GPU implementation and optimization of electromagnetic simulation using the FDTD method for antenna designing

Abstract

This paper describes electromagnetical field simulation using the 3D-FDTD method for antenna designing on a CUDAcompatible GPU. We use the Split Perfectly Matched Layer as an absorbing boundary condition. As is well known, the 3D-FDTD method is a kind of stencil computation and is considered better at GPU implementation. In order to find the best blocking size for the target GPU architecture, we empirically explore a design space of blocking size. We also propose a kernel fusing method as one of the efficient optimization methods, which improves the total performance about 10 % at the cost of a small increase in memory usage. As a result of evaluation, our implementation of the 3D-FDTD method on a GeForce GTX295 platform achieves about 130 times performance improvement compared to a simple CPU implementation, which is expected to be faster than an ideally parallelized CPU implementation using multicore and SIMD instructions.

Implementation of a GPU-Oriented Absorbing Boundary Condition for 3D-FDTD Electromagnetic Simulation

“…Many work including ours have reported the block size has a strong effect on the performance [7]. Although the best size may depend on the presence or absence of each optimization technique, we used the block size of (B x , B y , B z ) = (32, 4, 32) based our primary evaluation results.…”

“…The FDTD method is known as a kind of stencil computation that has a high degree of parallelism but requires a large memory bandwidth. While GPU implementation is attractive as a cost-effective acceleration approach, our earlier work has shown that GPU implementation of Absorbing Boundary Conditions (ABCs) tends to be a bottleneck of the simulation [7]. We discuss this issue in this paper.…”