GPU Implementation of the DP code
Laboratoire des Solides Irradies, Ecole Polytechnique, CNRS, CEA, UMR 7642, 91128 Palaiseau cedex, France
PRACE project WP67, 2013
@article{sottilea2013gpu,
title={GPU Implementation of the DP code},
author={Sottilea, F and Roedla, C and Slavnicb, V and Jovanovicb, P and Stankovicb, D and Kestenerc, P and Houssenc, F},
year={2013}
}
Main goal of this PRACE project was to evaluate how GPUs could speed up the DP code – a linear response TDDFT code. Profiling analysis of the code has been done to identify computational bottlenecks to be delegated to the GPU. In order to speed up this code using GPUs, two different strategies have been developed: a local one and a global one. Both strategies have been implemented with cuBLAS and/or CUDA C. Results showed that one can reasonably expect about 10 times speed-up on the total execution time, depending on the structure of the input and the size of datasets used, and speed-ups up to 16 have been observed for some cases.
June 23, 2013 by hgpu
