Dual-Rail Logic Synthesis for Semiconductor Timing and Area Tradeoffs

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The miniaturization of integrated circuits has led to stricter design and manufacturing specifications, requiring more efficient electronic design automation (EDA) processes for logic synthesis, particularly in optimizing semiconductor device designs for performance, area cost, and power consumption, while ensuring that existing methods fall short in effectively addressing these challenges.

Innovation Solution

The proposed solution involves a logic synthesis process that includes technology-independent and technology-dependent optimization stages, utilizing a cell library with dual-rail modules such as Pre-Charged Half Buffer (PCHB) and Weak Conditioned Half Buffer (WCHB) modules, along with hybrid modules, to optimize logic netlists through technology mapping and performance analysis, allowing for flexible optimization of timing, area, and power consumption.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Speed

If logic synthesis processes are optimized for performance, then device speed improves, but area cost increases

Engineering Contradiction:
Improvedevice speedVSAvoidarea cost
Core Design Contradiction:
SpeedVSArea of stationary object

Solution Approach 1:

The patent applies parameter changes by utilizing dual-rail modules (PCHB and WCHB) with different operational characteristics. The Pre-Charged Half Buffer (PCHB) module optimizes for speed with pre-charged nodes, while the Weak Conditioned Half Buffer (WCHB) module optimizes for area efficiency. The logic synthesis process dynamically selects between these modules based on timing constraints and area requirements, changing the implementation parameters to resolve the contradiction between speed and area cost.

Inventive Principle:
Principle #35Parameter changes

2Use of energy by moving object

If device miniaturization is pursued, then power consumption decreases, but manufacturing precision requirements increase

Engineering Contradiction:
Improvepower consumptionVSAvoidmanufacturing specification
Core Design Contradiction:
Use of energy by moving objectVSManufacturing precision

Solution Approach 1:

The patent applies segmentation by dividing the logic synthesis process into technology-independent optimization and technology-dependent optimization stages. This segmentation allows separate handling of different design concerns: the first stage focuses on logical optimization independent of manufacturing constraints, while the second stage addresses technology-specific implementation details including manufacturing precision requirements. This segmented approach enables miniaturization for lower power consumption while systematically managing the increased manufacturing precision requirements.

Inventive Principle:
Principle #1Segmentation

3Productivity

If EDA process complexity is reduced, then manufacturing efficiency improves, but design optimization capability deteriorates

Engineering Contradiction:
Improvemanufacturing efficiencyVSAvoidEDA process complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent applies preliminary action by performing technology-independent logic optimization before technology-dependent optimization. This preliminary optimization stage prepares the logic netlist with optimized structures and identifies key performance parameters in advance, reducing the complexity of subsequent technology-specific optimization. By doing the heavy logical optimization work first, the EDA process becomes more efficient in the manufacturing stage while maintaining comprehensive design optimization capability through the two-stage approach.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentUS11120183B2Computer product for making a semiconductor device
Publication Date: 2021.09.14 TAIWAN SEMICONDUCTOR MANUFACTURING CO LTD
  • US11120183B2 patent drawing
  • US11120183B2 patent drawing
  • US11120183B2 patent drawing

AI summary

A computer program product, including a non-transitory, computer-readable medium containing instructions therein which, when executed by at least one processor, cause the at least one processor to perform a performance analysis of a network of interconnected nodes, the nodes configured to perform corresponding logic functions. The performance analysis includes, for a pipeline node in the network, calculating a pre-charging finish time of the pipeline node based on an evaluation finish time of a fanout node of the pipeline node and an acknowledge output time parameter of the fanout node. The performance analysis further includes, for the pipeline node in the network, calculating a cycle time of the pipeline node based on the calculated pre-charging finish time and an evaluation finish time of a fanin node of the pipeline node.