LBM GPU Accelerator

CUDA C++ GPU

A high-performance implementation of the Lattice Boltzmann Method (D3Q19 scheme) using CUDA. The solver achieves real-time visualization of vortex shedding and turbulent flows.

LBM Evolution Equation

The collision step is local, making it ideal for the massive parallelism of GPUs. Memory coalescence is optimized by structure-of-arrays (SoA) data layout.

Visualizations


← Back to Projects