Integrated Circuit Cluster Persistence During Placement Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Contemporary integrated circuit design implementations often neglect the compact placement of large clusters, leading to sub-optimal results due to inferior clustering and handling during placement optimization, which affects objectives like cell density, congestion, and timing.
Innovation Solution
A method that utilizes a force-directed approach to group cells into mobs and simultaneously move them to optimize empty space, implementing a tri-force method with spreading, mob center of gravity, and mob COG direction forces to achieve physical persistence without relying on prior physical constraints or large-scale circuit clustering.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If large clusters of cells are grouped and placed together during circuit placement, then cell density and congestion are improved, but the complexity of handling and optimizing these large clusters increases
Solution Approach 1:
The patent segments the large cluster handling problem into multiple hierarchical levels. Cells are first grouped into small clusters, then these clusters are organized into larger mobs, which are further organized into clusters of mobs. This multi-level segmentation allows the placement optimizer to handle large numbers of cells indirectly through their organizational hierarchy, reducing the computational complexity while maintaining high cell density and proper congestion management.
2Reliability
If contemporary placement methods focus on multi-objective optimization, then timing and wire length are optimized, but large cluster compactness and physical persistence are neglected
Solution Approach 1:
The patent applies preliminary action by establishing the hierarchical cluster-mob-cluster structure and applying physical persistence constraints before the main multi-objective optimization begins. The force-directed placement method pre-organizes cells into compact clusters with defined centers of gravity, creating a structured foundation that maintains cluster integrity throughout subsequent timing and wire length optimizations. This preliminary structuring ensures that large clusters remain compact while still allowing the optimizer to achieve timing closure.
3Stability of the object's composition
If physical constraints and large-scale circuit clustering are applied, then cluster persistence is improved, but the adaptability to handle clusters of any size and complexity decreases
Solution Approach 1:
The patent implements universality through its hierarchical cluster-mob-cluster structure that can adapt to handle clusters of any size. The same organizational framework and force-directed placement methodology work whether dealing with small groups of cells or very large clusters. The system automatically adjusts the number and organization of mobs within clusters based on the actual cluster size, providing a universal solution that maintains cluster persistence across diverse cluster configurations without requiring size-specific constraints.
Data Source
AI summary
The disclosed herein relates to method for persistence during placement optimization of an integrated circuit design. The method comprises performing cluster operation by grouping of a plurality of cells into a plurality of mobs. The method further comprises performing a spreading operation by moving the plurality of mobs and the plurality of cells simultaneously to optimize empty space of the integrated circuit design.