DRAM Cache Tag Probing for Bandwidth Optimization

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current memory systems face inefficiencies in data caching and cache coherence management, particularly in multi-processor environments, where cache snooping operations can lead to unnecessary bandwidth usage and resource conflicts.

Innovation Solution

A dynamic random access memory (DRAM) device is configured to handle cache data access and probe commands, utilizing tag comparison circuitry to determine cache line presence and status without transferring data, thereby optimizing bandwidth usage and reducing unnecessary snooping operations through snoop-filtering functions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If cache snooping operations are performed to maintain cache coherence in multi-processor environments, then cache coherence management is improved, but bandwidth consumption increases and resource conflicts occur

Engineering Contradiction:
Improvecache coherence managementVSAvoidbandwidth consumption
Core Design Contradiction:
ReliabilityVSLoss of energy

Solution Approach 1:

The patent extracts the tag comparison function from the data path, allowing cache tag probing to occur independently without requiring data bus usage. This separates the coherence check operation from data transfer operations, enabling snooping without full cache line access and thereby reducing bandwidth consumption while maintaining coherence management.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent introduces a dedicated tag comparison mechanism that acts as an intermediary between cache coherence management and the data path. This intermediary allows processors to probe cache tags and determine presence/absence of cache lines without triggering full data access, thus maintaining coherence while minimizing bandwidth usage.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Loss of energy

If cache probe commands are implemented to check cache line presence without data transfer, then bandwidth consumption is reduced, but device complexity increases

Engineering Contradiction:
Improvebandwidth consumptionVSAvoiddevice complexity
Core Design Contradiction:
Loss of energyVSDevice complexity

Solution Approach 1:

The patent segments the cache access operation into two distinct functions: tag comparison (for presence detection) and data access (for actual data transfer). By separating these functions, the system can perform lightweight probe commands that only activate tag comparison logic without engaging the full data path, thus reducing bandwidth consumption while adding minimal complexity through modular design.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent implements preliminary tag comparison before data access operations. The probe command performs the tag matching step in advance to determine whether a cache line is present, allowing the system to avoid unnecessary data transfers. This preliminary action reduces bandwidth consumption while the added complexity is confined to the control logic for handling probe commands.

Inventive Principle:
Principle #10Preliminary action

3Device complexity

If traditional cache access commands are used for both data access and presence checking, then device complexity is reduced, but bandwidth consumption increases due to unnecessary data transfers

Engineering Contradiction:
Improvedevice complexityVSAvoidbandwidth consumption
Core Design Contradiction:
Device complexityVSLoss of energy

Solution Approach 1:

The patent makes the cache access mechanism dynamic by introducing different command types (probe commands vs. data access commands) that can be selected based on the actual need. When only presence checking is required, the dynamic probe command is used to activate only tag comparison. When data transfer is needed, the full data access command is used. This dynamic behavior optimizes bandwidth consumption while maintaining reasonable device complexity through command-based control.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentUS20240394195A1Dram cache tag probing
Publication Date: 2024.11.28 RAMBUS INC
  • US20240394195A1 patent drawing
  • US20240394195A1 patent drawing
  • US20240394195A1 patent drawing

AI summary

A dynamic random access memory (DRAM) device includes functions configured to aid with operating the DRAM device as part of data caching functions. The DRAM is configured to respond to at least two types of commands. A first type of command (cache data access command) seeks to access a cache line of data, if present in the DRAM cache. A second type of command (cache probe command) seeks to determine whether a cache line of data is present, but is not requesting the data be returned in response. In response to these types of access commands, the DRAM device is configured to receive cache tag query values and to compare stored cache tag values with the cache tag query values. A hit/miss (HM) interface/bus may indicate the result of the cache tag compare and stored cache line status bits to a controller.