Text this: Load Partitioning for Matrix-Matrix Multiplication on a Cluster of CPUGPU Nodes Using the Divisible Load Paradigm