Parallel Programming with OpenMP and MPI

This project demonstrates two approaches to parallel programming: OpenMP for shared memory systems and MPI for distributed memory systems. The implementations focus on efficient parallel computations for prefix sum and target searching within an array.

---

Features

OpenMP Implementation

• Function: prefixSumOMP
• Performs prefix sum computation on a shared memory system.
• Key highlights:
  • Thread-Level Parallelism: Splits the array into chunks, assigning each to a thread.
  • Synchronization: Utilizes barriers to synchronize thread execution.
  • SIMD Optimization: Applies SIMD directives to enhance performance during adjustment phases.

MPI Implementation

• Functionality: Distributed search for a target element in an array.
• Key highlights:
  • Data Distribution: Uses MPI_Scatter to distribute data among processes.
  • Broadcasting: Uses MPI_Bcast to share the target across all processes.
  • Reduction Operation: Uses MPI_Allreduce to find the global result across ranks.
  • Timing: Measures and reports execution time for performance analysis.

---

Setup

Prerequisites

• Compiler: Requires a C++ compiler with OpenMP and MPI support (e.g., GCC or Intel C++ Compiler).
• Libraries: MPI library (e.g., MPICH or OpenMPI).

Compilation

• For OpenMP:

  cd openmp/
  g++ -fopenmp -o ompprefixsum main.cpp

• For MPI:

  cd mpi/
  mpicxx -o mpi_search main.cpp

Usage

• For OpenMP:

  ./ompprefixsum

• For MPI:

  mpiexec -np <num_processes> ./mpi_search

---