We have worked with HPC machines where standard programming techniques fail and parallel programming models are needed. We have also used complete distributed computing frameworks and low-level code optimisations to achieve very substantial speedups.