搜索资源列表
18-1
- 矩阵转置并行计算,需要至少4个处理器,采用块棋盘划分方法
动态规划----矩阵连乘问题
- 动态规划----矩阵连乘问题 动态规划法是解决问题的一种方法。它不规定为了得到结果需如何将问题划分为子问题的固定方法,而是按不同输入给出问题的具体实例的子问题划分方法,然后再进行运算、解答问题。 矩阵连乘问题的主要思想如下: 1)设置大小为连乘个数的方阵 2)主对角线上方各元素Di,j(i<j)表示矩阵Mi连乘到Mj的最小工作量 3)下方元素Di,j(i>j)记录获得该最小工作量矩阵分组的第一组的最后一个矩阵的序列号 最后通过下方元素可知最终结果的分组方式。-dynamic prog
matri
- 行列划分并行矩阵乘积算法,在cluster上运行成功-matrix multiplition on cluster
transpose
- 矩阵转置,使用棋盘划分将矩阵通过划分为不同的小块,使用MPI讲其并行求小块的转置。-Matrix transpose, the matrix by using the board is divided into different small pieces, about its use of MPI parallel transpose small demand.
MPI_Bcast
- 矩阵相乘的LU算法的MPI并行实现源码,采用连续行划分方法。-MPI parallel implementation of matrix multiplication, LU algorithm source code, a continuous line of division method.
matrixAssign
- 最简单的CUDA实现的矩阵乘法,带状划分,没有进行优化的!-CUDA matrix multiplication easiest to achieve, striping, not optimized for!
bxMatrix_RowC
- 大数量级的矩阵乘法,按行划分,并行算法设计-Matrix multiplication, division by line, parallel algorithm design
bxMatrixMulti_new
- 并行算法中的矩阵乘法设计,按行列划分,高效求解矩阵运算-Parallel matrix multiplication algorithm design, according to the ranks of the division, and efficient for solving matrix operations
matrixTranspos
- 一个利用pthread进行基于直角划分的矩阵转置源码,内含功能源码,测试shell scr ipt,readme-A pthread-based matrix transpose source code implemented by using Rectangular partitioning, including source code, shell scr ipt for testing and Readme
矩阵向量
- 带状划分的矩阵向量乘法。数据文件是40*40的算例。(The file is prepared for solving A*b, in the Date there is an example for A which is a 40*40 matrix)
code
- 平行计算课程,并行程序转置矩阵,使用3种方法:线性转置,块棋盘划分,直角划分法(Parallel computing course, parallel program transpose matrix, using three methods: linear transpose, block chessboard division, right angle division)