该博客介绍了如何使用CUDA进行任意维度的矩阵相乘操作。通过设置合适的维度大小（wA, wB, hA），利用dim3结构定义线程网格和线程块，并调用matrixMul核函数在GPU上执行矩阵乘法。"
102824175,4755197,Excel列字母转换：从A到ZZZ的base26实现,"['编程算法', 'Excel', 'base26转换', '递归算法', '循环结构']
__global__ void matrixMul( float* A, float* B, float* C, int hA,int wA, int wB)
{
int bx = blockIdx.x;
int by = blockIdx.y;
int tx = threadIdx.x;
int ty = threadIdx.y;
float Csub = 0.0f;
for (int j=0;j<wA;j+=BLOCK_DIM)
{
__shared__ float AS[BLOCK_DIM][BLOCK_DIM];
__shared__ float BS[BLOCK_DIM][BLOCK_DIM];
if(((by*BLOCK_DIM+ty)<hA)&&((tx+j)<wA))
{
int bx = blockIdx.x;
int by = blockIdx.y;
int tx = threadIdx.x;
int ty = threadIdx.y;
float Csub = 0.0f;
for (int j=0;j<wA;j+=BLOCK_DIM)
{
__shared__ float AS[BLOCK_DIM][BLOCK_DIM];
__shared__ float BS[BLOCK_DIM][BLOCK_DIM];
if(((by*BLOCK_DIM+ty)<hA)&&((tx+j)<wA))
最低0.47元/天 解锁文章
扫一扫
459
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
