Performance Analysis of Different Parallelization Frameworks on Heat Flow Application

This repo contains a class project for the course EC526 - Parallel Algorithms for High Performance Computing by Dr. Richard Brower, Spring 2021 at Boston University.

Potential Parallelization Frameworks to be tried are:

• OpenACC
• C++ <threads>
• C++ <aysnc>
• C++ <packaged_task>
• OpenMP
• MPI
• CUDA ??
• GoLang
• OpenCL

The serial implementation the Heat Flow Application will be written and then set as a baseline for comparison with accelerated versions of the same code using some of the frameworks listed above.

Preliminary Investigation

Todo: Run some tests to see performance of library, effect of different threads running at the same time.

Todo: See how different scheduling variations of OpenMP (dynamic, static, chunked) may affect a typical parallelized application, or maybe just ours.

Things to look into:

• Find a code / write a code for the application.
• Verify that the code works as expected -> Discuss with Brower.
• Write/Find + Test the python (or something else) visualization for the output.