CSC 5001– High Performance Systems

The objective of the project is to design a parallel application on a target architecture (machine clusters, multicore machines, heterogeneous heterogeneous machines integrating CPU and GPU) starting from its sequential sequential version using the classical tools of parallelism parallelism (in this case OpenMP, MPI, CUDA).

You can choose between two typical HPC applications:

The project is to be carried out in pairs. Beyond the gain in performance, the objective of the project is that you follow a scientific approach that we ask you to explain during the defense (description of the parallelism, choice of the target platform, performance analysis, etc).

You are asked to return the source code of your parallelized programs, as well as the slides used during the defense.

During the defense, you will be asked to present how you parallelized the application. You must also also present and explain the measured performances.