Hybrid IMS CAM Cell Segmentation for Density and Power

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing data search technologies, such as ternary content addressable memory (TCAM) and in-memory searching (IMS), face limitations in memory density, power consumption, and matching flexibility, particularly in achieving high inference accuracy and stability for big data and artificial intelligence applications.

Innovation Solution

A hybrid in-memory search (IMS) content addressable memory (CAM) cell is introduced, comprising a combination of different types of IMS CAM cells, including a single-level cell (SLC) IMAS CAM cell and an analog IMS CAM cell, which store different parts of data and utilize distinct search voltages to enhance matching accuracy and stability.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If conventional TCAM is used for data search, then high parallel searching capability is achieved, but memory density is low and access power is high

Engineering Contradiction:

Improveparallel searching capabilityVSAvoidmemory density

Core Design Contradiction:

ProductivityVSQuantity of substance

Solution Approach 1:

The patent divides the search data into multiple parts and stores them in different IMS CAM cells (first IMS CAM cell stores first part, second IMS CAM cell stores second part). This segmentation allows the system to maintain high parallel searching capability while using lower-density non-volatile memory, thereby improving overall memory density without sacrificing search productivity.

Inventive Principle:

Principle #1Segmentation

2Productivity

If conventional TCAM is used for data search, then high parallel searching capability is achieved, but access power consumption is high

Engineering Contradiction:

Improveparallel searching capabilityVSAvoidaccess power consumption

Core Design Contradiction:

ProductivityVSUse of energy by moving object

Solution Approach 1:

The patent segments search data across multiple IMS CAM cells and uses non-volatile memory technology that consumes less power during access operations. By distributing the search functionality across segmented cells with lower individual power requirements, the system maintains parallel searching capability while reducing overall access power consumption compared to conventional TCAM.

Inventive Principle:

Principle #1Segmentation

3Quantity of substance

If NAND-based IMS is used for exact-matching, then power consumption is reduced through dense memory density, but matching flexibility is limited

Engineering Contradiction:

Improvememory densityVSAvoidmatching flexibility

Core Design Contradiction:

Quantity of substanceVSAdaptability or versatility

Solution Approach 1:

The patent segments storage data into multiple parts stored in different types of IMS CAM cells (first IMS CAM cell and second IMS CAM cell with different characteristics). This segmentation enables the system to perform both exact-matching and approximate-matching operations by selectively using different cell types, thereby maintaining high memory density while restoring matching flexibility.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The hybrid IMS CAM cell structure provides multi-functionality by enabling both exact-matching and approximate-matching operations within the same memory system. The first and second IMS CAM cells can be configured to perform different matching types, making the system universal enough to handle various search requirements while maintaining dense memory architecture.

Inventive Principle:

Principle #6Universality (Multi-functionality)

4Device complexity

If single type of IMS CAM cell is used, then device structure is simple, but inference accuracy and search stability are insufficient

Engineering Contradiction:

Improvecell structure simplicityVSAvoidsearch stability

Core Design Contradiction:

Device complexityVSReliability

Solution Approach 1:

The patent segments the search function across multiple IMS CAM cells with different characteristics (first IMS CAM cell and second IMS CAM cell). This segmentation allows each cell type to contribute different strengths to the search process, improving inference accuracy and search stability while keeping individual cell structures relatively simple.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent creates a hybrid IMS CAM cell structure that combines different types of IMS CAM cells (analogous to composite materials). This composite structure leverages the strengths of each cell type to achieve high inference accuracy and stable searching, while the modular design keeps the overall device complexity manageable.

Inventive Principle:

Principle #40Composite materials

Data Source

PatentUS12159672B2Hybrid IMS CAM cell, memory device and data search method

Publication Date: 2024.12.03 MACRONIX INTERNATIONAL CO LTD

AI summary

A hybrid in-memory search (IMS) content addressable memory (CAM) cell includes: a first IMS CAM cell; and a second IMS CAM cell, coupled to the first IMS CAM cell. The first IMS CAM cell and the second IMS CAM cell are of different types. When the hybrid IMS CAM cell stores a storage data, the first IMS CAM cell stores a first part of the storage data and the second IMS CAM cell stores the storage data or a second part of the storage data.