Scientific applications and AI´s huge requirements are driving the development of exascale architectures based on CPUs, GPUs, and exotic architectures such as PIM (processing in memory). Accurate performance evaluation and modelling are required to design and test the next generation of programming models that will drive these architectures. This project will design, implement, and evaluate techniques to enable the execution of the next generation of applications on future clusters based on various technologies, ranging from Vector and Matrix CPUs (such as those developed by EuroHPC projects such as DARE) to future GPUs and accelerators developed by the major players in the field (Nvidia, AMD, Intel, Google, AWS, etc.). We will also evaluate the potential of exotic architectures, such as PIM (processing in memory) and neuromorphic architectures such as Bayesian Confidence Propagation Neural Networks (BCPNN)