FuSe-TG: Open, Multiscale, Application-Agnostic Platform for Heterogeneous System-in-Package Co-Design

点击翻译按钮获取中文摘要

Heterogeneous integration (HI) has shown tremendous potential to overcome the cost and performance challenges of advanced monolithic integration technology and effectively combat the slow-down of Moore’s law. However, HI constitutes a grand challenge in engineering since it must simultaneously address the electrical, optical, thermal, mechanical, and material challenges of integrating separately manufactured/designed components into a higher-level System-in-Package (SiP), for a diverse range of technologies. Currently, the design and optimization of functional blocks and chips are implemented in different domains without self-consistent links between different models, and without co-design or co-optimization between them. This project, if successful, will translate to sustained innovations in the microelectronics industry at-large and permit the US to maintain global leadership in microelectronics, delivering systems that achieve more-than-Moore scaling for years to come. This teaming project will help meet the future workforce needs in microelectronics, utilizing the Nation’s First Semiconductor Degrees Program (SDP) just launched at Purdue University. The project will leverage partnerships with U of Central Florida, UC Santa Barbara, and UC San Diego, all minority-serving institutions to attract underrepresented students, and will also partner with Community Colleges to develop training programs.The technical objective of this teaming effort is to break the barriers across applications, design domains, and abstraction levels, in modeling, simulation, optimization, and design through an open-source, multiscale, application-agnostic platform (Open-MAP) for heterogeneous SiP co-design. Open-MAP encompasses three key components: (1) A multiscale multiphysics modeling and analysis seamlessly integrated with either first-principles accuracy or a user-desired level of accuracy, which is further enriched by data, ML-assisted, and uncertainty quantification techniques; (2) A suite of multiphysics-informed design and optimization tools to automate co-design from intent to fabrication; (3) A series of co-design projects that will be conducted to identify critical barriers to co-design, for informing and guiding the development of Open-MAP, and eventually demonstrating the co-design advantages using the Open-MAP capabilities.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.