EMT: Nanophotonic Ultra-Low latency Data Interconnection Network for High Performance Computing

EMT：用于高性能计算的纳米光子超低延迟数据互连网络

• 批准号: 0523771
• 负责人: Keren Bergman
• 金额: --
• 依托单位: Columbia University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2005
• 资助国家: 美国
• 起止时间: 2005-09-15 至 2008-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0523771&HistoricalAwards=false
• 关键词: EMT; Nanophotonic; Ultra; Low; latency

Intellectual MeritThe scaling of switching elements to nanometer lengthscales raises a formidable challenge to the internal processing of the optical data packets while maintaining a memory-free switching fabric. At the device level,nanophotonic logic elements for information processing will follow two parallel paths:electro-optic switch nodes employing high-index-contrast nanophotonics and an all-optical switch node based on photonic band gap nanostructures.Both implementations will demonstrate the feasibility for ultra-dense nanophotonic integration,towards the ultimate realization of an ultra-low latency computational network on the semiconductor chip.Broader Impact Specific outreach modules on optical networks and nanophotonics will be developed for K-12 school teachers,and presented at schools with a high proportion of minority and underrepresented students around the New York metropolitan area.Undergraduates from our summer research programs will be actively recruited to graduate studies at Columbia and Georgia Tech.The proposed program aims to harness the extraordinary capabilities of nanoscale photonics and theimmense communication capacities of optical networks to create a high-risk,revolutionary paradigm inhigh-performance computation.Our approach tackles what is perhaps the most critical performancechallenge to future large-scale computing systems,namely the mounting communications bottleneck.The uniquely integrated program merges expertise in fundamental nanophotonic physics, optical systems,network topology,and performance analysis,and thus presents an unparalleled opportunity for trulyrevolutionary advances in high-performance computing.Beyond the direct impact on future high-endcomputing systems,this proposed work may potentially introduce highly innovative photonictechnologies towards new commercial applications.By demonstrating the viability of optical packetswitching and nanophotonic subsystems businesses which currently deploy high-capacity fully electronicswitches may consider the future insertion of nanophotonic-based interconnection networks.

智能化的优点交换元件的缩放到纳米长度尺度提出了一个艰巨的挑战，内部处理的光数据包，同时保持一个无记忆的交换结构。在器件级，用于信息处理的纳米光子逻辑元件将遵循两条并行路径：采用高折射率对比度纳米光子学的电光开关节点和基于光子带隙纳米结构的全光开关节点。这两种实现都将证明超密集纳米光子集成的可行性，最终实现半导体芯片上的超低延迟计算网络。更广泛的影响将为K-12学校教师开发光网络和纳米光子学的特定外展模块，并在纽约大都会地区少数民族学生比例较高和代表性不足的学校进行展示。我们将积极招募暑期研究项目的本科生毕业研究在哥伦比亚和格鲁吉亚理工学院。拟议的计划旨在利用纳米光子学的非凡能力和光网络的巨大通信能力，以创造一个高风险，革命性的范式在高性能计算。我们的方法解决了什么可能是最关键的性能挑战，以未来的大规模计算系统，即安装通信瓶颈。独特的集成程序融合了基础纳米光子物理学，光学系统，网络拓扑和性能分析的专业知识，从而为高性能计算的真正革命性进步提供了无与伦比的机会。除了对未来高端计算系统的直接影响外，这项拟议中的工作可能会为新的商业应用引入高度创新的光子技术。通过演示目前部署的光学分组交换和纳米光子子系统的可行性，高容量全电子开关可能会考虑未来插入基于纳米光子的互连网络。