利用CUDA实现了遥感数字影像处理基本算法中的拉普拉斯算子，并分别实现了采用Globalmemory、Sharedmemory，Texturememory进行数据传输，还同CPU中的串行代码进行了时间比较和误差分析。为了方便在服务器环境下测试运行，我们去掉了在Windows环境下对第三方库遥感影像库GDAL的依赖，而将遥感影像数据转成了二进制图像格式，对于输出影像也以同样的二进制格式存储。图像效果图通过专业遥感软件PCI或通用图像处理软件PhotoshopCS4打开。最后我们对计算结果进行了误差分析和加速比分析。
CUDA enables the use of remote sensing digital image processing the basic algorithm in the Laplace operator, and were achieved using Globalmemory, Sharedmemory, Texturememory data transmission, but also with the serial code CPU time comparison and error analysis. In order to facilitate the test run in the server environment, we have removed in the Windows environment for third-party remote-sensing image library GDAL library dependencies, but will switch to remote sensing image data into a binary image format, for the output images are also stored in the same binary format . Image Effect Picture through professional remote-sensing software, PCI or universal image processing software PhotoshopCS4 open. Finally, we calculated the results of the error analysis and speedup analysis. (2010-02-04, Unix_Linux, 3KB, 下载5次)
pi的并行计算，以此例可以知道并行计算的基本结构，程序是用fortran加mpi编写的
pi of parallel computing, this Example can know the basic structure of parallel computing, the program is prepared fortran plus mpi (2008-05-27, Unix_Linux, 1KB, 下载5次)