DRAM Tag-Integrated Cache for Lower SCM Access Latency

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing DRAM systems face challenges in scaling and latency when combined with storage class memory (SCM) due to the longer latency and limited endurance of SCM, leading to inefficiencies in hybrid memory systems.

Innovation Solution

Modifying DRAM architecture to enable tag storage and comparison on the DRAM, allowing it to function as a multiway set associative cache, reducing latency and power overhead by performing tag matching within the DRAM rather than external circuits.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If SCM is used to increase memory capacity, then storage capacity is improved, but access latency increases

Engineering Contradiction:
Improvememory capacityVSAvoidaccess latency
Core Design Contradiction:
Quantity of substanceVSLoss of time

Solution Approach 1:

The memory system is segmented into two distinct parts: DRAM cache for fast access and SCM for large capacity storage. The DRAM cache holds frequently accessed data and tags, while SCM provides bulk storage. This segmentation allows the system to leverage the speed of DRAM for critical operations while maintaining the high capacity of SCM.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The DRAM cache acts as an intermediary between the processor and SCM. By storing copies of SCM data in DRAM along with associated tags, the system can serve requests from the fast DRAM cache when possible, reducing accesses to slow SCM. The tag storage within DRAM enables efficient cache hit/miss determination without external circuitry.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Adaptability or versatility

If external tag matching circuits are used, then cache functionality is achieved, but device complexity and power consumption increase

Engineering Contradiction:
Improvecache functionalityVSAvoidcircuit complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The tag storage and data storage are merged into the same DRAM array structure. Tags are stored in dedicated rows within the DRAM device, eliminating the need for separate external tag matching circuits. This integration reduces overall system complexity while maintaining full cache functionality.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The DRAM device performs multiple functions: it stores both data and tags, performs tag matching operations, and provides cache functionality. By making the DRAM device universal and multi-functional, the system eliminates dedicated external circuits for tag matching, reducing device complexity and power consumption.

Inventive Principle:
Principle #6Universality (Multi-functionality)

3Productivity

If DRAM is modified for tag storage, then cache efficiency is improved, but manufacturing complexity increases

Engineering Contradiction:
Improvecache efficiencyVSAvoidmanufacturing complexity
Core Design Contradiction:
ProductivityVSEase of manufacture

Solution Approach 1:

The invention changes the operational parameters of standard DRAM devices by repurposing existing redundant rows for tag storage and modifying the control logic to perform tag matching. This approach allows cache functionality to be achieved through parameter changes rather than fundamental structural modifications, easing manufacturing.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentUS12561251B2Storage and access of data and tags in a multi-way set associative cache
Publication Date: 2026.02.24 RAMBUS INC
  • US12561251B2 patent drawing
  • US12561251B2 patent drawing
  • US12561251B2 patent drawing

AI summary

Disclosed is a dynamic random access memory (DRAM) that includes a plurality of data rows and a plurality of tag rows. The DRAM includes a communication interface to receive a first group of address bits. The DRAM includes one or more comparators to generate a tag match indication and one or more set bits based on the first group of address bits and a first group of tag information bits from the plurality of tag rows. The one or more comparators are further to combine, based on the tag match indication, the one or more set bits and the first group of address bits to generate a second group of address bits.