This repository contains the code for optimizing a numerical simulation of Computational Fluid Dynamics (CFD). The simulation is based on the Lattice Boltzmann Method (LBM), a relatively recent technique which consists in replacing the Navier-Stokes equations with a discretization of the Boltzmann equation in order to simulate the complex behavior of fluids using streaming and collision (relaxation) processes.
The code is written in C and uses OpenMP for parallelization. The code can be compiled using the following command:
make allThe code is optimized using the following techniques:
- OpenMP parallelization
- SIMD vectorization (AVX2)
- Memory alignment
- Cache blocking
- Software pipelining
The code is benchmarked using execution time as metric. The benchmarking is performed on a 2.2 GHz Intel Xeon E5-2698 v4 processor with 20 cores and 40 threads. The following table shows the execution time for different optimization techniques:
| Optimization | Execution time (s) |
|---|---|
| Baseline | 119.5 |
| +OpenMP parallelization | 26.9 |
| +SIMD vectorization | 18.6 |
| +Memory Alignment | 14.8 |
| +Cache Blocking | 10.2 |
| +Software Pipelining | 5.8 |
This repository uses the implementation by Mousany and Clarivy. However, commits are modified to correspond to authors of the optimization techniques. The baseline code can be found in toast lab.