Force Directed Analytical Placer for Programmable ICs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current techniques for global placement of circuit designs on programmable integrated circuits do not account for control set constraints, leading to decreased resource utilization and infeasible placement solutions due to physical and control set overlaps.
Innovation Solution
A method and system that subdivides the programmable IC into a grid of cells, calculates forces based on control set overlap, and applies these forces to components to solve linear equations for optimal global placement, ensuring compliance with control set constraints and minimizing physical and control set overlaps.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional global placement techniques are used, then placement speed is maintained, but control set constraints are violated leading to infeasible solutions
Solution Approach 1:
The patent transforms the placement problem by changing the parameter space from direct component positioning to force-based continuous coordinates. By representing placement as a physics-inspired force equilibrium problem, the algorithm naturally satisfies control set constraints while maintaining feasibility. The force model incorporates control set density as a key parameter that guides components to appropriate control sets.
Solution Approach 2:
The patent introduces force vectors as intermediary elements between components and control sets. Rather than directly constraining component positions, the force model acts as a mediator that indirectly guides placement decisions. The control set density calculation serves as an intermediary metric that translates constraint violations into corrective forces.
2Productivity
If control set constraints are enforced, then resource utilization improves, but placement computation time increases
Solution Approach 1:
The patent performs preliminary calculation of control set density for each cell before the actual placement iteration. This pre-computed density information is then used to determine force directions, avoiding repeated complex calculations during each placement step. The force model leverages this preliminary density data to efficiently guide component placement while maximizing resource utilization.
Solution Approach 2:
The placement algorithm employs periodic iteration where forces are recalculated and applied in cycles. During each iteration period, components are moved based on current force directions, then forces are updated based on new density distributions. This periodic recalibration allows the system to converge to optimal placement while maintaining computational efficiency through structured repetition.
3Manufacturing precision
If force directed placement is applied, then physical overlap is reduced, but control set overlap may increase
Solution Approach 1:
The patent applies different force characteristics to different spatial regions based on local control set density. Rather than using uniform force fields, the algorithm calculates local density for each cell and adjusts force directions accordingly. This localized approach ensures that components are guided to appropriate control sets while maintaining physical placement accuracy, preventing both physical and control set overlaps.
Data Source
AI summary
A computer-implemented method of globally placing a circuit design on a programmable integrated circuit (IC) includes dividing, by a placement system, the programmable IC into a grid comprising a plurality of cells, assigning each component of a selected component type of the circuit design to one of a plurality of control set groups according to a control set of the component, and calculating a force including a control set force that depends upon overlap of control sets within the plurality of cells. The method further can include applying the force to at least one selected component of the circuit design and assigning components of the circuit design to locations on the programmable IC by solving a set of linear equations that depend upon application of the force to the at least one selected component to create a global placement. The circuit design including the global placement can be output.