Time Dependent Quantum Reactive Scattering on GPU
The computational core of the time dependent (TD) wavepacket program RWAVEPR has been implemented on a NVIDIA GPU of the GTX class. The TD program is a quantum wavepacket code that integrates the time-dependent Schrödinger equation for the generic atom-diatom reaction. In particular, the work has focused on the propagation procedure of the program, represented by the miham and lowpass routines, by implementing a fine grain model of parallelism on the GPU. Various features of the NVIDIA GPU have been exploited and different models of parallelism have been implemented and tested. Elapsed times and speed-ups for an atom-diatom chemical reaction have been calculated on the GPU and compared with the related CPU ones.
KeywordsElapse Time Single Instruction Multiple Data Computational Core High Level Programming Language Active Thread
Unable to display preview. Download preview PDF.
- 1.Kirk, D.B., Hwu, W.W.: Programming massively parallel processors. Morgan Kaufmann, San Francisco (2010)Google Scholar
- 2.CUDA ZONE website, http://www.nvidia.it/object/cuda_home_new_it.html (last access 06/04/2011)
- 3.Gervasi, O., Manuali, C., Laganà, A., Costantini, A.: On the restructuring of a molecular simulator as a Grid service in Chemistry and Material Science Applications on Grid Infrastructure. In: ICTP. Lecture Notes, vol. 24, pp. 63–81 (2009)Google Scholar
- 10.Balint-Kurti, G.G.: Time dependent quantum approaches to chemical reactivity. In: Laganà, A., Riganelli, A. (eds.), vol. 75, pp. 74–87. Springer, Heidelberg (2000)Google Scholar
- 11.COMPCHEM website, http://compchem.unipg.it (last access 06/04/2011)
- 12.EGI website, http://uf2011.egi.eu/ (last access 06/04/2011)
- 14.NVIDIA BLAS Library, http://developer.download.nvidia.com/.../CUBLAS_Library_2.0.pdf (last access 06/04/2011)
- 15.NVIDIA FFT Library, http://developer.download.nvidia.com/.../CUFFT_Library_1.1.pdf (last access 06/04/2011)
- 16.FFTW website, http://www.fftw.org (last access 06/04/2011)