Article information
2017 , Volume 22, ¹ 3, p.1631
Glinskikh V.N., Dudaev A.R., Nechaev O.V.
Highperformance CPU  GPU heterogeneous computing in resistivity logging of oil and gas wells
The work is concerned with the development of numerical algorithms for solving forward problems of borehole geoelectrics by applying high performance GPU computing. Studying new types of complex hydrocarbon reservoirs demands the improvement of logging tools and software for data processing. The improvement of processing accuracy is feasible with the use of numerical solutions in full mathematical formulations. Computational problems in resistivity logging are known to be resourceintensive and not very effective when used in practice. This research considers one of the ways to speed up calculations, namely through the application of NVIDIA graphics processors. We have developed and implemented into software an algorithm for the simulation of resistivity logging data from oil and gas wells, by making use of highperformance CPU  GPU heterogeneous computations. The numerical solution of the 2D forward problem is based on the finiteelement method and the Cholesky decomposition for solving a system of linear equations. The software implementations of the algorithm are made by means of NVIDIA CUDA technology and computing libraries making it possible to decompose the equation system and find its solution on CPU and GPU. The analysis of computing time as a function of the order of matrix and number of nonzero elements has shown that in the case at hand the computations are the most effective when decomposing on GPU and finding a solution on CPU. We have estimated the operational speed of CPU and GPU computations, including highperformance heterogeneous CPU  GPU ones. It is found that heterogeneous CPU  GPU computations enable speeding up in comparison with similar CPU or GPU calculations. Using the developed algorithm, we have simulated resistivity data in realistic models. The results of our investigation point to a high efficiency of the algorithm in respect to dealing with a wide variety of practical problems
[full text] Keywords: graphics processing units, CPU  GPU heterogeneous computing, parallel algorithm, finiteelement method, 2D forward problem, resistivity logging
Author(s): Glinskikh Viacheslav NikolaevichPhD. , Associate Professor Position: Head of Laboratory Office: Trofimuk Institute of Petroleum Geology and Geophysics SB RAS Address: 630090, Russia, Novosibirsk
Phone Office: (383) 3304505 Email: GlinskikhVN@ipgg.sbras.ru Dudaev Alexander RuslanovichPosition: assistant Office: Trofimuk Institute of Petroleum Geology and Geophysics SB RAS Address: 630090, Russia, Novosibirsk
Email: DudaevAR@ipgg.sbras.ru Nechaev Oleg ValentinovichOffice: Trofimuk Institute of Petroleum Geology and Geophysics SB RAS, Novosibirsk State University Address: 630090, Russia, Novosibirsk, Acad. Koptyug av. 3
Email: oleg.v.nechaev@gmail.com References: [1] Epov, M. I., Glinskikh, V. N., Nikitenko, M. N., Sukhorukova C. V. SKL complex for logging during one roundtrip operation: techniques and practical data interpretation. Puti realizatsii neftegazovogo i rudnogo potentsiala KhantyMansiyskogo avtonomnogo okruga – Yugry: Materialy Pyatnadtsatoy nauchnoprakticheskoy konferentsii. KhantyMansiysk: IzdatNaukaServis. 2012; (2):27–36. ( In Russ.) [2] Epov, M. I., Glinskikh, V. N. Fast twodimensional simulation of highfrequency electromagnetic field in induction logging. Russian Geology and Geophysics. 2003; 44(9):904–915. [3] Epov, M I., Glinskikh, V. N. Linearization of relative parameters of a highfrequency magnetic field in twodimensional conducting media. Russian Geology and Geophysics. 2004; 45(2):247–257. [4] Glinskikh, V. N., Epov, M. I. Locally nonlinear approximation of highfrequency electromagnetic field for logging applications. Russian Geology and Geophysics. 2006; 47(8):930–936. [5] Eremin, V. N., Nechaev, O. V., Haberkhauer, S., Shokina, N. Yu., Shurina, E. P. Parallel realization of mathematical modelling of electromagnetic logging processes using VIKIZ probe complex. Computational technologies. 2007; 12(6):18–33. (In Russ.) [6] Epov, M. I., Shurina, E. P., Arkhipov, D. A. Parallel numerical finite elements schemes for solving of geoelectric problems. Computational technologies. 2013; 18(2):95–112. (In Russ.) [7] Glinskikh, V. N., Nesterova, G. V. Epov, M. I. Forward modeling and inversion of induction logs from shaly sand reservoirs using petrophysical conductivity models. Russian Geology and Geophysics. 2014; 55(5–6):793–799. [8] Epov, M. I., Glinskikh, V. N., Nikitenko, M. N., Sukhorukova, C. V. Numerical simulation and analysis of electromagnetic loggingwhiledrilling data. Karotazhnik. 