GPU-Based Flow Simulation with Complex Boundaries
We have improved the implementation on the floating point pixel pipeline on the newest generation of GPU, using OpenGL and Cg, with many new optimizations and accurate boundary condition treatment. Preliminary results show that the speedup is about 10X on a Nvidia GeForceFX 5900 Ultra vs. a Pentium IV 2.53GHz CPU.
Here are two video clips recorded in real time.
Wei Li, Zhe Fan, Xiaoming Wei, and Arie Kaufman, Simulation with Complex Boundaries. “GPU Gems II”, edited by Matt Pharr(Nvidia) and published by Addison Wesley, Chapter 47. A previous version: Technical Report 031105, Computer Science Department, SUNY at Stony Brook, Nov 2003.(pdf).
Wei Li, Xiaoming Wei, and Arie Kaufman, Implementing Lattice Boltzmann Computation on Graphics Hardware. The Visual Computer, vol. 19, no.7-8, pp. 444-456, 2003. (pdf)