Grant Miller, S. Jain, Santosh Kelgeri, Pranav Ranganathan, A. Ceyhan
{"title":"Novel IR/EM-Aware Power Grid Design and Analysis Methodologies for Optimal PPA at Sub-10nm Technology Nodes","authors":"Grant Miller, S. Jain, Santosh Kelgeri, Pranav Ranganathan, A. Ceyhan","doi":"10.1109/IITC51362.2021.9537408","DOIUrl":null,"url":null,"abstract":"In this paper, we introduce novel, holistic, electromigration-(EM) and dynamic voltage drop-aware power grid (PG) design and analysis methods that can help resolve the critical limitations that guard-band-driven approaches of today’s modern design closure flows enforce upon physical designers in their quest to achieve the best possible power/performance/area (PPA). These methods can easily be integrated into any existing design flow. The proposed structured strategies to co-optimize inherent trade-offs in PG reliability and PPA improvement can help enable higher transistor density and accurately quantify the impact of IR drop for block-level timing within conventional automatic place-and-route (PnR) flows through the use of an exhaustive, but low-cost in-house solution. We demonstrate up to 9% area savings and up to 5% power reduction while maintaining achievable frequency. The proposed flow updates pave the path for future work to apply our machine-learning-enhanced design space exploration approaches to better control trade-offs between PG reliability and PPA improvement.","PeriodicalId":6823,"journal":{"name":"2021 IEEE International Interconnect Technology Conference (IITC)","volume":"11 1","pages":"1-3"},"PeriodicalIF":0.0000,"publicationDate":"2021-07-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2021 IEEE International Interconnect Technology Conference (IITC)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/IITC51362.2021.9537408","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
Abstract
In this paper, we introduce novel, holistic, electromigration-(EM) and dynamic voltage drop-aware power grid (PG) design and analysis methods that can help resolve the critical limitations that guard-band-driven approaches of today’s modern design closure flows enforce upon physical designers in their quest to achieve the best possible power/performance/area (PPA). These methods can easily be integrated into any existing design flow. The proposed structured strategies to co-optimize inherent trade-offs in PG reliability and PPA improvement can help enable higher transistor density and accurately quantify the impact of IR drop for block-level timing within conventional automatic place-and-route (PnR) flows through the use of an exhaustive, but low-cost in-house solution. We demonstrate up to 9% area savings and up to 5% power reduction while maintaining achievable frequency. The proposed flow updates pave the path for future work to apply our machine-learning-enhanced design space exploration approaches to better control trade-offs between PG reliability and PPA improvement.