Parallel Computing

National Tsing Hua University - CS 542200 Course's Instructor: Jerry Chou

This repo includes 4(+1) homeworks, which all cover on parallel computing techniques, such as MPI, OpenMP, Pthread, and CUDA.

The specific details of each homeworks (Homework Spec, My Report), please refer to each homework directory and find the PDF file prefixed with PP_2020 & hw.

Homework List:

1. HW1: [MPI] Odd-Even Sort

• If you wants to enhance the performance, please rewrite the logic by C++ and call the C++ native libraries(STL, ...), you will get greater performance than C.

2. HW2: [MPI+OpenMP+Vectorization] Mandelbrot Set

• My code is ranking top 3.
• Vectorization is Intel-based speedup method.

3. HW3: [Pthread] All-Pairs Shortest Path

• There're two way get more efficient code than my code:
  1. Implement Blocked Floyd-Warshall algorithm, and use Tenary Expression to implement logics. (You can check ./hw4-1/seq.cc to see how to implement Blocked Floyd-Warshall algorithm on CPU.)
  2. Implement Floyd-Warshall algorithm, and use Vectorization to implement logics.

4. HW4: [Cuda] All-Pairs Shortest Path (One GPU or Multiple GPUs)

• My code is ranking top 3.

There are also 5 labs for your reference.