Nowadays, the main defense against fault injection attacks targeting the execution unit of a processor is made by applying redundancy, meaning a simple duplication of the data and circuitry involved in computations. But, considering the growing use of open source processors, so we can assume the possibility for the attacker to perfectly know the system chosen to be attacked, and considering that fault injection attack are under constant evolution, redundancy measure are not longer effective. One possible solution is changing domain of data representation and perform calculations on those data in a parallel way instead of making a raw copy of the data and circuitry.