spmv_csr

  稀疏矩阵的DIA/ELLPACK/COO/CSR/HYB表示形式，以及各表示形式下的稀疏矩阵乘法（稀疏大矩阵*矢量）的CUDA实现。对于矩阵中每一行稀疏元素个数较统一的情况，ELLPACK表示最佳，其次是HYB（ELL+COO）。 CUDA™ 是一种由NVIDIA推出的通用并行计算架构，该架构使GPU能够解决复杂的计算问题。 它包含了CUDA指令集架构（ISA）以及GPU内部的并行计算引擎。 开发人员现在可以使用C语言来为CUDA™ 架构编写程序(Sparse matrix DIA/ELLPACK/COO/CSR/HYB representation, as well as the representation of the sparse matrix multiplication (large sparse matrix* vector)' s CUDA implementation. For each row of the sparse matrix representing the number of elements the case of unification, ELLPACK that the best, followed by HYB (ELL+COO). NVIDIA CUDA ™ is introduced by a general purpose parallel computing architecture that makes the GPU to solve complex computational problems. It contains the CUDA Instruction Set Architecture (ISA) and the GPU parallel computing engine. Developers can now use the C language to write programs for the CUDA ™ architecture)