Built-In Memory Repair Using Merged Defect Data Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Integrated circuits face challenges in achieving high device yield and minimizing overhead as device densities increase and sizes decrease, particularly in managing defective memory instances within digital signal processors.
Innovation Solution
A centralized memory repair system employing compression techniques to convert diverse memory instance data into a unified format, compress repair codes, and store them efficiently in non-volatile memory, enabling efficient repair and reducing memory requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If repair codes are stored for each memory instance separately, then repair accuracy is improved, but memory overhead increases
Solution Approach 1:
The patent merges repair data from multiple memory instances into a single consolidated repair code. Instead of storing separate repair codes for each memory instance, the system combines defect information from multiple instances and generates a unified repair code that can repair all instances simultaneously, thereby reducing memory overhead while maintaining repair accuracy.
Solution Approach 2:
The patent creates a universal repair code that serves multiple memory instances simultaneously. The repair code is designed to be multi-functional, capable of repairing defects across different memory instances with a single code, eliminating the need for instance-specific repair codes and reducing overall storage requirements.
2Productivity
If device density increases and device size decreases, then productivity is improved, but device complexity increases
Solution Approach 1:
The patent merges multiple memory instances and their associated repair mechanisms into a more compact integrated structure. By consolidating repair functionality and sharing resources across memory instances, the system achieves higher device density without proportionally increasing device size, as the repair infrastructure is shared rather than duplicated.
Solution Approach 2:
The patent implements universal repair logic and shared repair resources that can serve multiple memory instances. This multi-functional approach allows a single repair mechanism to handle defects across different memory instances, reducing the overall complexity and size of the repair infrastructure while supporting increased device density.
3Quantity of substance
If repair codes are compressed, then memory overhead is reduced, but repair time increases
Solution Approach 1:
The patent performs preliminary compression of repair codes during the manufacturing or initialization phase. By pre-compressing the repair codes and storing them in a compact format, the system reduces memory overhead without incurring compression overhead during actual repair operations. The compression is done in advance, so when repairs are needed, the pre-compressed codes can be directly applied.
Data Source
AI summary
Circuits and methods are directed to repairable memory systems and memory repair processes. An example circuit includes first and second logic coupled together. The first logic receives a plurality of instances of defect data from a plurality of memories, respectively, in which each of the plurality of instances of defect data has a memory-specific format. The first logic converts each of the plurality of instances of defect data to a common format and merges the plurality instances of defect data in the common format to generate merged data. The second logic receives the merged data and determines a plurality of instances of repair data for the plurality of instances of defect data, respectively, based on the merged data.