CUDA 第三方库 cula 应用示例
CULA 库， 说白了就是LAPACK库的cuda版， 实现了大部分LAPACK的函数功能， 而且函数命令以及参数都极其类似LAPACK， 不了解或者不知道LAPACK的， wiki 自己去wiki吧！
目前该库， 有free版可以下载， 此版只能实现6个函数， 而且是单精度计算，如下
|
Type
|
Description
|
Real
|
Complex
|
|---|---|---|---|
| General | Solves a general system of linear equations AX=B. | SGESV | CGESV |
|
|
Computes an LU factorization of a general matrix, using partial pivoting with row interchanges. | SGETRF | CGETRF |
| Computes a QR factorization of a general rectangular matrix. | SGEQRF | CGEQRF | |
| Computes the least squares solution to an over-determined system of linear equations, AX=B, ATX=B, or AHX=B, or the minimum norm solution of an under-determined system, where A is a general rectangular matrix of full rank, using a QR or LQ factorization. | SGELS | CGELS | |
| Solves the LSE (Constrained Linear Least Squares Problem) using the GRQ (Generalized RQ) factorization. | SGGLSE | CGGLSE | |
| Computes the singular value decomposition (SVD) of a general rectangular matrix. | SGESVD | CGESVD |
以下代码示例怎么使用其中一个函数 SGESVD， 矩阵奇异值分解， 应用广泛， 该分解旨在分解任意一个矩阵A大小mxn成， 三个矩阵的积， A=U*S*V， 其中U大小 mxm， S 是对角矩阵对角线上的特征值按照从大到小排列， V大小nxn.
CULA库的这个函数， 对matlab 相应的SVD函数加速比已经超过20倍， 让我们还是来看看怎么用吧！
头文件：
#include
#include
#include
#include
#include
#include
#include //cula 库的头文件
#include
#include
#include
/* Setup SVD Parameters */
int LDA;
int LDU;
int LDVT;
int i,j;
int* dev=NULL;
float* A = NULL;
float* S = NULL;
float* U = NULL;
float* VT = NULL;
char jobu = 'A';
char jobvt = 'A';
/* 参数初始化*/
/*LDA LDU LDVT 都是用来指定矩阵同行两个相邻元素之间的物理存储距离，由于cula lapack 都是按照列存储矩阵所以此处的LDA LDVT 是矩阵中一列中包含的元素个数， 也就是行数.*/
LDA = m; //m矩阵A的行数 n矩阵A的列数
LDU = m;
LDVT = n;
A = (float*)malloc(m*n*sizeof(float));
S = (float*)malloc(imin(m,n)*sizeof(float));
U = (float*)malloc(LDU*m*sizeof(float));
VT = (float*)malloc(LDVT*n*sizeof(float));
/*初始化cula库*/
/*culaSgesvd函数参数描述*/
/*Parameters
• jobu
– Type: char
– Direction: Input
Specifies options for computing all or part of the matrix U:
= ‘A’: all M columns of U are returned in array U:
= ‘S’: the first min(m,n) columns of U (the left singular vectors) are returned in the array U;
= ‘O’: the first min(m,n) columns of U (the left singular vectors) are overwritten on the array A;
= ‘N’: no columns of U (no left singular vectors) are computed.
• jobvt
– Type: char
– Direction: Input
Specifies options for computing all or part of the matrix VT :
= ‘A’: all N rows of VT are returned in the array VT;
= ‘S’: the first min(m,n) rows of VT (the right singular vectors) are returned in the array VT;
= ‘O’: the first min(m,n) rows of VT (the right singular vectors) are overwritten on the array A;
= ‘N’: no rows of VT (no right singular vectors) are computed.
JOBVT and JOBU cannot both be ‘O’.
• m
– Type: int
– Direction: Input
The number of rows of the input matrix A. M >= 0.
• n
– Type: int
– Direction: Input
The number of columns of the input matrix A. N >= 0.
• a
– Type: S/D/C/Z Pointer
– Direction: Input/Output
– Dimension: (LDA,N)
On entry, the M-by-N matrix A.
On exit,
if JOBU = ‘O’, A is overwritten with the first min(m,n) columns of U (the left singular vectors, stored
columnwise);
if JOBVT = ‘O’, A is overwritten with the first min(m,n) rows of VT (the right singular vectors,
stored rowwise);
if JOBU != ‘O’ and JOBVT != ‘O’, the contents of A are destroyed.
• lda
– Type: int
– Direction: Input
The leading dimension of the array A. LDA >= max(1,M).
• s
– Type: S/D Pointer
– Direction: Output
– Dimension: (min(M,N))
The singular values of A, sorted so that S(i) >= S(i+1).
• u
– Type: S/D/C/Z Pointer
– Direction: Output
– Dimension: (LDU,UCOL)
(LDU,M) if JOBU = ‘A’ or (LDU,min(M,N)) if JOBU = ‘S’. If JOBU = ‘A’, U contains the M-by-M
orthogonal/unitary matrix U; if JOBU = ‘S’, U contains the first min(m,n) columns of U (the left
singular vectors, stored columnwise); if JOBU = ‘N’ or ‘O’, U is not referenced.
• ldu
– Type: int
– Direction: Input
The leading dimension of the array U. LDU >= 1; if JOBU = ‘S’ or ‘A’, LDU >= M.
• vt
– Type: S/D/C/Z Pointer
– Direction: Output
– Dimension: (LDVT,N)
If JOBVT = ‘A’, VT contains the N-by-N orthogonal/unitary matrix VT ; if JOBVT = ‘S’, VT contains
the first min(m,n) rows of VT (the right singular vectors, stored rowwise); if JOBVT = ‘N’ or ‘O’,
VT is not referenced.
• ldvt
– Type: int
– Direction: Input
The leading dimension of the array VT. LDVT >= 1; if JOBVT = ‘A’, LDVT >= N; if JOBVT = ‘S’,
LDVT >= min(M,N)
*/
status=culaSelectDevice(2); // 选择运行cula库的GPU
status = culaInitialize(); //初始化
status = culaSgesvd(jobu, jobvt, m, n, A, LDA, S, U, LDU, VT, LDVT); //开始计算
本示例， 仅仅示范cula库的一个函数应用， 有需要源码和示例程序的， 上cula官方网站下载免费版， 里面有个小的sdk， 可以参考下http://www.culatools.com
来自 “ ITPUB博客 ” ，链接：http://blog.itpub.net/20259129/viewspace-662452/，如需转载，请注明出处，否则将追究法律责任。
转载于:http://blog.itpub.net/20259129/viewspace-662452/