Signoff-Driven Floorplanning Under Foundry and Technology Constraints for Multi-Voltage-Domain SoC Designs

Abstract

Advanced system-on-chip (SoC) designs increasingly rely on multi-voltage domains, aggressive power-performance-area-cost (PPAC) targets, and stringent foundry signoff requirements. At advanced technology nodes, late-stage design rule check (DRC), layout versus schematic (LVS), IR-drop, and electromigration (EM) violations frequently trigger expensive engineering change orders (ECOs), impacting time-to-market. Traditional physical design flows treat signoff as a post-routing activity, leading to poor convergence and closure. This paper presents a signoff-driven physical design methodology that integrates foundry and technology constraints directly into floorplanning, placement, clock tree synthesis (CTS), and routing for multi-voltage-domain SoC designs. Mathematical cost models, early power distribution network (PDN) estimation, voltage-island-aware placement, and ECO-aware routing strategies are introduced. The proposed methodology significantly reduces late signoff iterations and improves first-pass silicon success.

Introduction

Modern SoCs integrate CPUs, GPUs, AI accelerators, memory, and high-speed interfaces, operating under aggressive power budgets with multiple voltage domains, power gating, and fine-grained clocks. While these techniques reduce power, they greatly increase physical design complexity, especially at advanced nodes with thousands of signoff-driven design rules. Late-stage violations often lead to costly ECOs, emphasizing the need for signoff-aware floorplanning from the earliest stages.

Key Contributions of Signoff-Aware Floorplanning:

1. Cost models incorporating timing, IR-drop, congestion, DRC density, and voltage-island constraints.

2. Voltage-island-aware placement and level-shifter planning.

3. Early PDN (power distribution network) and IR-drop estimation coupled with placement.

4. Routing and CTS strategies to minimize late-stage ECOs.

Fundamentals of Multi-Voltage-Domain SoCs:

• Power domains are physical/logic partitions with dedicated VDD/VSS; voltage islands are their physical implementations.

• Types: Always-on, switchable, retention domains.

• Inter-domain signals require level shifters (LS), isolation cells (ISO), and retention registers (RET).

Floorplanning Strategies:

• Power-Grid-Centric Floorplanning: Define domain-specific grids early to handle IR-drop, EM, and noise.

• Voltage-Island-Aware Macro Placement: Cluster macros by voltage domain and current demand; isolate noise-sensitive blocks.

• Level Shifter & Isolation Cell Planning: Place LS near receiving domains and ISOs in always-on regions to reduce late-stage ECOs.

Signoff Considerations:

• IR-drop Budgets: Domains have tight voltage drop limits; high-performance areas need denser grids.

• Electromigration (EM): High-current paths and level shifters are EM hotspots; floorplans must minimize long high-current routes.

• Manufacturability & Yield: Voltage islands should be rectangular and aligned to routing grids; uneven densities require careful filler and decap planning.

Macro Placement Constraints:

• High-current macros must be clustered near power entry points and compatible with voltage domains.

• Macro placement is constrained by EM, DRC, and power-switch rules, making it a power-integrity-driven optimization.

Routing-Aware Floorplanning:

• Routing resources are reduced due to stricter spacing, wider power straps, and duplicated grids.

• Pre-reserve routing channels for clock trunks and high-fanout nets.

• Early global routing estimation guides channel widths, congestion mitigation, and macro positioning.

ECO-Aware Considerations:

• Reserve space for buffers and hold-fix paths.

• Mitigate cross-domain ECO issues by planning level shifters and isolation cells early.

Conclusion

This paper presented a signoff-driven floorplanning methodology for multi-voltage-domain (MVD) SoC designs under stringent foundry and advanced technology constraints. As semiconductor technologies scale into deeply advanced nodes, the interaction between physical layout decisions, manufacturing rules, and signoff requirements has become increasingly complex. Conventional floorplanning approaches, which primarily optimize for area and performance, are insufficient to address the compounded challenges of timing closure, power integrity, routability, and manufacturability in MVD designs. The analysis further demonstrates that floorplanning serves as the primary control point for downstream placement, CTS, and routing quality, especially in MVD SoCs where cross-domain interactions dominate physical design complexity. Early awareness of routing congestion, EM/IR margins, and timing criticality enables more balanced trade-offs between performance, power, and area while maintaining compliance with foundry signoff requirements. As a result, signoff-driven floorplanning improves implementation predictability, enhances first-pass silicon success, and shortens overall design cycles. In conclusion, signoff-driven floorplanning is no longer optional but a fundamental requirement for successful MVD SoC implementation at advanced technology nodes. Integrating foundry and technology constraints at the floorplanning stage establishes a robust foundation for downstream physical design stages and mitigates the risk of costly late-stage fixes. Future research directions include the integration of AI-assisted prediction models, adaptive floorplan refinement loops, and cross-domain optimization frameworks to further improve signoff robustness and scalability for next-generation SoC designs.

References

[1] J. Cong, M. Xie, and Y. Zhang, “Thermal-aware floorplanning for multi-core SoC designs,” IEEE Trans. Computer-Aided Design Integr. Circuits Syst., vol. 31, no. 10, pp. 1525–1538, Oct. 2012. [2] D. Sylvester, K. Keutzer, and C. Hu, “IR drop in advanced VLSI design,” IEEE Design & Test of Computers, vol. 21, no. 6, pp. 418–425, Nov.–Dec. 2004 [3] S. R. Nassif, “Power grid analysis benchmarks,” in Proc. IEEE Int. Symp. Quality Electronic Design (ISQED), 2008, pp. 376–381. [4] A. B. Kahng, S. Kang, R. Kumar, and J. Sartori, “Designing a physical design flow for advanced technology nodes,” IEEE Trans. Computer-Aided Design Integr. Circuits Syst., vol. 36, no. 8, pp. 1259–1272, Aug. 2017. [5] P. Saxena et al., “multi-voltage and power-gated SoC physical design challenges,” in Proc. IEEE Custom Integrated Circuits Conf. (CICC), 2015, pp. 1–8. [6] M. Cho, D. Z. Pan, and R. Puri, “Power-aware physical design for multi-voltage designs,” IEEE Trans. Very Large Scale Integr. (VLSI) Syst., vol. 20, no. 5, pp. 845–858, May 2012. [7] A. B. Kahng and Q. Wang, “Challenges and opportunities in DFM-aware physical design,” IEEE Design & Test, vol. 34, no. 4, pp. 6–15, Aug. 2017. [8] C. Visweswariah et al., “Signoff closure challenges at advanced nodes,” in Proc. IEEE Design Automation Conf. (DAC), 2018, pp. 1–6. [9] S. Hu et al., “Routing congestion prediction and mitigation in advanced technology nodes,” IEEE Trans. Computer-Aided Design Integr. Circuits Syst., vol. 39, no. 11, pp. 3958–3970, Nov. 2020. [10] K. Bernstein et al., “High-performance CMOS variability in the 65-nm regime and beyond,” IBM J. Res. Develop., vol. 50, no. 4–5, pp. 433–449, Jul.–Sep. 2006.

Copyright

Copyright © 2025 Mukul Anand, Muskan Singh. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.