Advances in AI Hardware Design in the Continuum: A Survey on Frameworks, Optimization, and Integration in Heterogeneous Systems

The increasing demand for optimized hardware solutions for artificial intelligence (AI) applications has driven the development of innovative approaches to hardware generation and system integration. This survey presents a review of current techniques used in hardware design, with an emphasis on frameworks addressing the key challenges in AI-oriented systems, including Design Space Exploration (DSE), multi-objective optimization, and dynamic adaptability. By analyzing existing methodologies for hardware construction, such as dedicated AI accelerators, we highlight the strengths and limitations of current solutions. The analysis underscores the opportunity to develop adaptive and self-aware frameworks for AI-oriented hardware generation, adopting a holistic approach that leverages design space exploration to optimize performance, energy efficiency, precision, and area. These frameworks should support real-time dynamic adaptation, ensure fault tolerance, and enable efficient integration between hardware and software in the computational continuum.