A two-plane rotation (TPR) approach to Gaussian elimination (Jacobi) is used for computational efficiency in determining rotation parameters. A rotation processor is constructed using the TPR approach to perform singular value decomposition (SVD) on two by two matrices yielding both eigenvalues and left and right eigenvectors. The rotation processor can then be replicated and interconnected to achieve higher dimensioned matrices. For higher dimensional matrices, the rotation processors on the diagonal solve the 2×2 rotation angles, broadcast the results to off-diagonal processors, whereby all processors perform matrix rotations in parallel.