C Image Processing SIMD

This repository contains a cleaned C project around image processing and performance-oriented optimization.

What it demonstrates

• BMP image loading and writing in C
• scalar image-processing implementations
• OpenMP parallelization
• ARM NEON SIMD-oriented code paths when compiled on a NEON-capable target
• comparison of baseline and optimized implementations
• low-level C programming for performance-oriented workloads

Repository structure

include/              Header files
src/                  C source files
.github/workflows/    Continuous-integration workflow
Makefile              Local build entry point
README.md             Project documentation

Current source files

• include/lib_bmp.h
• src/lib_bmp.c
• src/tp8_etudiants.c
• src/main_tp10.c

src/tp8_etudiants.c contains the optimized image-processing work, including OpenMP parallel sections and an ARM NEON implementation guarded by __ARM_NEON.

Build

Requirements:

• GCC or a compatible C compiler
• OpenMP support (-fopenmp)
• make

Build the default TP8 executable:

make

Equivalent explicit target:

make tp8

Build the TP10 executable:

make tp10

Remove generated binaries:

make clean

The Makefile currently produces:

• image_processing_tp8
• image_processing_tp10

The GitHub Actions workflow builds the default make target on Ubuntu, then cleans the generated binary. This validates the portable scalar/OpenMP build path. The ARM NEON path requires a compatible ARM target and is not exercised by the current x86 GitHub-hosted runner.

Platform note

The SIMD-specific code is conditionally compiled under __ARM_NEON. On non-ARM/NEON systems, the project remains buildable through the scalar/OpenMP path without requiring NEON compiler flags.

Scope

This repository focuses on the implementation and optimization work itself. Course PDFs, reports, binaries, assistant configuration files, and generated build artifacts are intentionally excluded.