Acceleration of Higher Order FEM Matrix Filling by OpenMP Parallelization of Volume Integrations
Апстракт
Parallelization of time-consuming volume integrations in the matrix filling is performed using OpenMP to provide a smooth transition of a higher order finite element method (FEM) code for electromagnetic analysis to the multi-core hardware architecture. In the process, the carefully constructed original computational scheme suited to maximally avoid the redundant computations in the higher order FEM is not affected. The results, in analysis of an electrically large dielectric scatterer, demonstrate acceleration of up to 2.28 times in a four-thread architecture.
Кључне речи:
finite element methods / higher order methods / OpenMP / parallelizationИзвор:
20th Telecommunications Forum (TELFOR), 2012, 1183-Финансирање / пројекти:
- Алгоритми и софтвер за симулације у фреквенцијском и временском домену RF подсистема и електромагнетских сензора у ICT (RS-MESTD-Technological Development (TD or TR)-32005)
- National Science Foundation [ECCS-1002385]
Напомена:
- 20th Telecommunications Forum, Proceedings, Nov 20-22, 2012, Belgrade, Serbia
Колекције
Институција/група
VinčaTY - CONF AU - Savić, Slobodan V. AU - Ilić, Anđelija Z. AU - Notaros, Branislav M. AU - Ilić, Milan M. PY - 2012 UR - https://vinar.vin.bg.ac.rs/handle/123456789/6995 AB - Parallelization of time-consuming volume integrations in the matrix filling is performed using OpenMP to provide a smooth transition of a higher order finite element method (FEM) code for electromagnetic analysis to the multi-core hardware architecture. In the process, the carefully constructed original computational scheme suited to maximally avoid the redundant computations in the higher order FEM is not affected. The results, in analysis of an electrically large dielectric scatterer, demonstrate acceleration of up to 2.28 times in a four-thread architecture. C3 - 20th Telecommunications Forum (TELFOR) T1 - Acceleration of Higher Order FEM Matrix Filling by OpenMP Parallelization of Volume Integrations SP - 1183 UR - https://hdl.handle.net/21.15107/rcub_vinar_6995 ER -
@conference{ author = "Savić, Slobodan V. and Ilić, Anđelija Z. and Notaros, Branislav M. and Ilić, Milan M.", year = "2012", abstract = "Parallelization of time-consuming volume integrations in the matrix filling is performed using OpenMP to provide a smooth transition of a higher order finite element method (FEM) code for electromagnetic analysis to the multi-core hardware architecture. In the process, the carefully constructed original computational scheme suited to maximally avoid the redundant computations in the higher order FEM is not affected. The results, in analysis of an electrically large dielectric scatterer, demonstrate acceleration of up to 2.28 times in a four-thread architecture.", journal = "20th Telecommunications Forum (TELFOR)", title = "Acceleration of Higher Order FEM Matrix Filling by OpenMP Parallelization of Volume Integrations", pages = "1183", url = "https://hdl.handle.net/21.15107/rcub_vinar_6995" }
Savić, S. V., Ilić, A. Z., Notaros, B. M.,& Ilić, M. M.. (2012). Acceleration of Higher Order FEM Matrix Filling by OpenMP Parallelization of Volume Integrations. in 20th Telecommunications Forum (TELFOR), 1183. https://hdl.handle.net/21.15107/rcub_vinar_6995
Savić SV, Ilić AZ, Notaros BM, Ilić MM. Acceleration of Higher Order FEM Matrix Filling by OpenMP Parallelization of Volume Integrations. in 20th Telecommunications Forum (TELFOR). 2012;:1183. https://hdl.handle.net/21.15107/rcub_vinar_6995 .
Savić, Slobodan V., Ilić, Anđelija Z., Notaros, Branislav M., Ilić, Milan M., "Acceleration of Higher Order FEM Matrix Filling by OpenMP Parallelization of Volume Integrations" in 20th Telecommunications Forum (TELFOR) (2012):1183, https://hdl.handle.net/21.15107/rcub_vinar_6995 .