Improved parallel computations with matrices and polynomials