Parallel LU-SGS method implementation for gas dynamics problems on GPU-accelerated computing systems |
| 1 | |
| 2013 |
| INFORMATION TECHNOLOGIES |
| scientific article | 519.612 | ||
| 213-218 | parallel algorithms, numerical methods in gas dynamics, LU-SGS, high performance computing, GP-GPU. |
| An effective algorithm for parallel computing gasdynamic flows using GPU-accelerated computing systems is proposed. The algorithm is based on an implicit scheme that leads to large-size linear systems with sparse matrices solved by the LU-SGS (Lower-Upper Symmetric Gauss-Seidel) approximate factorization method. The parallel algorithm duplicates exactly the work of a successive one and possesses a high level of scalability. |
 |