This paper provides a comprehensive survey on the convergence of edge computing and deep learning (DL), highlighting the integration of these technologies to enable more pervasive and efficient AI services. The authors discuss the application scenarios, practical implementation methods, and enabling technologies for DL in edge computing, including DL training and inference in customized edge computing frameworks. They also address the challenges and future trends in realizing more fine-grained intelligence. The paper emphasizes the mutual benefits of edge intelligence and intelligent edge, where edge intelligence aims to facilitate the deployment of DL services by edge computing, and intelligent edge incorporates DL into edge computing frameworks for dynamic and adaptive maintenance and management. The survey covers key technologies such as DL applications on edge, DL inference in edge, edge computing for DL services, DL training at edge, and DL for optimizing edge. The authors aim to help readers understand the connections between enabling technologies and promote further discussions on the fusion of edge intelligence and intelligent edge, known as Edge DL.This paper provides a comprehensive survey on the convergence of edge computing and deep learning (DL), highlighting the integration of these technologies to enable more pervasive and efficient AI services. The authors discuss the application scenarios, practical implementation methods, and enabling technologies for DL in edge computing, including DL training and inference in customized edge computing frameworks. They also address the challenges and future trends in realizing more fine-grained intelligence. The paper emphasizes the mutual benefits of edge intelligence and intelligent edge, where edge intelligence aims to facilitate the deployment of DL services by edge computing, and intelligent edge incorporates DL into edge computing frameworks for dynamic and adaptive maintenance and management. The survey covers key technologies such as DL applications on edge, DL inference in edge, edge computing for DL services, DL training at edge, and DL for optimizing edge. The authors aim to help readers understand the connections between enabling technologies and promote further discussions on the fusion of edge intelligence and intelligent edge, known as Edge DL.