The side-channel attack is an attack method based on the information gained about implementations of computer systems, rather than weaknesses in algorithms. Information about system characteristics such as power consumption, electromagnetic leaks and sound can be exploited by the side-channel attack to compromise the system. Much research effort has been directed towards this field. However, such an attack still requires strong skills, thus can only be performed effectively by experts. Here, we propose SCNet, which automatically performs side-channel attacks. And we also design this network combining with side-channel domain knowledge and different deep learning model to improve the performance and better to explain the result. The results show that our model achieves good performance with fewer parameters. The proposed model is a useful tool for automatically testing the robustness of computer systems.