Performance Optimisations for a Numerical Solution to a 3D Model of Tumour-Induced Angiogenesis on a Parallel Programming Platform

Abstract: The challenging issues of cancer prevention and cure lie in the need for a more detailed knowledge of the dynamic processes and mechanisms of cellular behaviour and tumour growth dynamics. In this paper we extend a previous 2D parallel implementation of a continuous-discrete model of tumour-induced angiogenesis to the more realistic 3D case. In particular, we look in-depth at available performance optimisation techniques to further improve the computational method and explore in more detail the hardware architecture. Recent evidence clearly indicates that GPU-accelerated computing can greatly facilitate researchers, clinicians and oncologists by performing time-saving in-silico experiments that have the potential to assist in quantifying cellular parameters, highlight model features, and help explore new cancer treatments and therapies