Wavefront Computation: Wavefront Computation is a parallel application implemented in C++ that performs dependency-based computations following a wavefront pattern. This project leverages a hybrid parallel programming model:

Shared Memory Parallelism using FastFlow Distributed Memory Parallelism using MPI It was developed as part of my exam project in the SPM (Parallel and Distributed Paradigm) course.

Features: Hybrid Parallelism: Combines shared memory (via FastFlow) and distributed memory (via MPI) techniques. Wavefront Pattern: Computes data along diagonals (wavefronts) where each computation depends on the previous one—ideal for dynamic programming, simulations, or matrix-based computations. Scalability: Designed to scale across multi-core machines and distributed clusters. Performance Optimization: Efficient scheduling and communication strategies to optimize both intra-node and inter-node performance.

Requirements: C++ Compiler: Supporting C++20 FastFlow Library: For shared memory parallelism (FastFlow GitHub) MPI Library: Such as OpenMPI or MPICH CMake: For building the project (optional but recommended)

Install Dependencies: Download and install FastFlow, or include its headers in the project Fastflow Library Github: https://github.com/fastflow/fastflow/tree/master/ff Ensure an MPI library (e.g., OpenMPI) is installed and available (mpicc/mpic++).

Implementation Details Wavefront Algorithm: The computation follows a wavefront pattern where each element is computed after its dependencies (typically the elements from the previous diagonal) are ready. Hybrid Approach: The program dynamically selects between MPI and FastFlow depending on whether it is running on a multi-node cluster or a single multi-core system. Synchronization: MPI is responsible for synchronizing boundaries between nodes, while FastFlow handles parallelism within a node.

Acknowledgments FastFlow Library: For providing a robust shared memory parallel framework. MPI Community: For the scalable distributed computing tools. Developed as part of the SPM Parallel and Distributed Paradigm exam.

Skills: Parallel Programming (FastFlow & MPI) · Distributed Systems · High Performance Computing (HPC) · Performance Optimization · C/C++ Programming.