%0 Journal Article
%T FDTD Acceleration for Cylindrical Resonator Design Based on the Hybrid of Single and Double Precision Floating-Point Computation
%A Hasitha Muthumala Waidyasooriya
%A Masanori Hariyama
%A Yasuhiro Takei
%A Michitaka Kameyama
%J Journal of Computational Engineering
%D 2014
%R 10.1155/2014/634269
%X Acceleration of FDTD (finite-difference time-domain) is very important for the fields such as computational electromagnetic simulation. We consider the FDTD simulation model of cylindrical resonator design that requires double precision floating-point and cannot be done using single precision. Conventional FDTD acceleration methods have a common problem of memory-bandwidth limitation due to the large amount of parallel data access. To overcome this problem, we propose a hybrid of single and double precision floating-point computation method that reduces the data-transfer amount. We analyze the characteristics of the FDTD simulation to find out when we can use single precision instead of double precision. According to the experimental results, we achieved over 15 times of speed-up compared to the CPU single-core implementation and over 1.52 times of speed-up compared to the conventional GPU-based implementation. 1. Introduction Computational electromagnetic simulation shows a rapid development recently due to the introduction of processors that have parallel processing capability such as multicore CPUs and GPUs (graphic processing units). The FDTD (finite-difference time-domain) algorithm [1, 2] is one of the most popular methods of computational electromagnetic simulations due to its simplicity and very high computational efficiency. It has been widely used in many applications such as coil modeling [3] and resonance characteristics analysis of a cylindrical cavity [4, 5]. Many of these applications require double precision floating-point computation to satisfy the stability condition [6]. The FDTD simulation requires a large amount of data. When more processor cores are used in parallel, more data transfers occur between the memory and the processor cores. Therefore, the memory-bandwidth limitation is a major problem in the FDTD simulation using computers. To overcome this problem, we have to reduce the data transfer amount, so that we can use more cores in parallel. To do this, we propose a hybrid precision computation method that uses both single and double precision. Single precision data use 4 bytes compared to 8 bytes used in double precision data. Therefore, using single precision reduces the data amount and increases the processing speed. However, using single precision could bring inaccurate results. In some cases, the FDTD simulation does not converge when it is executed for a large number of iterations. In this paper, we consider the FDTD simulation of a cylindrical resonator [5]. It is one of the most fundamental types of resonant cavities
%U http://www.hindawi.com/journals/jcengi/2014/634269/