Optimising the use of all the cores of a hybrid multi-core CPU and its accelerating GPUs is becoming increasingly important as such combined systems become widely available. We show how a complex interplay of cross-calling kernels and host components can be used to support good throughput performance on hybrid simulation tasks that have inherently serial analysis calculations that must be run alongside more easily parallelisable simulation time-stepping calculations. We present results for a cluster component-labelling analysis performed during simulation of a Potts lattice simulation model. We discuss how these hybrid techniques can be more broadly applied to this class of numerical simulation experiments in computational science.
