High-Performance time-dependent Lattice Boltzmann Navier-Stokes flow on GPU !

Performed on a Laptop (Yes we can !). GPU : nVIDIA GTX 580M, Compute capability 2.1, Clock rate 1240 Mhz,384 CUDA cores, 8 multiprocessors, Warp size 32. Throughput 38 GB/s Device-to-Device, CUDA-Z (http://cuda-z.sourceforge.net) returns 630 GFlops in single-precision ! OS : Windows 7, MS Visual Studio 10 used, CUDA 4.1 Please visit : http://www.cmla.ens-cachan.fr