HIGH PERFORMANCE COMPUTATION OF INCOMPRESSIBLE FLOW BY LATTICE BOLTZMANN METHOD ON MULTI-NODE GPU CLUSTER

摘要

GPGPU has drawn much attention on accelerating non-graphic applications. The simulation by D3Q19 model of Lattice Boltzmann method was executed successfully on multi-node GPU cluster by using CUDA programming and MPI library. Comparison and analysis were made among the parallel results by ID, 2D and 3D domain decompositions. With 384 x 384 x 384 mesh system and 96 GPUs, the performance by 3D decomposition is about 3~4 times higher than that of ID decomposition. In order to hide the communication time, we introduced the overlapping technique between computation and communication. Using 8-96 GPUs, the performances increase by a factor about 1.1～1.3 with overlapping mode. As large-scaled computation of a flow around a sphere at Re=13000 was carried on successfully using mesh system 2000 x 1000 x 1000 and 100 GPUs. As a result, 6.0 hours were used for processing 100,000 time steps. Under this condition, the computational time (2.79 hours) and data communication time (3.06 hours) are almost same.