Enhancing Machine Learning Performance: The Role of GPU-Based AI Compute Architectures

This paper advances the field of GPU-based embedded intelligence (EI) by providing a comprehensive review of current and emerging architectures and applications. It covers key paradigms in GPU-based EI, focusing on architecture, technologies, and practical applications. The paper is structured as follows: (1) An overview and classification of GPU-based EI research, providing a broad perspective and concise summary of the paper's scope; (2) An in-depth discussion of various architectural technologies for GPU-based deep learning techniques and applications; and (3) A detailed examination of architectural technologies for GPU-based machine learning techniques and applications. This paper aims to offer valuable insights into the research area, encouraging further development of GPU-based EI for practical deployment and applications.