Caching Home Agent Pipeline Architecture for Processor Latency Reduction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current cache management systems in multicore processors face inefficiencies in handling cache misses and coherence operations, leading to increased latency and power consumption due to the need for multiple traversals through both the cache and home agent pipelines.

Innovation Solution

The implementation of a caching home agent with a combined cache and home agent pipeline architecture that optimizes processing by receiving and handling completion messages directly within the home agent pipeline, reducing the need for cache pipeline traversal and offloading cache pipeline activity, thereby minimizing latency and power consumption.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of time

If cache misses and coherence operations are handled through separate cache and home agent pipelines, then completeness of processing is ensured, but latency and power consumption increase due to multiple traversals

Engineering Contradiction:
ImprovelatencyVSAvoidpipeline traversal complexity
Core Design Contradiction:
Loss of timeVSDevice complexity

Solution Approach 1:

The patent combines the cache pipeline and home agent pipeline into a unified caching home agent pipeline. This integration allows cache misses and coherence operations to be handled in a single travers al, eliminating the need for separate pipeline passes and reducing latency while maintaining processing completeness

Inventive Principle:
Principle #5Merging (Combining)

2Loss of energy

If completion messages are handled by the cache pipeline, then cache state updates are maintained, but power consumption increases due to unnecessary cache pipeline traversal

Engineering Contradiction:
Improvepower consumptionVSAvoidcache coherence
Core Design Contradiction:
Loss of energyVSReliability

Solution Approach 1:

The patent extracts the handling of completion messages from the cache pipeline and assigns it to the home agent pipeline. This extraction allows completion messages to be processed without traversing the cache pipeline, reducing power consumption while cache coherence is maintained through the home agent's coherence protocol handling

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent implements partial action by having the home agent pipeline handle only the specific task of completion message processing, rather than the cache pipeline handling all cache-related operations. This specialization reduces unnecessary cache pipeline activity and power consumption while maintaining required cache coherence functionality

Inventive Principle:
Principle #16Partial or excessive action

3Productivity

If the cache pipeline handles all cache-related operations, then comprehensive cache management is achieved, but bandwidth and latency are degraded due to pipeline congestion

Engineering Contradiction:
Improvecache bandwidthVSAvoidpipeline operation complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent segments the cache pipeline operations into distinct functional components: the caching home agent pipeline handles cache misses and completion messages, while the cache pipeline focuses on cache hit operations and data forwarding. This segmentation distributes the workload, improving cache bandwidth and reducing pipeline congestion

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP3274860B1A method, apparatus and system for optimizing cache memory transaction handling in a processor
Publication Date: 2019.11.27 INTEL CORP
  • EP3274860B1 patent drawingFigure 1
  • EP3274860B1 patent drawingFigure 2
  • EP3274860B1 patent drawingFigure 3

AI summary

In one embodiment, a processor includes a caching home agent (CHA) coupled to a core and a cache memory and includes a cache controller having a cache pipeline and a home agent having a home agent pipeline. The CHA may: receive, in the home agent pipeline, information from an external agent responsive to a miss for data in the cache memory; issue a global ordering signal from the home agent pipeline to a requester of the data to inform the requester of receipt of the data; and report issuance of the global ordering signal to the cache pipeline, to prevent the cache pipeline from issuance of a global ordering signal to the requester. Other embodiments are described and claimed.