In the field of high-performance computing, with its myriads of architecture, hardware and communication protocols, programs are often written to be highly specialized to extract maximal performance from the particular environment. Therefore, in-depth knowledge of the hardware and software tools is often required even for simple programs, a hurdle that may be daunting for newer and experienced programmers alike. In this thesis, we attempt to solve this problem by providing an efficient and versatile abstraction layer that is easy and fast to use. The actor library provided separates computation and communication routines and adapts to many HPC environments. The library uses the GPI library as the communication API, and aims to replicate close to native performance on various problems. We test the library with a tsunami simulation model, and we see that our implementation performs and scales comparably to Actor-UPC++, the original library that the work was ported from.