Power grids are an essential part of the energy infrastructure, acting as a bridge between power generation and the end consumers. Over the years, the grids have become more reliable, partly due to the introduction and usage of intelligent electronic devices (IED). These devices collect data from the grid and execute protection functionswhich help maintain the grid and avoid possible faults. Nowadays, chip manufacturers produce system on chipswith multiple computing units used in the IED. One such unit is the digital signal processor (DSP), specially tailored for digital signal processing. Hitachi ABB Power Grids produce IEDs and wanted to investigate whether offloading the execution of protection functions from the main CPU to the DSP would improve the performanceof the device, thus allowing for more complex configurations of functions. The work performed revolved around finding valid function candidates for the DSP by measuring the execution times on both processors, as well as evaluating the communication methods available between the CPU and DSP based on requirements of the protection functions. A mathematical model was created to help utilize the heterogeneous system efficiently. Results show that data-intensive functions generally perform better on the DSP, while shared memory is the preferable method for transferring data between the processors. The model can be used to determine if an offloading would benefit the performance of the IED and therefore improve the reliability of power grids.