基于CUDA C的程序，可以对任意矩阵进行矩阵乘法，目前仅用全局存储器进行数据计算，相比于传统CPU程序，计算效率明显提升。程序结构简洁，内有注释，方便学习，如有不足之处，方便批评指正。-Based on CUDA C programs can be arbitrary matrix matrix multiplication, the current global memory for data only, compared to traditional CPU program, computational efficiency improved significantly. Program structure is simple, there are notes, easy to learn, if inadequate, easy criticism.