Reconfigurable Lookup Engine Tiles for Low Latency Router Memory
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current router technologies face inefficiencies in memory lookup operations due to the need for rapid processing of diverse lookup tasks and the limitations of conventional memory architectures, which lead to high power consumption and heat generation, especially in large data centers.
Innovation Solution
A network router architecture utilizing a packet processing engine with intercommunicating computational tiles, each equipped with functional units, a configuration store, and a programmable multi-way switch, allowing for flexible and efficient memory lookups by optimizing the memory topology for different tasks through a tile-based system with high-speed gate-array structures.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If ternary content addressable memories (TCAM) are used for rapid memory lookups, then lookup speed is improved, but power consumption and heat generation increase
Solution Approach 1:
The patent segments the monolithic TCAM lookup operation into multiple stages distributed across a network of interconnected tiles. Each tile performs a portion of the lookup task using smaller, more energy-efficient memory structures, collectively achieving rapid lookup without the prohibitive power consumption of a single large TCAM.
Solution Approach 2:
The patent extracts the memory lookup function from conventional processor-based implementations and from single-architecture TCAM systems. By creating a specialized tile-based architecture with configurable memory structures, it separates the lookup function from general-purpose processing and from fixed memory architectures, enabling optimized energy-efficient lookup operations.
2Speed
If specialized hardware is created for rapid memory lookup, then lookup performance is improved, but adaptability to different lookup tasks deteriorates
Solution Approach 1:
The patent creates a universal tile-based architecture where each tile can be configured to perform different lookup functions. The interconnected tiles can be programmed to implement various memory structures (e.g., TCAM, CAM, RAM, hash tables) and lookup algorithms, allowing the same hardware platform to adapt to different lookup tasks including IP address lookup, Ethernet address lookup, and other network functions.
Solution Approach 2:
The patent implements dynamic reconfigurability in the tile-based architecture, allowing the memory structures and data flow paths to be dynamically adjusted based on the specific lookup task requirements. This enables the system to optimize its structure for different protocols and lookup patterns while maintaining high performance.
3Device complexity
If a fixed memory architecture is used, then device complexity is reduced, but productivity for evolving network tasks deteriorates
Solution Approach 1:
The patent segments the memory architecture into multiple independent tiles, each with its own configurable memory structure. This segmentation allows individual tiles to be optimized for specific tasks while maintaining overall system simplicity through modular design, enabling easy adaptation to new network protocols and lookup requirements.
Data Source
AI summary
An architecture for a specialized electronic computer for high-speed data lookup employs a set of tiles each with independent logic elements lookup memory portions. The tiles may each comprise gate-array-like functional units that may be wired together by a multi-way switch for extremely low latency.


