首页> 外文学位 >Efficient full-wave simulation for very large scale off-chip interconnects.
【24h】

Efficient full-wave simulation for very large scale off-chip interconnects.

机译:针对超大规模片外互连的高效全波仿真。

获取原文
获取原文并翻译 | 示例

摘要

The requirement to simulate larger and more complex interconnect circuits is being driven by the rapid developments that are taking place in the integrated circuit industry, where more complex circuits are continually being designed. Since full-wave analyses rigorously account for all the higher-order modes, in addition to the transmission line mode (i.e., the Transverse ElectroMagnetic (TEM) mode), they provide more accurate results than conventional 2-D analysis tools, which are based on the assumption that only a TEM mode exists. Furthermore, a full-wave analysis is required to accurately model the physics of complex 3-D interconnects.; In order to address this need a Full-Wave Layered Interconnect Simulator (UA-FWLIS) was previously developed. UA-FWLIS is a Method of Moments (MoM) based tool for the analysis of stripline interconnects. However, UA-FWLIS could only handle a maximum of 10000 unknowns for signal traces in a single layer. Our final goal is to simulate complex practical systems, which have hundreds of thousands of unknowns and consist of multiple layers with vias interconnecting the different layers. In this dissertation, we extend the prototype full-wave simulator so that it can handle reactions between signal traces and vias, as well as reactions between two multiple layers. This is accomplished by employing analytical techniques to the reactions elements, thereby avoiding the use of inefficient numerical integration algorithms. This leads to substantial reductions in the matrix fill time, e.g., two orders of magnitude reductions for moderate size problems.; In addition to improving the matrix fill time, we also dramatically reduce the matrix solution time by employing sparse matrix solution techniques. We demonstrate that sparse reaction matrices are produced when modeling stripline interconnects provided that a parallel-plate Green's function is employed in the analysis. We found that by applying sparse matrix storage techniques and a sparse matrix solver, it is possible to dramatically improve the matrix solution time when compared with a commercial MoM-based simulator. This also makes it possible to solve much larger problems. The contribution of this dissertation empowers the current full-wave simulator to handle more realistic problems and makes full-wave simulations of very large scale stripline interconnect structures feasible.
机译:集成电路行业中的快速发展推动了对更大,更复杂的互连电路进行仿真的需求,在该行业中,不断设计更复杂的电路。由于全波分析严格考虑了所有高阶模式,因此除了传输线模式(即横向电磁(TEM)模式)外,它们还提供了比传统的二维分析工具更准确的结果假设仅存在TEM模式。此外,需要进行全波分析以精确地建模复杂的3-D互连的物理过程。为了满足该需求,先前已经开发了全波分层互连模拟器(UA-FWLIS)。 UA-FWLIS是基于矩量法(MoM)的工具,用于分析带状线互连。但是,UA-FWLIS在单层中最多只能处理10000条信号迹线的未知数。我们的最终目标是模拟复杂的实际系统,该系统具有成千上万的未知数,并且由多层结构以及通过通孔互连的不同层组成。在本文中,我们扩展了原型全波模拟器,使其能够处理信号走线和过孔之间的反应,以及两个多层之间的反应。这是通过对反应元素采用分析技术来完成的,从而避免了使用无效的数值积分算法。这导致矩阵填充时间的大量减少,例如,对于中等大小的问题,减少了两个数量级。除了改善矩阵填充时间外,我们还采用稀疏矩阵求解技术来显着减少矩阵求解时间。我们证明,如果在分析中采用平行板格林函数,则对带状线互连建模时会产生稀疏反应矩阵。我们发现,通过应用稀疏矩阵存储技术和稀疏矩阵求解器,与基于MoM的商用仿真器相比,可以大大缩短矩阵求解时间。这也使解决更大的问题成为可能。本文的贡献使当前的全波仿真器能够处理更现实的问题,并使超大型带状线互连结构的全波仿真成为可能。

著录项

  • 作者

    Wang, Xing.;

  • 作者单位

    The University of Arizona.;

  • 授予单位 The University of Arizona.;
  • 学科 Engineering Electronics and Electrical.
  • 学位 Ph.D.
  • 年度 2006
  • 页码 232 p.
  • 总页数 232
  • 原文格式 PDF
  • 正文语种 eng
  • 中图分类 无线电电子学、电信技术;
  • 关键词

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号