Data Search Apparatus Using Address Transformation and Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data search methods in hardware/software co-design face challenges with high complexity and power consumption in Content Addressable Memory (CAM) and inefficiency in conventional Linear Search, necessitating a more efficient search solution.
Innovation Solution
A data search apparatus and method that partitions memory into multiple banks using an address transformation procedure, allowing for a search complexity of O(log N) by utilizing a comparator-based search module and address transformation, reducing the number of logic units required.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If Content Addressable Memory (CAM) is used for data search, then search time complexity is O(1), but hardware cost and power consumption increase significantly
Solution Approach 1:
The patent divides the memory into multiple banks (e.g., 4 banks) and uses multiple comparators to search parallel segments. This segmentation allows the system to achieve O(log N) search complexity without requiring a single large CAM structure, thereby reducing hardware cost while maintaining efficient search performance.
Solution Approach 2:
The patent uses multiple comparators that replicate the search function across different memory banks. Instead of using a single complex CAM unit, multiple simpler comparator units work in parallel, achieving similar search efficiency with reduced individual component complexity and lower overall hardware cost.
2Device complexity
If conventional Linear Search is used, then hardware cost is reduced, but search time complexity increases to O(N)
Solution Approach 1:
The patent performs preliminary sorting of data in memory and pre-organizes memory addresses through transformation procedures. This preliminary organization enables the search algorithm to efficiently narrow down search spaces using binary search logic, achieving O(log N) complexity without requiring complex hardware structures.
Solution Approach 2:
The patent introduces address transformation that maps logical addresses to physical memory locations in a transformed dimension. By organizing data in sorted order and transforming addresses, the system enables logarithmic search complexity using simple comparator hardware, effectively adding an organizational dimension without increasing hardware complexity.
3Loss of time
If Binary Search is used with 64 data entries, then search rounds are reduced to 6, but memory access complexity increases
Solution Approach 1:
The patent divides 64 data entries into 4 memory banks with 16 entries each, and uses 4 comparators to search parallel segments. This reduces the search rounds from 6 (sequential binary search) to 3 (parallel search), while distributing memory access complexity across multiple simpler bank-access operations rather than one complex sequential access.
Solution Approach 2:
The patent combines multiple comparator operations and memory bank accesses into a single synchronized search cycle. By merging the search functions of multiple comparators working in parallel, the system achieves reduced search rounds while the overall memory access complexity is managed through coordinated bank access rather than complex sequential access patterns.
Data Source
AI summary
A data search apparatus and method are disclosed for searching for a target address of a target data in a memory. The data search apparatus includes a data sort module, an address assignment module, an address transformation module, and at least one comparative module.The data sort module sorts a plurality of data in said memory. The address assignment module assigns an address to each of said plurality of data. The address transformation module transforms said address into a new address according to an address transformation procedure. And the at least one comparative module obtains a portion of bits of said target address of said target data according to a comparative data and said target data. Accordingly, the time complexity for data search is then reduced.