TCAM Memory Array for Large Search Word Similarity Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems face inefficiencies in processing large data sets due to the need for substantial data movement and high power consumption when transferring data between nonvolatile memory and digital processing systems, limiting their throughput and efficiency.
Innovation Solution
A ternary content addressable memory (TCAM) technology that supports very large search words by using a memory array with TCAM cells configured to store ternary states, allowing for similarity matching rather than exact matches, which reduces data transfer requirements and improves processing efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If data is transferred from nonvolatile memory to digital processing system for processing, then processing capability is improved, but data transfer time and power consumption increase
Solution Approach 1:
The patent combines the storage function (nonvolatile memory) and processing function (TCAM similarity matching) into a single integrated system. The TCAM is built directly within the nonvolatile memory array, allowing similarity matching operations to be performed in-place without transferring data to external digital processing systems, thus eliminating data transfer time while maintaining processing capability
Solution Approach 2:
The patent introduces TCAM as an intermediary component between nonvolatile memory and digital processing systems. The TCAM performs preliminary similarity matching filtering on stored data, identifying only those data items that match the search criteria within a specified mismatch threshold. This intermediary processing step reduces the volume of data that needs to be transferred to digital systems, thereby reducing data transfer time
2Productivity
If data is transferred from nonvolatile memory to digital processing system for processing, then processing capability is improved, but power consumption increases
Solution Approach 1:
The patent combines the storage function (nonvolatile memory) and processing function (TCAM similarity matching) into a single integrated system. The TCAM is built directly within the nonvolatile memory array, allowing similarity matching operations to be performed in-place without transferring data to external digital processing systems, thus eliminating data transfer time while maintaining processing capability
Solution Approach 2:
The patent introduces TCAM as an intermediary component between nonvolatile memory and digital processing systems. The TCAM performs preliminary similarity matching filtering on stored data, identifying only those data items that match the search criteria within a specified mismatch threshold. This intermediary processing step reduces the volume of data that needs to be transferred to digital systems, thereby reducing data transfer time
3Measurement precision
If exact match is required for search operations, then search precision is improved, but data transfer volume increases
Solution Approach 1:
The patent applies partial matching instead of requiring complete exact matches. The TCAM allows searches with a specified number of mismatches (threshold), returning data items that partially match the search criteria. This partial matching approach maintains sufficient search precision for many applications while dramatically reducing the number of data items that need to be transferred for further processing
Data Source
AI summary
A memory array is arranged to store data words in respective sets of TCAM cells, where each TCAM cell is configured to store ternary states of a bit of the stored word. A circuit to select a set of TCAM cells in the set of TCAM cells, such as decoders and drivers for word lines, bit lines, block select gates. A circuit to apply an input search word to the TCAM cells in the selected set of TCAM cells, such as a search word buffer or driver on one of word lines or bit lines for the array. A circuit to generate an output indicating similarity of the stored word in the selected set of TCAM cells to the input search word, based on mismatch or possible mismatch of more than one bit of the search word.


