Academic High Performance Computing Projects

Description:

This repository contains two academic projects focusing on High Performance Computing (HPC), specifically utilizing MPI (Message Passing Interface) and OpenMP with intrinsic optimizations (AVX and NEON). And Both were implemented in C.

Project Details:

• Supervisor: Charles Bouillaguet
• Email: [email protected]
• Project Contributors: Yannick Zhang, Romain Capron
• University/Institution: Polytech Sorbonne, Université Pierre et Marie Curie (UPMC)

Project Structure:

This repository consists of two main repertory, each containing:

• Initial problem statements and sequential codes
• Parallel implementations using MPI or OpenMP and intrinsic optimizations (AVX and NEON)
• Python scripts for result analysis
• Detailed reports documenting the methodology and interpretation of results

Running the Codes:

To utilize the provided codes, follow these steps:

Prerequisites:

1. Clone the repository:

   git clone https://github.com/Owkly/HPC

2. Navigate to the desired project directory:

   cd <project_directory>

   Replace <project_directory> with either MPI or OMP_Intrinsic.

3. Compile the codes using :

   make

4. Optionally, clean the directory (remove executables and temporary files) using:

   make clean

Execute the desired program:

MPI:

• For the sequential version:

  ./heatsink_sequential

• For the parallel version:

  make run NUM=<number_of_processes> TARGET=<executable_name>

  Replace <number_of_processes> with the desired number of processes and <executable_name> with the name of the target executable, either heatsink_1D or heatsink_2D.

• For graphical representation:

  make graph MODE=<mode>

  Replace <mode> with either FAST, MEDIUM, or NORMAL. (Default mode is FAST)

  NOTE: If you want to change the mode, you have to modify the specific C file (in the #define on line 23) and recompile the code using the make command.

OpenMP and Intrinsic:

• For sequential execution:

  make run-seq

• For parallel version:

  make run TARGET=<executable_name> THREADS=<number_of_threads>

  Replace <number_of_threads> with the desired number of threads and <executable_name> with the name of the target executable (fft_openmp or fft_intrinsic_vec).