Collaborative Research: DESC: Type II: Multi-Function Cross-Layer Electro-Optic Fabrics for Reliable and Sustainable Computing Systems

点击翻译按钮获取中文摘要

With the explosion of computing devices and systems in everyday consumer electronics, the amount of energy required to power information and computing technologies (ICT) is reaching close to 5% of worldwide carbon emissions. Embodied emissions originate from the manufacturing of hardware and from infrastructure-related energy such as procurement of raw materials, fabrication, packaging, and assembly. This is distinct from operational energy, which stems from using the hardware. Sustainability in computing should be based on three universally accepted tenets – reuse, reduce and recycle – of hardware components and systems. From a sustainability perspective, we need to reuse or repurpose existing hardware platforms for multiple functionalities to reduce embodied emissions, recycle hardware to extend or prolong lifetime reliability, and reduce the operational or use-phase energy required of the hardware platform. Therefore, future hardware platforms based on chiplets, - small, integrated circuits with defined functions - should be carefully designed to balance power, performance, and reliability when designed for sustainability. This research project establishes very important connections between electronic computing systems, photonic technology, computer architecture, machine learning, and sustainability requirements. The research will foster new research directions in several areas, spanning computer architecture, silicon photonics, algorithms, and applications, with the potential to significantly transform the design of next-generation sustainable chiplet-based heterogenous computing systems. All the research findings and simulation results will be shared with the community via conference/journal publication, professional meetings, and a dedicated website. The team is committed and will continue to expand on outreach activities, education, training, and broadening participation in computing as part of the project by making the necessary efforts to attract and train minority students in this field. This research will design multi-functional, self-healing and cross-layer optimized electro-optic fabric for computing architectures and accelerators to improve reliability, performance and sustainability. The overarching goal of the project is to enhance the sustainability of computing systems by reducing the impact of embodied energy and extending the operational lifetime of hardware systems. The team will explore a combination of electronics and silicon photonics comprehensively and systematically for both communication and computation in one integrated system to dramatically improve performance-per-Watt resource utilization, reliability, and sustainability of future computing systems. This research will result in (1) novel electrical and photonic interconnect-based architectures that have multiple functionality for communication, computation and storage, (2) self-healing and fault-tolerant electrical and optical fabric that improve the reliability of the heterogeneous computing chiplets, (3) hardware and cross-layer techniques to dynamically adapt to application demands to reduce operational energy, (4) an extensive modeling and simulation framework for evaluating embodied and operational energy of the proposed architectures, and (5) proof-of-concept testbed implementation.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.