Authors D.A. Sorokin, A.Yu. Matrosov, E.E. Semernikova, K.N. Alekseev
Month, Year 12, 2016 @en
Index UDC 004.382.2
DOI 10.18522/2311-3103-2016-12-1628
Abstract The paper covers peculiarities of implementation of the multiple prediction problem on high-performance computer systems with the help of the SRMP algorithm. The algorithm belongs to the class of computationally laborious tightly coupled tasks. For such tasks the number of data interprocessor exchange, and the number of data exchange between processors and memory units are similar or even exceed the number of operations. Effective implementation of this task requires combination of multiple channels and non-linear data access. Such requirements cannot be fulfilled if we use computer systems with traditional (cluster) architecture. Therefore we suggest an alternative approach to the SRMP problem, based on an idea of design of a single computational circuit on the base of reconfigurable computer systems (RCS). The principal feature of the RCS is the use of FPGAs as the main computational component capable to adapt to the computational structure of the solving task. Structural procedural organization of calculations is one of the most effective for RCS, and it implies direct mapping of the basic information graph of the task on the computational field of the system. Such implementation of the SRMP problem requires computa-tional resource, which is not available in modern RCS. Taking into account all peculiarities of the algorithm, we have suggested a method, owing to which it is possible to transform the basic infor-mation graph of the task and to develop the structure of the computational pipeline. In this case it is possible to synthesize several pipelines in each computational FPGA according to its available computational resource. Besides, due to non-linear access to initial data we suggest a specific organization of calculations which allows each pipeline to operate independently. The suggested structural procedural single computational circuit implementation of the SRMP problem can be easily scaled. It provides real-time data processing, does not require high-rate inter-chip exchange and keeping intermediate results of calculations. As a result, when the available computational resource grows, the performance of the task will grow practically linearly.

