作者: Luigi Genovese , Matthieu Ospici , Thierry Deutsch , Jean-François Méhaut , Alexey Neelov
DOI: 10.1063/1.3166140
关键词: Central processing unit 、 Wavelet 、 Computational science 、 Software 、 Double-precision floating-point format 、 CUDA 、 Code (cryptography) 、 Convergence (routing) 、 Physics 、 Speedup
摘要: We present the implementation of a full electronic structure calculation code on hybrid parallel architecture with graphic processing units (GPUs). This is performed free software based Daubechies wavelets. Such shows very good performances, systematic convergence properties, and an excellent efficiency computers. Our GPU-based acceleration fully preserves all these properties. In particular, able to run many cores which may or not have GPU associated, thus massive machines. With double precision calculations, we achieve considerable speedup, between factor 20 for some operations 6 whole density functional theory code.