Design automation and analysis of three-dimensional integrated circuits
Author(s)Das, Shamik, 1977-
Massachusetts Institute of Technology. Dept. of Electrical Engineering and Computer Science.
MetadataShow full item record
This dissertation concerns the design of circuits and systems for an emerging technology known as three-dimensional integration. By stacking individual components, dice, or whole wafers using a high-density electromechanical interconnect, three-dimensional integration can achieve scalability and performance exceeding that of conventional fabrication technologies. There are two main contributions of this thesis. The first is a computer-aided design flow for the digital components of a three-dimensional integrated circuit (3-D IC). This flow primarily consists of two software tools: PR3D, a placement and routing tool for custom 3-D ICs based on standard cells, and 3-D Magic, a tool for designing, editing, and testing physical layout characteristics of 3-D ICs. The second contribution of this thesis is a performance analysis of the digital components of 3-D ICs. We use the above tools to determine the extent to which 3-D integration can improve timing, energy, and thermal performance. In doing so, we verify the estimates of stochastic computational models for 3-D IC interconnects and find that the models predict the optimal 3-D wire length to within 20% accuracy. We expand upon this analysis by examining how 3-D technology factors affect the optimal wire length that can be obtained. Our ultimate analysis extends this work by directly considering timing and energy in 3-D ICs. In all cases we find that significant performance improvements are possible. In contrast, thermal performance is expected to worsen with the use of 3-D integration. We examine precisely how thermal behavior scales in 3-D integration and determine quantitatively how the temperature may be controlled during the circuit placement process. We also show how advanced packaging(cont.) technologies may be leveraged to maintain acceptable die temperatures in 3-D ICs. Finally, we explore two issues for the future of 3-D integration. We determine how technology scaling impacts the effect of 3-D integration on circuit performance. We also consider how to improve the performance of digital components in a mixed-signal 3-D integrated circuit. We conclude with a look towards future 3-D IC design tools.
Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Electrical Engineering and Computer Science, 2004.Includes bibliographical references (p. 165-176).
DepartmentMassachusetts Institute of Technology. Dept. of Electrical Engineering and Computer Science.
Massachusetts Institute of Technology
Electrical Engineering and Computer Science.