The landscape of processor architectures in the upcoming Exascale and pre-Exascale systems is proving to be very diverse. The Fugaku system in Japan is a homogeneous CPU -only system, whilst the US systems are heterogeneous systems of CPUs and GPUs from an assortment of vendors. Applications therefore must be portable, and as HPC codes must achieve good performance too. In this talk, I will show how the performance portability of codes can be measured, analysed and visualised. These approaches provide tools and techniques for proactively understanding and improving the performance portability of your applications.