论文信息 - Electromagnetic Device Optimization: The Forking of Already Parallelized Threads on Graphics Processing Units

Electromagnetic Device Optimization: The Forking of Already Parallelized Threads on Graphics Processing Units

─ In light of the new capability to fork an already parallelized kernel on a GPU, this paper shows how the use of the parallelization capabilities of a PC’s Graphics Processing Unit (GPU) makes the finite element design of coupled problems (such as the electroheat shape optimization problems we work with) realistic and practicable in terms of computational time. Index Terms Finite elements, GPU computing, inverse problems, parallelization.

[1] Eric Darve,et al. Assembly of finite element methods on graphics processors , 2011 .

[2] Tien-Tsin Wong,et al. Implementation of parallel genetic algorithms on graphics processing units , 2009 .

[3] S. Gyimothy,et al. Parallel Realization of the Element-by-Element FEM Technique by CUDA , 2012, IEEE Transactions on Magnetics.

[4] S. Ratnajeevan H. Hoole,et al. CUDA memory limitation in finite element optimization to reconstruct cracks , 2014 .

[5] T. Hughes,et al. An element-by-element solution algorithm for problems of structural and solid mechanics , 1983 .

[6] Lalita Udpa,et al. Faster, more accurate, parallelized inversion for shape optimization in electroheat problems on a graphics processing unit (GPU) with the real-coded genetic algorithm , 2015 .

[7] E. R. Laithwaite. The goodness of a machine , 1965 .

[8] S. Hoole. Computer-aided analysis and design of electromagnetic devices , 1989 .

[9] S.R.H. Hoole,et al. Unconstrained optimization of coupled magneto-thermal problems , 1995 .

[10] Cyril Fonlupt,et al. Genetic programming on graphics processing units , 2009, Genetic Programming and Evolvable Machines.

[11] David R. Kincaid,et al. Algorithm 586: ITPACK 2C: A FORTRAN Package for Solving Large Sparse Linear Systems by Adaptive Accelerated Iterative Methods , 1982, TOMS.