Skip to content

README.md

Latest commit

 

History

History
59 lines (40 loc) · 1.66 KB

README.md

File metadata and controls

59 lines (40 loc) · 1.66 KB

CPU OpenMP Version

This folder contains the CPU-only implementation using OpenMP for multi-threading parallelization.

Features

  • OpenMP Parallelization: Multi-threaded execution with configurable thread count
  • SIMD Vectorization: Compiler optimizations for enhanced performance
  • Standard C++17: Works on any platform with g++ or clang++
  • Memory Safe: Uses std::vector for automatic memory management
  • Configurable Threading: Thread count specified in parameter.txt (11th parameter)

Requirements

  • g++-14 (or any modern C++ compiler)
  • OpenMP support
  • macOS, Linux, or Windows (with appropriate compiler)

Building

make clean
make

Running

./abp_3D_confine.out

Parameters

Edit parameter.txt to change simulation parameters (11 tab-separated values):

epsilon  delta  Particles  Dt  De  vs  Wall  height  N  output_interval  N_thread
0.01     1e-4   200        10.1 0.0 0.0 15.0  15.0    1000 100             6

The N_thread parameter controls the number of OpenMP threads used for parallel execution.

Performance

Particles 1 Thread 6 Threads Speedup
100 0.026s 0.103s 0.25×
200 ~0.5s ~0.2s ~2.5×
1000 ~16s ~6s ~2.7×
5000 34.4s 14.3s 2.4×

Benchmarks on Apple M2, 1000 timesteps

Scaling: O(N²) due to pairwise particle interactions

Note: For N < 100, single-threaded execution is faster due to OpenMP overhead.

Output

Results are saved to data/simulation.csv with particle positions and orientations at each output interval.