Description: OpenACC is a high-level compiler directive-based (using pragmas in C/C++) programming approach for parallel computing on GPUs. Compared to other low-level GPU programming models such as CUDA and OpenCL, porting of legacy code with OpenACC does not require significant structure changes, which allows considerable simplification and productivity improvement for heterogenous HPC applications.

The objectives of this training are to provide an introductory overview of the OpenACC programming model and assist HPC users in gaining experience on OpenACC as a “multicore/GPU programming model” using hands-on exercises with incremental changes of sample codes and profiling comparisons using OpenACC directives, data management and loop optimization.

Presenter(s): Dr. Shaolin Mao, GDIT / PET Location: Webcast Date & Time: February 29, 2024, 2:00p - 3:30p ET

Controlled by: DoD HPCMP Controlled by: PET Program CUI Category: OPSEC Limited Dissemination Control: FEDCON POC: Mr. Ronald Hedgepeth, pet@hpc.mil

---

CUI