Senior Director 3DFabric Alliance TSMC San Jose, CA, United States
As the demand for faster and more efficient computing continues to grow, the traditional approach of relying on Moore's Law to increase the performance of individual chips is reaching its limits. A new paradigm shift is embodied in the concept of chiplets, which are modular components that can be combined to create powerful and flexible computing systems. The potential of chiplets lies in their ability to enable the integration of diverse technologies on a single chip. One of the key advantages of chiplets is their ability to facilitate the integration of specialized accelerators and memory modules, which are essential for AI and hyperscale computing tasks. By combining these components in a modular fashion, chiplet-based systems can be tailored to the requirements of different workloads, resulting in improved performance and energy efficiency. This presentation will cover the technical challenges and opportunities associated with chiplet-based technologies, as well as the implications for the future of computing.
