Unified Improvement Scoring for IC Rebuffering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional ROI algorithms for rebuffering in electronic design automation (EDA) often result in negative improvement scores for rebuffering candidates with lower buffering costs, preventing their selection despite potential benefits, due to their comparison to initial buffer trees.
Innovation Solution
A unified improvement score calculation method is introduced, where a reference buffer tree with the lowest buffering cost is selected, allowing for the computation of improvement scores relative to this reference, ensuring the selection of the best rebuffering candidate with the highest positive ROI, regardless of cost differences.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional ROI algorithms compare rebuffering candidates to initial buffer trees, then improvement scores can be calculated, but rebuffering candidates with lower buffering costs receive negative scores and are prevented from selection
Solution Approach 1:
The patent inverts the conventional ROI calculation approach by selecting the rebuffering candidate with the lowest buffering cost as the reference buffer tree, rather than using the initial buffer tree as reference. This inversion ensures that all other candidates are compared against a lower-cost baseline, eliminating negative scores and enabling flexible selection of candidates that optimize design objectives.
2Quantity of substance
If rebuffering candidates with lower buffering costs are selected, then cost is reduced, but conventional algorithms assign negative improvement scores preventing their selection
Solution Approach 1:
The patent changes the reference parameter for ROI calculation from the initial buffer tree to the rebuffering candidate with the lowest buffering cost. This parameter change transforms the ROI calculation so that candidates with lower costs receive positive scores, enabling their selection while reducing overall buffering costs.
3Stability of the object's composition
If the initial buffer tree is used as reference for ROI calculation, then conventional scoring is maintained, but the best rebuffering candidate may not be selected due to negative scores
Solution Approach 1:
The patent performs a preliminary action by identifying and selecting the rebuffering candidate with the lowest buffering cost before conducting the ROI calculation. This preliminary selection establishes a new reference buffer tree that ensures all subsequent candidates will receive positive improvement scores, enabling optimal design achievement.
Data Source
AI summary
Aspects of the present disclosure address improved systems and methods for rebuffering an integrated circuit (IC) design using a unified improvement scoring algorithm. A plurality of rebuffering candidates are generated based on an initial buffer tree in an integrated circuit (IC) design. A rebuffering candidate in the plurality of rebuffering candidates comprises a modified buffer tree based on the initial buffer tree. A buffering cost of each rebuffering candidate is determined. A reference buffer tree is selected from among the rebuffering candidates based on the buffering cost of each rebuffering candidate. An improvement score of each rebuffering candidate is determined based on the buffering cost of each rebuffering candidate relative to the reference buffer tree. A new buffer tree is selected from among the plurality of rebuffering candidates to replace the initial buffer tree based on the improvement score of each rebuffering candidate.