Agricultural mechanization so far has increasingly led to larger machines with large working widths to optimize their area output and which can be considered a single point of failure. Contrastingly, swarms of smaller machines provide an alternative for a better scalable and more resilient agricultural technology. Following the principles of swarm intelligent systems, the resulting decentralization promises improved robustness, i.e., through the possibility of the swarm to compensate for single failing machines in a self-organizing manner. In this paper, we propose a swarm simulation testbed for assessing the potential of robot swarms in agricultural fields at the process-level. In initial experiments to demonstrate the suitability of our testbed, we explore the scaling potential of a robot swarm compared to a single conventional machine during the process of seeding. We compared three coordination approaches to intelligently control the robot swarm. The approaches vary between: 1) A centralized control unit coordinating the swarm, 2) a decentralized swarm coordination, and, 3) a hybrid variant. The algorithms comprise combinations of the meta-heuristics simulated annealing, a genetic algorithm and a simple rule-based heuristic followed by the individual robots. Our experiments showcase the suitability of our swarm simulation testbed to explore the potential of swarm robotic solutions in agricultural field logistic problems. They also indicate that the decentralized swarm coordination approach results in a higher area output for higher number of machines working on small fields.