Parallel Layout File Decompression for IC Mask Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing IC manufacturing process is hindered by the time-consuming and resource-intensive decompression and recompression of large layout files during the import and merging of layout files in the mask house, which increases processing time and resource utilization.
Innovation Solution
The method involves using a host computing system to create sub-threads that decompress and combine layout information in parallel, eliminating the need for redundant decompression and compression steps, thereby directly appending processed files from computing nodes to obtain a single corrected layout file.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If layout files are decompressed and recompressed during import and merging operations, then the layout information can be processed and combined, but the processing time and resource utilization increase significantly
Solution Approach 1:
The layout file is pre-compressed using a reversible compression algorithm before being distributed to computing nodes. This preliminary compression reduces the initial file size for transmission and storage, while the reversibility ensures that decompression can be efficiently performed only when needed, rather than continuously during processing operations.
Solution Approach 2:
The layout file processing is divided into multiple computing nodes that work in parallel. Each node receives a portion of the compressed layout data, performs local processing, and contributes to the final merged result. This segmentation eliminates the need for sequential decompression and recompression operations, significantly reducing processing time.
2Ease of manufacture
If layout files are decompressed and recompressed during import and merging operations, then the layout information can be processed and combined, but resource utilization increases
Solution Approach 1:
The layout file is pre-compressed using a reversible compression algorithm before being distributed to computing nodes. This preliminary compression reduces the initial file size for transmission and storage, while the reversibility ensures that decompression can be efficiently performed only when needed, rather than continuously during processing operations.
Solution Approach 2:
The layout file processing is divided into multiple computing nodes that work in parallel. Each node receives a portion of the compressed layout data, performs local processing, and contributes to the final merged result. This segmentation eliminates the need for sequential decompression and recompression operations, significantly reducing processing time.
3Ease of manufacture
If traditional import and merging methods are used, then layout files can be processed, but the manufacturing and mask production speed decreases
Solution Approach 1:
The layout file is pre-compressed using a reversible compression algorithm before being distributed to computing nodes. This preliminary compression reduces the initial file size for transmission and storage, while the reversibility ensures that decompression can be efficiently performed only when needed, rather than continuously during processing operations.
Solution Approach 2:
The layout file processing is divided into multiple computing nodes that work in parallel. Each node receives a portion of the compressed layout data, performs local processing, and contributes to the final merged result. This segmentation eliminates the need for sequential decompression and recompression operations, significantly reducing processing time.
Data Source
AI summary
A computer-implemented method includes executing, using a computer, a process including a main thread that receives a layout file. The layout file includes a first plurality of tags and compressed information blocks. Each tag of the first plurality is associated with a compressed information block. The method further includes decompressing the compressed information blocks using sub-threads and thereby obtaining decompressed information blocks. The sub-threads are created by the main thread, and each sub-thread corresponds to a compressed information block. The decompressed information blocks are combined into decompressed layout information. The decompressed file is partitioned and each partition is provided to a node of a distributed computing system for performing layout correction. Multiple result files each in a compressed format are obtained from the distributed computing system and the result files are combined to obtain a single result file without decompressing and re-compressing the results from the distributed computing system.


