-2022-126-19

Аuthors

Sokolova Y. V.^*, Leun E. V.^**, Primakov P. V.^***, Samoylov S. Y.^****

Lavochkin Research and Production Association, NPO Lavochkin, 24, Leningradskay str., Khimki, Moscow region, 141400, Russia

*e-mail: jv.sokolova@mail.ru
**e-mail: leunev@laspace.ru
***e-mail: pavel.primakov@laspace.ru
****e-mail: SS2916@mail.ru

Abstract

One of the priority areas in economically developed countries is the security of critically important, especially complexly organized systems. Examples of such systems are a variety of robotic production; nuclear power plant control systems; onboard computing systems; groups of unmanned aerial, land and water robots; global navigation satellite system (GLONASS); large software systems of high importance and many other systems.

The main problem of building computing systems at all times remains the task of ensuring their long-term functioning. This task has three components: reliability, availability and serviceability.

Of particular relevance is the use of computer systems for managing critical objects operating in real time.

The main difference between real-time operating systems and any other operating systems is the guarantees for the start or end time of processes that are provided by real-time systems.

In the event of a failure, such systems are subject to high requirements for operability, non-failure operation, safety, security, etc. Obviously, the most important thing is to minimize the time and hardware costs required for the response of a multiprocessor system to an emergency situation.

One of the options for solving this issue may be planning the load of processors in multiprocessor systems. In this case, you can avoid simultaneous loading of several processors by one task (program, subroutine, algorithm, file, etc.) and, at the same time, schedule the queue of incoming tasks in such a way that they are served simultaneously. This allows you to reduce unplanned downtime and at the same time increase its availability along with increased speed.

The article is devoted to multiprocessor systems. The issue of compiling a plan for loading processors in them is touched upon. It is supposed to use the so-called real-time systems.

Keywords:

multiprocessor system, scheduling, loading, high availability, assignment, method, algorithm, schedule

References

1. Tsil’ker B.Ya. Organizatsiya EVM i system (Organization of computers and systems), Saint Petersburg, Piter, 2007, 668 p.
2. Ore O. Teoriya grafov (Theory of Graphs), Moscow, Nauka, 1968, 352 p.
3. Kormen T.M. et al. Algoritmy: postroenie i analiz. Algoritmy dlya raboty s grafami. Ch. VI. (Algorithms: construction and analysis. Algorithms for working with graphs), Moscow, Izdatel’skii dom «Vil’yams», 2006, 1296 p.
4. Stallings W. Computer organization and architecture, Prentice-Hall, 1999, 881 p.
5. Levin I.I., Shteinberg B.Ya. Iskusstvennyi intellect, 2001, no. 3, pp. 234-242.
6. Slyusar V. Elektronika: Nauka, tekhnologiya, biznes, 2007, no. 1 (75), pp. 92-97.
7. John L. Hennessy, David A. Patterson. Computer Architecture: A Quantitative Approach, Morgan Kaufmann, 2003, 883 p.
8. Mikushin A.V., Sazhnev A.M., Sedinin V.I. Tsifrovye ustroistva i mikroprotsessory (Digital devices and microprocessors), Saint Petersburg, BKhV-Peterburg, 2010, 832 p.
9. Busurin V.I., Medvedev V.M., Karabitskii A.S., Groppa D.V. Trudy MAI, 2017, no. 97. URL: https://trudymai.ru/eng/published.php?ID=87277
10. Basov R.G., Borzov D.B. XIV Mezhdunarodnaya nauchno-tekhnicheskaya konferentsiya «Optiko-elektronnye pribory i ustroistva v sistemakh raspoznavan iya obrazov, obrabotki izobrazhenii i simvol’noi informatsii»: sbornik materialov, Kursk, Yugo-Zapadnyi gosudarstvennyi universitet, 2018, pp. 64-66.
11. Basov R.G., Borzov D.B., Loktionova O.G. Svidetel’stvo o gosudarstvennoi registratsii programm na EVM № 2019616927 RF, 30.05.2019.
12. Trakhtengerts E.A. Vvedenie v teoriyu analiza i rasparallelivaniya programm EVM v protsesse translyatsii (Introduction to the theory of analysis and parallelization of computer programs in the translation process), Moscow, Nauka, 1981, 254 p.
13. Tyutlyaeva E.O., Konyukhov S.S., Odintsov I.O., Moskovsky A.A. Seismic Pro-cessing Performance Analysis on Different Hardware Environment, Seismic Pro-cessing Performance Analysis on Different Hardware Environment, 2017, vol. 4, no. 3, pp. 80 — 90. DOI: 10.14529/jsfi170305
14. Khokni R., Dzhesskhoup K. Parallel’nye EVM. Arkhitektura, programmirova-nie i algoritmy (Parallel computers. Architecture, programming and algorithms), Moscow, Radio i svyaz’, 1986, 392 p.
15. Kalyaev A.B., Levin I.I. Modul’no-narashchivaemye mnogoprotsessornye si-stemy so strukturno-protsedurnoi organizatsiei vychislenii (Modularly expandable multiprocessor systems with structural and procedural organization of computations), Moscow, Yanus-K, 2003, 379 p.
16. Borzov D.B., Chernetskaya I.E. Proektirovanie protsessora EVM (Computer processor design), Kursk, Yugo-Zapadnyi gosudarstvennyi universitet, 2020, 199 p.
17. Momose S. SX-Aurora TSUBASA. Brand-new Vector Supercomputer, SC’17 Supercomputer Forum, 2017. URL: https://www.osp.ru/os/2018/01/13053934
18. Kuz’minskii M. Otkrytye sistemy SUBD, 2016, no. 3, pp. 4–6. URL: https://www.osp.ru/os/2016/03/13050252
19. Markaryan A.O., Churkov I.S. Trudy MAI, 2020, no. 113. URL: http://trudymai.ru/eng/published.php?ID=118150. DOI: 10.34759/trd-2020-113-10
20. Nabatov A.N., Vedenyapin I.E., Mukhtarov A.R. Trudy MAI, 2018, no. 102. URL: http://trudymai.ru/eng/published.php?ID=99177

Download