Download PDF

Keywords Reconfigurable computer systems; FPGA; prediction of ringing interfering waves (multiple prediction); SRMP-algorithm; structural procedural organization of calculations.
References 1. Khmelevskoy V.K. Geofizicheskie metody issledovaniya zemnoy kory [Geophysical methods of exploration of the earth's crust]. Book 1. Dubna: Mezhdunarodnyy universitet prirody, obshchestva i cheloveka "Dubna", 1997, 203 p.
2. Denisov M.S., Finikov D.B. Metody podavleniya kratnykh voln v seysmorazvedke [Methods of suppression of multiples in seismic prospecting]. Part 1, Tekhnologii seysmorazvedki [Seismic Technology], 2007, No. 1, pp. 5-16.
3. McHugo S., Webb B., Grechishnikova T., Whitebread R. Revealing the reservoir through 3D multiple attenuation // ROGTEC. WesternGeco. – 2014. – No. 10. – P. 22-34.
4. Kurin E.A., Denisov M.S. Primenenie vysokoproizvoditel'nykh vychislitel'nykh sistem v zadache podavleniya mnogokratno otrazhennykh voln-pomekh [The use of high performance computing systems in the task of suppressing multiply reflected waves-interference], Tekhnologii seysmorazvedki [Seismic Technologies], 2011, No. 4, pp. 35-40.
5. Kurin E.A. Seysmorazvedka i superkomp'yutery [Seismic and supercomputers], Vychislitel'nye metody i programmirovanie [Computing methods and programming], 2011, No. 12, pp. 34-39.
6. Verschuur D.J., Berkhout A.J., Wapenaar C.P. A. Adaptive surface-related multiple elimination, Geophysics, 1992, No. 9, pp. 1166-1177.
7. Berkhout, A.J., Verschuur D.J. Estimation of multiple scattering by iterative inversion, Part I: theoretical considerations, Geophysics, 1997, No. 5, pp. 1586-1595.
8. Barnes S., van Borselen R., Salazar H., Vàzquez A., Ronzón I., Martinez R. Application of True-Azimuth 3D SRME to an Onshore Mexican Data Set // Society of exploration geophysi-cists: Сollected papers of the International conference, Houston, Texas, USA, 2011.
9. van Dedem E.J., Verschuur D.J. 3D surface-related multiple prediction: A sparse inversion approach, Geophysics, 2005, No. 3, pp. 31-43.
10. Huang X., Sun C., Niu B., Wang H., Zeng M. Surface-related multiple prediction and suppres-sion based on data-consistence: a theoretical study and test, Chinese journal of geophysics, 2005, No. 1, pp. 188-196.
11. Artman B. Passive seismic imaging: cand. of eng. sc. diss. Stanford, 2007, pp. 126-132.
12. Artman B., Alvarez G., Matson K. Image-space surface-related multiple prediction, Geophysics, 2007, No. 2, pp. 113-122.
13. Wang J., Wang S. Practical implementation of SRME for land multiple attenuation, GeoConvention: Integration: Sbornik statey mezhdunarodnoy konferentsii [GeoConvention: Integration: Collected papers of the international conference], Calgary, Canada, 2013.
14. Verschuur E., Dragoset B., Moore I., Bisley R. A perspective on 3D surface-related multiple elimination, Geophysics, 2010, No. 5, pp. 245-261.
15. Kalyaev A.V., Levin I.I. Modul'no-narashchivaemye mnogoprotsessornye sistemy so strukturno-protsedurnoy organizatsiey vychisleniy [Modular-scalable multiprocessor system with structural-procedural organization of computing]. Moscow: Yanus-K, 2003, 380 p.
16. Levin I.I. Rekonfiguriruemye mul'tikonveyernye vychislitel'nye struktury [Multiconference reconfigurable computing structure]. Rostov-on-Don: YuNTs RAN, 2008, pp. 3-13.
17. Levin I.I., Pelipets A.V., Sorokin D.A. Reshenie zadachi LU dekompozitsii na rekonfiguriruemykh vychislitel'nykh sistemakh: otsenka i perspektivy [Estimation and prospects of solving LU-decomposition on reconfigurable computer systems], Izvestiya YuFU. Tekhnicheskie nauki [Izvestiya SFedU. Engineering Sciences], 2015, No. 7 (168), pp. 62-70.
18. Kalyaev I.A., Levin I.I. Rekonfiguriruemye mul'tikonveyernye vychislitel'nye sistemy dlya resheniya potokovykh zadach [Multiconference reconfigurable computing systems for the so-lution of flow problems], Informatsionnye tekhnologii i vychislitel'nye sistemy [Information technology and computer systems], 2011, No. 2, pp. 12-22.
19. Sorokin D.A., Dordopulo A.I. Metodika sokrashcheniya apparatnykh zatrat v slozhnykh sistemakh pri reshenii zadach s sushchestvenno-peremennoy intensivnost'yu potokov dannykh [A methodology of hardware overhead decrease in complex systems while solving tasks with considerably variable data flow density], Izvestiya YuFU. Tekhnicheskie nauki [Izvestiya SFedU. Engineering Sciences], 2012, No. 4 (129), pp. 213-219.
20. Sorokin D.A. Metody resheniya zadach s peremennoy intensivnost'yu potokov dannykh na rekonfiguriruemykh vychislitel'nykh sistemakh: dis. … kand. tekhn. nauk [Methods of solution of problems with variable intensity data streams in reconfigurable computing systems. Cand. of eng. sc. diss.]. Taganrog, 2012, pp. 51-58.
21. Rabiner L., Gold B. Teoriya i primenenie tsifrovoy obrabotki signalov [Theory and application of digital signal processing]. Moscow: Mir, 1978, 848 p.
22. Vvedenie v tsifrovuyu fil'tratsiyu [Introduction to digital filtering], ed. by R. Bognera and
A. Konstantinidisa. Moscow: Mir, 1976, 216 p.
23. Semernikov E.A., Doronchenko Yu.I., Trunov I.L. Makroprotsessor tsifrovoy obrabotki signalov dlya mnogoprotsessornykh vychislitel'nykh sistem so strukturno-protsedurnoy organizatsiey vychisleniy [Macro processor digital signal processing for multiprocessor systems with structural-procedural organization of computing], Materialy Mezhdunarodnoy nauchnoy konferentsii “Iskusstvennyy intellekt. Intellektual'nye i mnogoprotsessornye sistemy-2004” [Materials of the International scientific conference “Artificial intelligence. Intelligent and multiprocessor systems-2004”], 2004, pp. 80-84.
24. Semernikov E.A., Doronchenko Yu.I. Konveyernyy makroprotsessor tsifrovoy obrabotki signalov so strukturno-protsedurnoy organizatsiey vychisleniy [Conveyor macro processor digital signal processing with structural-procedural organization of computing], Vestnik komp'yuternykh i informatsionnykh tekhnologiy [Vestnik of computer and information tech-nology], 2005, No. 8, pp. 49-55.

Comments are closed.