The industrial internet of things (IIoT) is facing enormous challenges, including limited computing resources and network bandwidth and high energy consumption. To facilitate the realisation of IIoT-enabled manufacturing, several problems concerning computing modes and interaction patterns among fog nodes and the cloud should be clarified: what the typical computing modes are; what their advantages and drawbacks are; and how the tasks are calculated in different computing modes. The above-listed problems are the focus of this paper. Four computing modes are listed accordingly, namely, local computing, remote computing, cloud computing and fog computing. The system model is established for each kind of computing mode, and the models for delay and energy consumption are formulated. A task processing experiment platform is setup in simulation software. The experimental results show that compared with other computing modes, fog computing exhibits better performance in terms of delay and energy consumption. Copyright © 2019 Inderscience Enterprises Ltd.