Cache Buffer Segmentation for Concurrent Request Servicing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing cache systems face inefficiencies in servicing multiple requests while maintaining cache performance, particularly as cache size increases, due to locking issues and inefficient data eviction and retrieval processes.

Innovation Solution

Implementing a cache buffer system with searchable buffers that allow data to be stored and retrieved independently of cache line locking, enabling simultaneous servicing of requests and scaling with cache size, using a combination of hardware, firmware, and software algorithms to manage buffer allocation and masking.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If traditional cache locking mechanisms are used to service requests, then data consistency is maintained, but cache performance degrades and request servicing speed decreases

Engineering Contradiction:
Improvedata consistencyVSAvoidrequest servicing speed
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The cache system is segmented into multiple independent buffers (first buffer, second buffer) that can operate concurrently. By dividing the cache into separate buffer regions with independent access paths, the system eliminates the need for global locking mechanisms, allowing multiple requests to be serviced simultaneously without compromising data consistency.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

A buffer management mechanism acts as an intermediary between the cache and request servicing operations. The controller uses buffer status indicators and linked list structures to mediate access to buffered data, enabling concurrent requests to be handled without direct locking of cache lines, thus maintaining both consistency and performance.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Quantity of substance

If cache size is increased to improve storage capacity, then more data can be cached, but efficiency and performance degrade due to locking issues

Engineering Contradiction:
Improvecache storage capacityVSAvoidcache efficiency
Core Design Contradiction:
Quantity of substanceVSProductivity

Solution Approach 1:

The cache is organized into multiple independent buffers that can be simultaneously accessed. This segmentation allows the cache to scale in size by adding more buffer regions without introducing global locking overhead, as each buffer can be managed independently with its own status indicators and access control.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces a new dimension of buffer management using linked lists and status indicators that operate independently of the traditional cache hierarchy. This additional dimensional approach to organizing and accessing cached data allows efficient management of large cache sizes without the performance penalties associated with traditional locking mechanisms.

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

3Productivity

If data eviction and retrieval processes are simplified to improve speed, then request servicing is faster, but data management flexibility and consistency are compromised

Engineering Contradiction:
Improvedata retrieval speedVSAvoiddata management complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

Data is pre-loaded into buffers and organized in linked list structures before retrieval is needed. The buffer management system maintains status indicators and organized data structures in advance, allowing rapid retrieval operations without complex real-time management decisions, thus achieving both speed and consistency.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The buffer management mechanism uses self-contained linked lists and status indicators that automatically track data location and availability. This self-service approach eliminates the need for complex centralized management logic, enabling fast data retrieval while maintaining data consistency through autonomous buffer state tracking.

Inventive Principle:
Principle #25Self-service

Data Source

PatentEP3676715B1Cache buffer
Publication Date: 2024.03.27 MICRON TECHNOLOGY INC
  • EP3676715B1 patent drawingFigure 1
  • EP3676715B1 patent drawingFigure 2~3
  • EP3676715B1 patent drawingFigure 4

AI summary

The present disclosure includes apparatuses and methods related to a cache buffer. An example apparatus can store data associated with a request in one of a number of buffers and service a subsequent request for data associated with the request using the one of the number of buffers. The subsequent request can be serviced while the request is being serviced by the cache controller.