Download presentation
Presentation is loading. Please wait.
Published byRiley Mallatt Modified over 9 years ago
1
PAGE 1 Design-for-Si Initiatives - Process-Design Integration - M Nowak / Riko R March 2007
2
2 Riko R Demo at 65 Deploy at 45 Demo at 45 Deploy at ? -3 -4 -2 -1 Design–for–Si Initiatives for Post Design-Rule Era Its the Process-Design Integration, Stupid !!! Improve product cost-performance by alignment of process and design sweet spots Yield & Layout Its the DFM Simulators ! Yield & Layout Its the DFM Simulators ! Variability & Corners Its Geometry Aware Timing ! Variability & Corners Its Geometry Aware Timing ! Physical Design & Cost Its the RDR Opportunity ! Physical Design & Cost Its the RDR Opportunity ! Architecture Design & Cost Its VAM & VSP ! Architecture Design & Cost Its VAM & VSP ! NOW-ish
3
3 Riko R 1 : Zero Physical DFM Margin Use “DFM ECO” Loop to add the DFM Pad only where it may be necessary. Use the DFM Simulators to identify where are the opportunities Value Proposition : Smaller Die with Managed Functional Yield Risk Start with an Aggressive Library that excludes (most of ) DFM Rule padding Its all about Layout Polishing
4
4 Riko R Tighten corners to exclude Shape Tighten corners to exclude Shape, Thickness, OCV ECO to correct impact of block level interactions on yield or performance DFM LPE to account for non- litho systematics OCV to account for global systematics used in SSTA Tighten corners to exclude Thickness & Shape 2 : Zero Electrical DFM Margin Value Proposition : Tighter Margins with Managed Parametric Yield Risk Its all about Geometry Awareness in the Timing Flow
5
5 Riko R 3 : Correct by Construction Layout [= RDR ??] Not a new concept Rigorous gridded design has been around for a while Traditionally it cost area vs. more flexible rules – Hence rejected in 65 and 45 (so far) – especially by volume shops Reason for this (inevitable) conclusion – Comparison based on same set of Design Rules – Comparison based on same Design Methodology New RDR (Compelling ? ) Drivers Some say RDR is inevitable – question is when (32 ?) – But then that is what they said about it at 65 and 45 and … – Dual Exposure Demands Design Fragmentation ? – EUV Negates -29% / year Cost / Gate Reduction ? – Strain Variability Management ? New RDR Proposition (s) Negate area loss through better yield, or Tighter lay out rules, or Better design methodologies Establishing Collaborative efforts with select supply chain partners to explore trade-offs Value Proposition : DFM-Correct by Construction Layout It (should be) all about manufacturable topographies
6
6 Riko R 4 : Holistic Process-Design Co Optimization Our Strategy : IFM Fabless entities tend to Have an Intrinsic Disadvantage in this Arena – Less intimate with the process world Morph the fabless model to address the challenge : Integrated Fabless Manufacturer – Necessary at the bleading edge Requirement : A Structured Methodology for Co-Optimization Analyses To steer the design To steer the process Two Exploratory Co-Optimization Initiatives In Parametric Domain : – Performance vs power vs variability In Area (Cost) Domain : – Area vs Yield vs Design Content …. Process Design Value Proposition : Optimized Process & Design “Sweet Spot” Alignment It (should be) all about manufacturable architectures
7
7 Riko R Restricted Design Rules = Designability Risks Must manage via use of pushed rules or custom techniques to negate area loss ADVANTAGE : better manufacturability and yields, and easy design interface DISADVANTAGE : Cost Benefits vs Process Maturity 3 : DFM Junction in the Road ? Flexible Design Rules = Manufacturability Risks Must manage via Use of DFM Rules or DFM Simulators & Models to “Polish” the Layout ADVANTAGE : Minimum overhead for Area or margin for Performance DISADVANTAGE : it is complex and carries overhead in the design flow Product Specific Cost-Benefit Trade Off Analyses
Similar presentations
© 2024 SlidePlayer.com Inc.
All rights reserved.