US8458442B2 - Method and Structure of Using SIMD Vector Architectures to Implement Matrix Multiplication