Novel methodologies for fast electromagnetic modeling of signal propagation on interconnects

消费电子和工业电子对更高功能和更低成本的持续增长的需求决定了微系统设计方法的演变。随着微系统设计呈现出更高的数据交换率和更紧密的组件集成，建模技术必须充分改变，以考虑到设备之间发生更紧密的物理耦合。因此，必须同时对微系统设计的大片段进行建模，同时保持分析的电磁准确性。为了克服使用简化的传统方法对分析施加的限制，这些方法只能对小的微系统片段进行建模，我们建议开发能够方便地处理相关计算复杂性的新一代原型方法。这些方法将使当今最快的计算电磁学方法适用于集成电子学的建模，并将它们映射到强大的高性能超级计算架构上。研究的主旨将是将最近提出的条件良好的积分公式推广到多层衬底的情况，用集肤效应包含的电流模型来增强它们，并将由此创建的算法框架移植到多处理器机器上。这种专门的软硬件协同将使电子设计人员能够对前所未有的规模和复杂性的集成电路进行电磁建模。这种建模方法有望被电子设计自动化行业采用，并建立一个新的标准，因为现有的商业工具严重缺乏解决相关复杂性的能力。拟议的研究计划将有助于缩小电子设计师的需求和能力之间的差距，并有可能减少当今大多数设计公司和半导体铸造厂因昂贵的重新设计和延迟的上市时间而造成的数十亿美元的损失。

The continuously increasing demand for higher functionality and lower cost of consumer and industrial electronics stipulates the evolution in microsystem design methodologies. As microsystem designs are exhibiting higher data exchange rates and tighter component integration, the modeling techniques must be changed adequately to account for the occurrence of tighter physical coupling between the devices. Thus, the large fragments of the microsystem designs must be modeled simultaneously while maintaining the electromagnetic accuracy of analysis. To overcome the constraints imposed on the analysis by the use of simplified traditional approaches capable of modelling only small microsystem fragments, we propose to develop a new generation of prototyping methodologies which can expediently handle the associated computational complexity. These methodologies will adapts today's fastest methods of computational electromagnetics to modeling of integrated electronics and map them onto powerful high-performance supercomputing architechtures. The main thrust of research will be in generalizing the recently proposed well-conditioned integral formulation to the case of multilayered substrates, augmenting them with skin-effect inclusive current flow models, and porting thus created algorithmic framework to the multiprocessor machines. This specialized software-hardware synergy will allow electronics designers to conduct electromagnetic modeling of integrated circuts of unprecedented sizes and complexity. Such modeling methods are expected to be adopted by the electronic design automation industry and establish a new standard as the available commercial tools greatly lack the capicity to address the associated complexity. The proposed research program will help close the gap between the needs and capabilities of electronics designers and has a potential to cut the multi-billion dollar losses incured by the majority of today's design houses and semiconductor foundries due to costly re-designs and delayed time-to-market.