VLSI Design Optimization Techniques

Very-large-scale integration (VLSI) design optimisation encompasses a suite of algorithmic and heuristic strategies aimed at enhancing performance, power efficiency and area minimisation of integrated circuits. Key stages of the physical design flow—partitioning, placement and routing—present tightly coupled constraints in advanced technology nodes. Partitioning techniques divide a design into manageable blocks, balancing communication overhead and resource utilisation. Placement algorithms assign circuit cells to discrete locations on the silicon die, targeting minimal wirelength and congestion while respecting timing and power budgets. Routing methods establish the interconnecting wires between placed cells, further constraining delay and manufacturability. Recent trends have emphasised co-optimisation of placement and routing to avoid conservative margin allocation, the integration of machine-learning-driven predictors to accelerate decision-making and the adoption of mixed-integer and quadratic programming for more accurate modelling of complex constraints. These advances support globally significant applications from high-performance computing to energy-harvesting devices, ensuring that VLSI solutions meet stringent reliability and cost requirements in an era of ever-shrinking feature sizes.

Research from Nature Portfolio

No recent Nature Portfolio content available.

Research from all publishers

In placement optimisation, comprehensive reviews have charted the evolution of techniques that accelerate physical design by combining algorithmic refinements with computational acceleration. Emerging flows employ learning-based predictors to anticipate congestion hotspots and guide iterative refinement, achieving significant reductions in runtime without compromising quality. Generative adversarial learning has been harnessed to transfer placement expertise from commercial tools to open-source GPU-accelerated placers, yielding up to 8% improvements in wirelength and power metrics and demonstrating persistent gains through to the post-route stage. Concurrently, cooperative frameworks that tightly couple routing and placement stages have been developed for advanced nodes; by embedding routing feedback into placement engines via integer linear programming and cost-caching strategies, these approaches reduce global routing runtimes by almost 30% and improve detailed routing outcomes with minimal cell movement. Together, these studies illustrate the potency of co-design and machine learning in advancing VLSI physical design optimisation.

VLSI Design Optimization Techniques publication trend

The graph below shows the total number of articles in vlsi design optimization techniques across all publications each year (not limited to Nature Index journals).

Technical terms

Placement: Assignment of circuit cells or modules to specific locations on the silicon chip to minimise interconnect length and timing delays.

Routing: Determination of paths for electrical connections between placed cells, subject to congestion, delay and design-rule constraints.

Wirelength: Total length of interconnects in a design, often used as a proxy for signal delay and power consumption.

Power–Performance–Area (PPA): A comprehensive metric assessing an integrated circuit’s energy usage, speed and silicon footprint.

Co-optimisation: A methodology that simultaneously considers two or more stages (e.g., placement and routing) to reduce conservative design margins and enhance overall efficiency.

References

1. Detailed Placement and Global Routing Co-Optimization with Complex Constraints. Electronics (2021).
2. Progress of Placement Optimization for Accelerating VLSI Physical Design. Electronics (2023).
3. GAN-Place: Advancing Open Source Placers to Commercial-quality Using Generative Adversarial Networks and Transfer Learning. ACM Transactions on Design Automation of Electronic Systems (2024).
4. CRP2.0: A Fast and Robust Cooperation between Routing and Placement in Advanced Technology Nodes. ACM Transactions on Design Automation of Electronic Systems (2023).