Industrial edge computing, combing various smart devices such as smart sensors, manufacturing equipment, and Internet of Things, has an ultimate goal to provide the industrial edge intelligence. Assigning large-scale tasks with multi-devices connection property for distributed edge servers is one of the main challenges to realize this goal. To address this question, a generative-coding group evolution algorithm, which involves a coding-based operator to approximate different sorts of evolutionary operators, is proposed in this paper. A simple grouping strategy is also introduced to accelerate the optimization process. Experimental results on three cases show that this algorithm is able to provide near-optimal solutions for large-scale tasks within a very short time compared with some traditional approaches.