Thermally-Aware Power Distribution Networks for Vertically Integrated Systems

电子产品在我们的社会中扮演着举足轻重的角色。在过去的二十年里，便携式和手持设备的发展进一步扩大了电子产品在我们日常活动中的普及程度。一般来说，大多数现代电子产品需要一组不同种类的元件。这些产品包括医学和工业中用于安全、监测、预防或治疗的常见便携式设备或复杂系统。这种系统的异构性在未来将会加剧，因为对信息更快处理的主要需求被对环境刺激的准确感知和检测的需求增加。尽管这些设备创造了一个蓬勃发展的市场，但设计超出消费者需求的高性能和低功耗计算系统仍然是一个无处不在的挑战。更强大的计算机的设计不仅是为了回答重要的科学问题，也是为了满足新的或正在进行的社会需求。一个典型的例子是他们提供各种服务的数据中心，但相关的能源成本正在以惊人的速度增长。为了缓解这些问题，科学家和研究人员探索了颠覆性技术。一种很有前途的技术是三维(3-D)或垂直一体化。出于几个原因，这项新兴技术吸引了工业界和学术界。例如，考虑到便携式产品的情况，垂直集成系统可以通过将这些子系统堆叠成多层结构来大幅减小承载各种组件的电路板的尺寸，从而减小整体尺寸和提高功耗。该项目将通过为垂直集成系统的配电网络提供新的设计技术和创新的分析工具，为这些系统的发展做出贡献。为了更好地解释这项任务的复杂性和重要性，考虑一下为我们的住宅供电的电网。类似但受到严格约束的栅极向集成电路内的每个晶体管提供电流。这种垂直多层电路网络的设计复杂性要高得多，类似于多层建筑的电气安装比房屋的电气安装复杂得多的情况。这个问题的规模如此之大，必须为数十亿“消费者”(即晶体管)提供充足的电流，这突出表明需要更快、更准确的分析方法。在本项目中，由于电网资源有限且与其他资源竞争，优化方法将作为分析的补充。使用我们的民用类比，想一想这样一种情况，即电话和电源线竞争有限的开放空间。由于问题的规模以及设想的垂直系统的异质性，解决这个问题的有效方案不能仅仅求助于现有的设计方法和/或数值技术。因此，这项研究的目的是在考虑电力系统的特点和约束的同时，提出综合系统的电力网络分析和设计方法。该项目的成果将推动这一新兴技术的发展，使垂直整合更接近经济的大批量制造。

Electronic products have played an overarching role in our societies. The evolution of portable and handheld devices in the past two decades has further augmented the pervasiveness of electronics into our daily activities. In general, the majority of modern electronic products require a heterogeneous set of components. These products include common portable devices or sophisticated systems used in medicine and industry for safety, monitoring, prevention, or therapy. This system heterogeneity will be intensified in the future, since the primary demand for faster processing of the information is augmented by the requirement for accurate sensing and detection of environmental stimuli.Although these devices have created a flourishing market, the design of high performance and low power computing systems beyond consumer demands remains an omnipresent challenge. The design of more powerful computers is driven not only by the need to answer important scientific questions but also to address new or on-going societal needs. A characteristic example is the data-centers which they offer a variety of services but the relating energy-cost is increasing at an alarming rate. To mitigate these issues scientists and researchers explore disruptive technologies. A promising technology is three-dimensional (3-D) or vertical integration. This emerging technology has appealed to both industry and academia for several reasons. For example, considering the case of portable products, a vertically integrated system can drastically reduce the size of the board, which hosts the various components, by stacking these subsystems into a multi-tier structure reducing the overall size and improving power consumption.This project will contribute to the evolution of vertically integrated systems by providing new design techniques and innovative analysis tools for the power distribution network of these systems. To better explain the complexity as well as the significance of this task, consider the power grid that supplies our residences. A similar yet severely constrained grid provides current to each transistor within an integrated circuit. The design complexity of this network for vertical multi-tier circuits is much higher similar to the case where the electric installation of a multi-storey building is much more complicated than that of a house. The scale of the problem where billion of "consumers" (i.e. transistors) must be provided with abundant current underlines the need for faster and accurate analysis methods. In this project, the analysis will be complemented by optimization methods, as the resources for the power network are limited and compete with other resources. Using our civic analogy, think of a situation where phone and power cables compete for a limited open space.Efficient solutions to this problem cannot resort solely to existing design methodologies and/or numerical techniques due to the size of the problem as well as the heterogeneity of the envisioned vertical systems. The proposed research therefore aims at advancing the analysis and design methods of power networks for integrated systems while considering the particular traits and constraints relating to these systems. The results of the project will boost this emerging technology, bringing vertical integration a stride closer to economical high-volume manufacturing.