Evaluating the performance of estimation algorithms is not a trivial task. If considering a single simulation of a system it is hard to say if the observed performance is due to a modified algorithm or the inherent randomness in the considered system itself.
In order to do correct performance evaluations, the simulation of the performance of the algorithm must instead be conducted multiple times to attain statistical sound results. This is known as Monte Carlo simulation and it often consumes a lot of computing hours.
By putting the Monte Carlo simulation in a computing facility, such as via SNIC, new ideas and systems can be tested faster and more effectively to a better statistical certainty.