当我读到《
Directable, High-Resolution Simulation of Fire on the GPU》这篇论文的时候,它说它用一个基于GPU的雅个比迭代器求解速度场的散度场。于是我就用CUBLAS自己做了一个。下面就是GPU小波变换的实现了,比较头大,要学的还有很多。
When I read the
"Directable, High-Resolution Simulation of Fire on the GPU", it said that it used a GPU-based Jacobi iterator to solve the non-divergence equation for velocity field, so I made one based on CUBLAS. I think it should be much faster when the linear system is quite large. The next problem is GPU wavelet transform, there are still lot's of things to learn.
转载于:https://www.cnblogs.com/Jedimaster/archive/2009/09/14/1566454.html