A comparative analysis between FPGA and GPU for solving large numbers of linear equations
- 1 December 2017
- conference paper
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE) in 2017 29th International Conference on Microelectronics (ICM)
Abstract
Electromagnetic (EM) simulation is currently a highly-needed planning tool in high frequency systems. The main objective of EM simulation is to find an approximate solution for Maxwell's equations that satisfies the given boundary conditions and a set of initial conditions. The most famous methods for solving EM problems are finite element method (FEM). Computations involved in FEM consume too much time. In this paper, the performance improvement from using hardware emulation (FPGAs) and GPUs to solve large numbers of linear equations is evaluated. The results show that GPUs have superior performance over FPGAs.Keywords
This publication has 3 references indexed in Scilit:
- High-Performance Architecture for the Conjugate Gradient Solver on FPGAsIEEE Transactions on Circuits and Systems II: Express Briefs, 2013
- 3D parallel conjugate gradient solver optimized for GPUsPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2010
- Finite Volume Methods for Hyperbolic ProblemsPublished by Cambridge University Press (CUP) ,2002