GPU-Accelerated Optical Proximity Correction for Semiconductor Mask Design
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing complexity and computational intensity of optical proximity correction (OPC) processes in semiconductor manufacturing lead to significant delays in chip design and production, as current CPU processing power cannot keep pace with the growing computational demands, especially with the need for multiple iterations and larger chip designs.
Innovation Solution
The use of specialized processing units such as GPUs, CPUs, and other hardware configurations to execute OPC algorithms, allowing for the distribution of computational tasks and leveraging parallel processing to improve efficiency and reduce processing time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Power
If CPU processing power is increased to handle OPC calculations, then computational capability improves, but cost and power consumption increase significantly
Solution Approach 1:
The patent segments the OPC computational workload into multiple independent tasks that can be distributed across multiple GPUs. Each GPU processes a portion of the mask features or pattern data independently, allowing parallel execution without requiring a single high-power CPU to handle the entire computation sequentially or with limited parallelism.
Solution Approach 2:
The patent replaces the traditional CPU-based computational system with a GPU-based system. GPUs are specifically designed for parallel processing of computationally intensive tasks and consume less power per unit of computational work compared to high-performance CPUs, thus substituting the mechanical/computational approach while reducing energy consumption.
2Measurement precision
If more computational resources are allocated to OPC, then processing accuracy improves, but processing time increases
Solution Approach 1:
The computational domain is segmented into multiple independent regions or feature sets that can be processed simultaneously. By dividing the mask data into chunks assigned to different GPUs, the system achieves both high accuracy (through adequate computational resources per task) and short processing time (through parallel execution of multiple tasks simultaneously).
Solution Approach 2:
The patent introduces parallelism as an additional dimension for processing. Instead of sequentially increasing computational resources to improve accuracy, the system adds a temporal dimension by executing multiple calculations simultaneously across multiple GPUs, thereby achieving both accuracy and speed through dimensional expansion of the computational architecture.
3Manufacturing precision
If OPC is applied to more features and regions, then pattern fidelity improves, but computational complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the entire mask pattern into smaller feature groups or regions that can be processed independently. This allows OPC to be applied comprehensively across all features (improving pattern fidelity) while managing computational complexity through modular processing where each segment is handled by dedicated computational units (GPUs) working in parallel.
Solution Approach 2:
The patent implements partial action by applying OPC selectively to different feature types or regions based on their importance and sensitivity to optical effects. Not all features require the same level of correction, so the system applies appropriate computational effort to each region, achieving sufficient pattern fidelity without unnecessarily increasing overall computational complexity.
4Ease of manufacture
If traditional CPU-based OPC processing is used, then implementation simplicity is maintained, but productivity decreases
Solution Approach 1:
The patent employs GPUs which are universal parallel processing units capable of handling various OPC algorithms and workloads. The same GPU hardware can execute different OPC methods (such as rule-based, model-based, or iterative optimization approaches) by loading appropriate software routines, thus maintaining implementation flexibility while dramatically increasing processing throughput compared to traditional CPU systems.
Data Source
AI summary
Optical proximity correction techniques performed on one or more graphics processors improve the masks used for the printing of microelectronic circuit designs. Execution of OPC techniques on hardware or software platforms utilizing graphics processing units. GPUs may share the computation load with the system CPUs to efficiently and effectively execute the OPC method steps.