Beams Document 3902-v1

DocDB Home ]  [ Search ] [ Last 20 Days ] [ List Authors ] [ List Topics ] [ List Events ]

Scientific Computing on Graphics Processor Units: An Application in Time-Domain Electromagnetic Simulations

Document #:
Document type:
Submitted by:
Camille M. Ginsburg
Updated by:
Camille M. Ginsburg
Document Created:
13 Jul 2011, 09:29
Contents Revised:
13 Jul 2011, 09:29
Metadata Revised:
13 Jul 2011, 09:29
Viewable by:
  • Public document
Modifiable by:

Quick Links:
Latest Version

Recent developments in the design of graphic processing units (GPU's) have been occurring at a much greater pace than with central processor units (CPU's). The computation power due to the parallelism provided by the graphics cards got the attention of communities dealing with high performance scientific computing. The computational electromagnetics community as well has started to utilize the computational power of graphics cards for computing and, in particular, several implementations of finite-difference time-domain (FDTD) method have been developed. FDTD has been used to solve numerous types of problems arising while studying many applications, including the following applications: scattering, radar cross-section, microwave circuits, waveguides, fiber optics, antennas (radiation, impedance), propagation, medical applications, shielding, coupling, electromagnetic compatibility (EMC), electromagnetic pulse (EMP) protection, nonlinear and other special materials, geological applications, inverse scattering, and plasma. In this talk the basics of the FDTD method will be introduced, and an implementation of FDTD based on Compute Unified Device Architecture (CUDA) development environment from NVIDIA will be presented.
Files in Document:
  • powerpoint file (VEYSEL_DEMIR_PRESENTATION_FERMILAB.ppt, 9.1 MB)
DocDB Home ]  [ Search ] [ Last 20 Days ] [ List Authors ] [ List Topics ] [ List Events ]

DocDB Version 8.8.9, contact Beams Document Database Administrators