Author:
Chen Yuqiao,Liang Jianhan,Ding Meng,Zhang Lin,Guan Qingdi,Wang Xinxin
Abstract
Abstract
To develop high performance computing methods for compressible flow calculation, a GPU-accelerated compressible flow solver is developed with Compute Unified Device Architecture (CUDA). The WENO5 scheme is adopted for spatial discretization, and the third-order Runge-Kutta scheme is used for time discretization. According to the algorithm and programming model, the heterogeneous computing method of the solver is designed. Different kernels are designed to implement different computing functions, and shared memory is used for time-advanced computations. The solver is verified by the one-dimensional shock tube case, and a good acceleration effect is obtained with the increase of the grid size. And the impact of execution configuration on kernel performance was investigated. When the block size is reduced under different grid sizes, the speedup changes in the same way, but the performance parameters change differently.
Subject
General Physics and Astronomy
Reference8 articles.
1. An mpi-cuda approach for hypersonic flows with detailed state-to-state air kinetics using a gpu cluster;Bonelli;Computer Physics Communications,2017
2. Accelerating cardiac bidomain simulations using graphics processing units;Neic;IEEE Transactions on Biomedical Engineering,2012
3. Hybrid core acceleration of UWB SIRE radar signal processing;Song;IEEE Transactions on Parallel & Distributed Systems,2010
4. Acceleration of a two-dimensional Euler flow solver using commodity graphics hardware;Brandvik;Proceedings of the Institution of Mechanical Engineers, Part C: Journal of Mechanical Engineering Science,2007