Tier Control Units for Cache Coherency

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional cache coherency systems in multi-controller architectures face issues with increased wire lengths, circuit area, power consumption, and timing congestion due to the use of global interconnects for maintaining cache coherency across multiple domains, which limits performance and increases costs.

Innovation Solution

The implementation of tier control units (TCUs) that maintain cache coherency by operatively coupling to local interconnects within each domain, allowing for a tiered coherency structure that reduces circuit area, power consumption, and bandwidth requirements, while supporting a tiered architecture that efficiently handles cache line requests and responses.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If global interconnects are used to maintain cache coherency across multiple domains, then cache coherency can be maintained, but wire lengths increase, circuit area increases, power consumption increases, and timing congestion occurs

Engineering Contradiction:
Improvecache coherencyVSAvoidcircuit area
Core Design Contradiction:
ReliabilityVSArea of stationary object

Solution Approach 1:

The system segments the cache coherency management into hierarchical domains with local interconnects handling domain-level coherency and global interconnects handling cross-domain coherency. This segmentation reduces the burden on global interconnects and allows for more efficient local management, thereby reducing overall wire lengths and circuit area while maintaining coherency.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces a hierarchical dimension to cache coherency management by organizing controllers into domains with local interconnects, adding a layer between individual controllers and the global interconnect. This dimensional change reduces the direct connectivity requirements and allows for more compact circuit design with reduced area.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Reliability

If global interconnects are used to maintain cache coherency across multiple domains, then cache coherency can be maintained, but wire lengths increase, circuit area increases, power consumption increases, and timing congestion occurs

Engineering Contradiction:
Improvecache coherencyVSAvoidwire lengths
Core Design Contradiction:
ReliabilityVSLength of stationary object

Solution Approach 1:

The system segments the cache coherency management into hierarchical domains with local interconnects handling domain-level coherency and global interconnects handling cross-domain coherency. This segmentation reduces the burden on global interconnects and allows for more efficient local management, thereby reducing overall wire lengths and circuit area while maintaining coherency.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Local interconnects act as intermediaries between controllers within domains and the global interconnect. This intermediary layer reduces the need for long direct wire connections between all controllers and the global interconnect, thereby reducing overall wire lengths and timing congestion.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Reliability

If global interconnects are used to maintain cache coherency across multiple domains, then cache coherency can be maintained, but power consumption increases

Engineering Contradiction:
Improvecache coherencyVSAvoidpower consumption
Core Design Contradiction:
ReliabilityVSUse of energy by stationary object

Solution Approach 1:

The system segments the cache coherency management into hierarchical domains with local interconnects handling domain-level coherency and global interconnects handling cross-domain coherency. This segmentation reduces the burden on global interconnects and allows for more efficient local management, thereby reducing overall wire lengths and circuit area while maintaining coherency.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent implements local coherency management within domains using local interconnects, allowing each domain to manage its own coherency requirements independently. This local quality approach reduces the need for frequent global interconnect transactions, thereby reducing overall power consumption.

Inventive Principle:
Principle #3Local quality

4Reliability

If global interconnects are used to maintain cache coherency across multiple domains, then cache coherency can be maintained, but timing congestion occurs

Engineering Contradiction:
Improvecache coherencyVSAvoidtiming congestion
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The system segments the cache coherency management into hierarchical domains with local interconnects handling domain-level coherency and global interconnects handling cross-domain coherency. This segmentation reduces the burden on global interconnects and allows for more efficient local management, thereby reducing overall wire lengths and circuit area while maintaining coherency.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Local interconnects act as intermediaries between controllers within domains and the global interconnect. This intermediary layer reduces the need for long direct wire connections between all controllers and the global interconnect, thereby reducing overall wire lengths and timing congestion.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentUS11604731B2Systems and methods for maintaining cache coherency
Publication Date: 2023.03.14 SEAGATE TECH LLC
  • US11604731B2 patent drawing
  • US11604731B2 patent drawing
  • US11604731B2 patent drawing

AI summary

Cache coherency of a global address space of a cache can be maintained with one or more tier control units (TCUs). The global address space of the cache may be shared by multiple domains. Domains may include multiple controllers and a local interconnect operatively coupling the controllers to the cache. The local interconnect of each domain may maintain a cache coherency of a local address space of the cache shared by the controllers of the domain. The one or more TCUs may be operatively coupled to the local interconnects of the domains to maintain the cache coherency of the global address space.