Optimisation of Parallel Scientific Applications on Highly Heterogeneous Modern Multicore and Multi-GPU Computing Nodes