Unstructured computational aerodynamics on many integrated core architecture
详细信息 查看全文
- 作者：Mohammed A. Al Farhan ; a ; mohammed.farhan@kaust.edu.sa" class="auth_mail" title="E-mail the corresponding author ; Dinesh K. Kaushikb ; dkaushik@qf.org.qa" class="auth_mail" title="E-mail the corresponding author ; David E. Keyesa ; david.keyes@kaust.edu.sa" class="auth_mail" title="E-mail the corresponding author
- 关键词：Intel Xeon Phi coprocessor ; Native mode ; Offload mode ; PETSc-FUN3D ; Thread affinity ; Unstructured meshes
- 刊名：Parallel Computing
- 出版年：2016
- 出版时间：November 2016
- 年：2016
- 卷：59
- 期：Complete
- 页码：97-118
- 全文大小：2799 K
文摘
- •
Porting implicit unstructured mesh code onto Intel Xeon Phi “Knights Corner”.
- •
Employing aggressive optimizations to improve the CFD flux kernel on Phi hardware.
- •
Exploring different thread affinity modes with different programming paradigms.
- •
Achieving 3.8x speedup using the offload mode compare to the baseline.
- •
Achieving 5x speedup using the native mode compare to the baseline.