In this document we describe the performance-critical numerical kernels extracted from a number of community codes, proposed by the scientific communities, which were identified in the first month of PRACE-2IP work package 8. The kernels have been selected according to the results of the performance analysis previously accomplished (and reported in deliverable D8.1.2). A number of approaches are envisaged for refactoring the kernels refactoring, in order to allow them to efficiently and effectively exploit the coming generation of HPC systems. This was part of the performance modelling methodology adopted in this work package in order to quantitatively evaluate the potential performance benefits that can be achieved on the new hardware architectures if novel software solutions are adopted. This final performance evaluation will be estimated by month six of the project (M6, )along with the specification of the final refactoring plan.