OpenMP has been successfully implemented in a FORTRAN program for hydrodynamic analysis. Background of the numerical codes will be introduced at the beginning of the talk. In the following slides, the original sequential codes will be analyzed. Details of OpenMP implementation for the specific codes will be explained in the main part of presentation. Emphasis will be put on how to formulate the matrix equation in parallel and solve it using Intel Math Kernel Library (Intel MKL). The key issue of data race will be also highlighted. Some tests for the modified codes have been carried out on HPC. Computational time for small and large scale problems has been considered. Finally, some concluding remarks will be given based on the tests.
|Publication status||Published - 4 Jun 2013|
|Event||HPC symposium - Bath, UK United Kingdom|
Duration: 4 Jun 2013 → 4 Jun 2013
|Country||UK United Kingdom|
|Period||4/06/13 → 4/06/13|
- Intel MKL
- hydrodynamic analysis