CUDA based parallel Monte-Carlo computing methods for heat radiation calculations with a large quantity of bundles
Abstract：The use of NVIDIA＇s CUDA parallel computing graphic processing unit（GPU） architecture was studied.Flame radiation from a pool fire was used as the computational example.Three GPU＇s having different capabilities（v.1.1,1.2,and 2.1） were used.A Monte-Carlo heat radiation calculation of the fire was performed and the computation times determined.If several blocks,the number being an integral multiple of the processor number,were used with the maximum allowed number of threads and a high capability GPU the computation speed is greater.For a five million beams calculation the timing for the 1.1,1.2,and 2.1 GPU was 0.44,0.167,and 0.084 seconds,respectively.This is 62,167,and 333 times faster than the central processing unit.It is concluded that GPU based,CUDA parallel computing significantly increases the computational efficiency of the Monte-Carlo method while still providing equal accuracy of the results.
Keywords：Monte-CarloCUDAheat radiation
Publication Date：2011-01-01
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)