Design Enablement of Fine Pitch Face-to-Face 3D System Integration using Die-by-Die Place & Route

G. Sisto, P. Debacker, Rongmei Chen, G. V. D. Plas, R. Chou, E. Beyne, D. Milojevic
{"title":"Design Enablement of Fine Pitch Face-to-Face 3D System Integration using Die-by-Die Place & Route","authors":"G. Sisto, P. Debacker, Rongmei Chen, G. V. D. Plas, R. Chou, E. Beyne, D. Milojevic","doi":"10.1109/3DIC48104.2019.9058901","DOIUrl":null,"url":null,"abstract":"We present extensions to commercially available EDA tools to support Wafer-to-Wafer (W2W), Face-to-Face (F2F), hybrid bonding 3D integration with ~1μm pitch Cupad structures. Proposed flow is based on Innovus™ Place & Route (P&R) tool from Cadence Design Systems and allows functional 3D system partitioning with user specified partitioning information and automated netlist split. Due to fine 3D pitch, the partitioning can occur at lower levels of system hierarchy, resulting in significant number of 3D pins and with die-crossing critical paths. Proposed flow has been validated using memory-on-logic split of a single OpenSPARC-T2 core. L1 memory macros have been extracted from the post-synthesized gate-level netlist using a dedicated automated netlist partitioner. Per die netlists and top-level system have been fed into the P&R flow, memory die being implemented first. 3D structures have been assigned to 3D nets automatically, allowing their optimal placement with respect to the memory macros pins. 3D pin positions of the memory die have been propagated as a constraint for the implementation of the logic die, allowing optimized standard cell placement with respect to the 3D pins. Finally, we have enabled a cross-die timing analysis to assess the improvements of 3D in comparison to 2D. Our results show up to 40% of the total wirelength savings and 25 % on the maximum one, resulting in a 19% timing improvement on critical path.","PeriodicalId":440556,"journal":{"name":"2019 International 3D Systems Integration Conference (3DIC)","volume":"32 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"6","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2019 International 3D Systems Integration Conference (3DIC)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/3DIC48104.2019.9058901","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 6

Abstract

We present extensions to commercially available EDA tools to support Wafer-to-Wafer (W2W), Face-to-Face (F2F), hybrid bonding 3D integration with ~1μm pitch Cupad structures. Proposed flow is based on Innovus™ Place & Route (P&R) tool from Cadence Design Systems and allows functional 3D system partitioning with user specified partitioning information and automated netlist split. Due to fine 3D pitch, the partitioning can occur at lower levels of system hierarchy, resulting in significant number of 3D pins and with die-crossing critical paths. Proposed flow has been validated using memory-on-logic split of a single OpenSPARC-T2 core. L1 memory macros have been extracted from the post-synthesized gate-level netlist using a dedicated automated netlist partitioner. Per die netlists and top-level system have been fed into the P&R flow, memory die being implemented first. 3D structures have been assigned to 3D nets automatically, allowing their optimal placement with respect to the memory macros pins. 3D pin positions of the memory die have been propagated as a constraint for the implementation of the logic die, allowing optimized standard cell placement with respect to the 3D pins. Finally, we have enabled a cross-die timing analysis to assess the improvements of 3D in comparison to 2D. Our results show up to 40% of the total wirelength savings and 25 % on the maximum one, resulting in a 19% timing improvement on critical path.
查看原文
分享 分享
微信好友 朋友圈 QQ好友 复制链接
本刊更多论文
采用逐模位置和路径的小间距面对面3D系统集成设计实现
我们提供扩展的商用EDA工具,以支持晶圆对晶圆(W2W),面对面(F2F),混合键合3D集成与~1μm间距Cupad结构。拟议的流程基于Cadence Design Systems的Innovus™Place & Route (P&R)工具,允许使用用户指定的分区信息和自动网表分割进行功能性3D系统分区。由于精细的3D间距,划分可以发生在较低的系统层次上,导致大量的3D引脚和模交叉关键路径。使用单个OpenSPARC-T2内核的内存逻辑分裂验证了所提出的流程。L1内存宏是使用专用的自动化网络列表分区器从后合成的门级网络列表中提取出来的。每个模具网络列表和顶层系统都被输入到P&R流程中,内存模具首先被实现。3D结构被自动分配到3D网,允许它们相对于内存宏引脚的最佳位置。存储芯片的3D引脚位置已被传播为逻辑芯片实现的约束,允许相对于3D引脚优化标准单元放置。最后,我们启用了交叉模定时分析,以评估与2D相比3D的改进。我们的结果显示,总长度节省了40%,最大长度节省了25%,从而在关键路径上改善了19%的时间。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 去求助
来源期刊
自引率
0.00%
发文量
0
期刊最新文献
A Graph-Based Model of Micro-Transfer Printing for Cost-Optimized Heterogeneous 2.5D Systems Growth Optimization of Multi-Layer Graphene for Thermal-TSV Application in 3D-LSI Investigation of the Influence of Material Properties on Warpage and Solder Joint Reliability of 2.5D & FO Package Fraunhofer's Initial and Ongoing Contributions in 3D IC Integration Electrostatic Shield TSVs to Suppress Coupling Among Stacked ICs
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
现在去查看 取消
×
提示
确定
0
微信
客服QQ
Book学术公众号 扫码关注我们
反馈
×
意见反馈
请填写您的意见或建议
请填写您的手机或邮箱
已复制链接
已复制链接
快去分享给好友吧!
我知道了
×
扫码分享
扫码分享
Book学术官方微信
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术
文献互助 智能选刊 最新文献 互助须知 联系我们:info@booksci.cn
Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。
Copyright © 2023 Book学术 All rights reserved.
ghs 京公网安备 11010802042870号 京ICP备2023020795号-1