Differentiable content-addressable memory
The memristor-based aCAM addresses the limitations of conventional CAMs by storing analog values and using a learning algorithm to optimize operations, achieving efficient and accurate low-power operation for enhanced functionality.
Patent Information
- Authority / Receiving Office
- DE · DE
- Patent Type
- Patents
- Current Assignee / Owner
- HEWLETT PACKARD ENTERPRISE DEV LP
- Filing Date
- 2022-10-17
- Publication Date
- 2026-07-02
AI Technical Summary
Conventional content-addressable memories (CAMs) are large, power-consuming, and expensive, limiting their use to specific applications where their performance, efficiency, and speed are desirable, and they lack scalability due to non-ideal circuitry and memory components that cause errors in large arrays.
A memristor-based analog CAM (aCAM) that operates with multi-level voltages and stores analog values, reducing the need for digital conversion and enabling low-power operation with higher accuracy by using memristors to define resistance ranges for storing values between Rhigh and Rlow, and incorporating a learning algorithm to optimize operational goals.
The aCAM achieves significant area and power savings, enabling applications in general calculations and novel scenarios by eliminating the need for analog-to-digital conversion and reducing operational errors through differential operation.
Smart Images
Figure 00000000_0000_ABST