This work presents a highly optimized computational framework for theDiscrete Dipole Approximation, a numerical method for calculating the opticalproperties associated with a target of arbitrary geometry that is widely usedin atmospheric, astrophysical and industrial simulations. Core optimizationsinclude the bit-fielding of integer data and iterative methods that complementa new Discrete Fourier Transform (DFT) kernel, which efficiently calculates thematrix vector products required by these iterative solution schemes. The newkernel performs the requisite 3-D DFTs as ensembles of 1-D transforms, and bydoing so, is able to reduce the number of constituent 1-D transforms by 60% andthe memory by over 80%. The optimizations also facilitate the use of paralleltechniques to further enhance the performance. Complete OpenMP-basedshared-memory and MPI-based distributed-memory implementations have beencreated to take full advantage of the various architectures. Several benchmarksof the new framework indicate extremely favorable performance and scalability.OpenDDA is available following the usual open source regulations fromhttp://www.opendda.org
展开▼
