Implementing CUDA for optimizing matrix-vector multiplication (SGEMV) to achieve cuBLAS-ish performance. - View it on GitHub