DRAM ECC Flag Bits for Selective Syndrome Refresh

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing memory systems face challenges in reducing power consumption during refresh operations in DRAM devices, as current error correction techniques require significant time and power to generate and store syndromes, leading to increased latency and power usage, especially when entering or exiting reduced power refresh modes.

Innovation Solution

The implementation of an ECC method and system that uses a flag bit to determine if data has been modified since the last refresh, allowing for the generation and storage of new syndromes only when necessary, thereby reducing the need for frequent syndrome generation and storage, and enabling efficient data integrity checking and correction during both reduced power refresh and normal operations.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If traditional ECC syndrome generation and storage is performed for every refresh operation, then data integrity is maintained, but power consumption increases significantly

Engineering Contradiction:
Improvedata integrityVSAvoidpower consumption
Core Design Contradiction:
ReliabilityVSUse of energy by moving object

Solution Approach 1:

Instead of performing full ECC syndrome generation and storage for every refresh operation, the patent applies partial action by only generating and storing syndromes when data is actually modified. The system selectively performs ECC operations based on whether write operations have occurred since the last refresh, thereby reducing unnecessary syndrome generation and storage while still maintaining data integrity for modified data.

Inventive Principle:
Principle #16Partial or excessive action

2Use of energy by moving object

If refresh rate is reduced to lower power consumption, then power usage decreases, but data loss risk increases

Engineering Contradiction:
Improvepower consumptionVSAvoiddata retention
Core Design Contradiction:
Use of energy by moving objectVSReliability

Solution Approach 1:

The patent applies preliminary action by generating and storing ECC syndromes in advance when data is written or modified, before the actual refresh operation occurs. This allows the system to use these pre-generated syndromes during reduced-power refresh operations to detect and correct errors, thereby maintaining data retention reliability even when refresh rates are reduced to lower power consumption.

Inventive Principle:
Principle #10Preliminary action

3Reliability

If syndrome generation is performed frequently to ensure data correctness, then data integrity is improved, but latency increases

Engineering Contradiction:
Improvedata correctnessVSAvoidlatency
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent reduces latency by applying partial action - only performing syndrome generation and storage when data modifications are detected via the dirty bit mechanism. This selective approach avoids the time-consuming full syndrome generation for every refresh operation, thereby reducing latency while still ensuring data correctness for modified data through targeted ECC operations.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentUS8601341B2Memory system and method using ECC with flag bit to identify modified data
Publication Date: 2013.12.03 MICRON TECHNOLOGY INC
  • US8601341B2 patent drawing
  • US8601341B2 patent drawing
  • US8601341B2 patent drawing

AI summary

A DRAM device includes an ECC generator/checker that generates ECC syndromes corresponding to items of data stored in the DRAM device. The DRAM device also includes an ECC controller that causes the ECC syndromes to be stored in the DRAM device. The ECC controller also causes a flag bit having a first value to be stored in the DRAM device when a corresponding ECC syndrome is stored. The ECC controller changes the flag bit to a second value whenever the corresponding data bits are modified, this indicating that the stored syndrome no longer corresponds to the stored data. In such case, the ECC controller causes a new ECC syndrome to be generated and stored, and the corresponding flag bit is reset to the first value. The flag bits may be checked in this manner during a reduced power refresh to ensure that the stored syndromes correspond to the stored data.