Accelerating lattice Boltzmann method using C++ standard parallelization on GPU

This presentation will address the technical challenges encountered when applying parallel algorithms from the C++ standard library to solve calculations in immersed-boundary lattice Boltzmann method problems. The proposed solutions involve the utilization of GPU buffers to mitigate communication costs and the implementation of GPU-GPU direct communication. Additionally, the presentation will delve into the performance evaluation of each of these approaches.