2014; 11(245):29–41. (In Russ.) [9] Glinskiy, B. M., Kostin, V. I., Kuchin, N. V., Solovyev, S. A., Cheverda, V. A. Aspects of parallel computing to solve Helmholtz equation by a direct solver with lowrank approximation and the HSS format of data storage. Numerical Methods and Programming. 2015; (16):607–616. (In Russ.) [10] Kostin, V. I., Reshetova, G. V., Tcheverda, V. A. Parallel numerical simulation of 3D acoustic logging. Matematicheskoe Modelirovanie. 2008; 20(9):51–66. (In Russ.) [11] Belonosov, M. A., Kostov, C., Reshetova G. V., Solovev, S. A., Cheverda, V. A. Parallel numerical simulation of seismic waves propagation with Intel Math Kernel Library. Lecture Notes in Computer Science. Applied Parallel and Scientific Computing, 2013; (7782):153–167. [12] Puzyrev, V., Koric, S., Wilkin, St. Evaluation of parallel direct sparse linear solvers in electromagnetic geophysical problems. Computers and Geosciences. 2016; (89):79–87. [13] Glinskikh, V. N., Epov, M. I., Labutin, I. B. Simulation of electromagnetic logs using graphics processing unit. Computational technologies. 2008; 13(6):50–60. (In Russ.) [14] Glinskikh, V. N., Bulantseva, Yu. O. Mathematical simulation of electromagnetic logs using highperformance coprocessor Intel Xeon Phi. Vestnik of NGU. Series: Mathematics, Mechanics, Informatics. 2014; 14(4):11–22. (In Russ.) [15] Glinskikh, V. N., Gorbatenko, V. A. Electromagnetic logging data inversion on GPU. Computational technologies. 2015; 20(1):25–37. (In Russ) [16] Labutin, I. B., Surodina, I. V. Algorithm for Sparse Approximate Inverse Preconditioners in the Conjugate Gradient Method. Reliable Computing. 2013; (19):120–126. [17] Dudaev, A. R., Nechaev, O. V., Glinskikh, V. N. Highperformance computing on GPU for 2D electric logging problem based on the finite element methods. Proceedings of the 8th International Siberian Early Career GeoScientists Conference. Novosibirsk; 390–391. [18] Ñóðîäèíà È. Â., Íåñòåðîâà Ã. Â. Ìîäåëèðîâàíèå ïîêàçàíèé çîíäîâ ÂÈÊÈÇ è ÁÊÇ íà ãðàôè÷åñêèõ ïðîöåññîðàõ // Ïåòðîôèçèêà ñëîæíûõ êîëëåêòîðîâ: ïðîáëåìû è ïåðñïåêòèâû 2015: Ñáîðíèê ñòàòåé.  Ì.: ÅÀÃÅ Ãåîìîäåëü, 2015. Ñ. 85–94. Surodina I. V., Nesterova G. V. Modelirovanie pokazaniy zondov VIKIZ i BKZ na graficheskikh protsessorakh [Simulation of VIKIZ and BKZ logs on GPUs]. Petrophysics of complex reservoirs: challenges and opportunities 2015: collection of articles. Ìoscow: EAGE Geomodel; 2015:85–94. (In Russ) [19] Epov, M. I., Sukhorukova, C. V., Glinskikh, V. N., Nikitenko, M. N., Nechaev, O. V., Surodina, I. V. Effective electromagnetic log data interpretation in realistic reservoir models. Open Journal of Geology. 2013; 3(2B):81–86. [20] Davis, T. A., Rajamanickam, S., SidLakhdar, W. M. A survey of direct methods for sparse linear systems. Acta Numerica. 2016; (25):383–566. [21] Pierce, D. J., Hung, Y., Liu, C. C., Tsai, Y. H., Wang, W., Yu, D. Sparse multifrontal performance gains via NVIDIA GPU. Workshop on GPU Supercomputing, 2009, National Taiwan University, Taipei. Available at: http://cqse.ntu.edu.tw/cqse/download_file/DPierce_20090116.pdf (accessed 30.09.16). [22] Lacoste, X., Ramet, P., Faverge, M., Ichitaro, Y., Dongarra, J. Sparse direct solvers with accelerators over DAG runtimes. Research Report RR7972, INRIA, Bordeaux, France; 2012: 11. Available at: https://hal.inria.fr/hal00700066v2/document (accessed 30.09.16). [23] Mittal, S., Vetter, J. S. A survey of CPUGPU heterogeneous computing techniques. ACM Computing Surveys. 2015; 47(4):69:1–69:35. [24] Sao, P., Vuduc, R., Li, X. S. A distributed CPUGPU sparse direct solver. Proc. EuroPar 2014 Parallel Processing, 2014. Silva F. et al (eds) Lecture Notes in Computer Science. 2014; (8632):487–498. [25] Dakhnov, V. N. Elektricheskie i magnitnye metody issledovaniya skvazhin [Electrical and magnetic borehole survey techniques]. Moscow: Nedra; 1981: 344. (In Russ.) [26] Shaydurov V. V. Mnogosetochnye metody konechnykh elementov [Multigrid methods for finite elements]. Moscow: Nauka; 1989: 288. (In Russ.) [27] CUDA C Programming Guide. Design Guide // NVIDIA CUDA. V 8.0.61. 2017. Available at: http://docs.nvidia.com/cuda/pdf/CUDA_C_Programming_Guide.pdf (accessed 30.09.16). [28] cuSOLVER // NVIDIA CUDA Toolkit Documentation. 2017. Available at: http://docs.nvidia.com/cuda/cusolver (accessed 30.09.16). [29] Chen, Y., Davis, T. A., Hager, W. W., Rajamanickam, S. Algorithm 887: CHOLMOD, supernodal sparse Cholesky factorization and update/downdate. ACM Transactions on Mathematical Software. 2008; 35(3): 22:1–22:14. [30] SuiteSparse: A Suite of Sparse Matrix Software v.4.5.5. Davis T.A. SuiteSparse. 2017. Available at: http://www.suitesparse.com (accessed 30.09.16) ïåðåâîäèò íà http://faculty.cse.tamu.edu/davis/welcome.html
Bibliography link: Glinskikh V.N., Dudaev A.R., Nechaev O.V. Highperformance CPU  GPU heterogeneous computing in resistivity logging of oil and gas wells // Computational technologies. 2017. V. 22. ¹ 3. P. 1631
