Memory Subsystem Data Retention via Targeted Refresh
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Memory sub-systems face data loss events due to temperature fluctuations and power interruptions, which existing active cooling measures cannot mitigate when disconnected from the main power supply, leading to unexpected data loss.
Innovation Solution
A data retention component utilizes an auxiliary power supply to perform targeted refresh operations on non-volatile memory components, conserving power and preventing data loss by predicting impending events and refreshing data on a component-by-component or page-by-page basis during power interruptions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If active cooling measures are used to mitigate temperature fluctuations, then data retention is improved, but the system requires continuous main power supply which is not available during power interruptions
Solution Approach 1:
The memory sub-system is segmented into multiple independent memory components, allowing selective refresh of individual components based on their specific temperature exposure and data retention risk. This enables the system to conserve power by only refreshing necessary portions rather than requiring full-system active cooling
Solution Approach 2:
The system performs preliminary monitoring of temperature characteristics and predicts impending data loss events before they occur. By identifying at-risk memory components in advance, the system can proactively refresh only those specific components during power interruptions, preventing data loss without requiring continuous full-system power
2Reliability
If blanket refresh operations are performed on all memory components, then data retention is improved, but auxiliary power is consumed excessively reducing operational duration
Solution Approach 1:
The refresh operation is applied locally only to specific memory components that are predicted to experience data loss, rather than uniformly to all components. This targeted approach maintains data retention for at-risk data while minimizing auxiliary power consumption by excluding components that do not require refresh
Solution Approach 2:
The system performs partial refresh operations on only the necessary portion of memory components rather than complete refresh of all components. This partial action is sufficient to prevent data loss events while conserving auxiliary power resources for extended operational duration
3Measurement precision
If continuous monitoring of all memory components is performed, then data loss prediction accuracy is improved, but processing resources and power consumption increase
Solution Approach 1:
The monitoring function is segmented to track temperature characteristics of individual memory components or groups rather than treating the entire memory sub-system as a single unit. This enables selective monitoring and prediction for specific components, reducing overall processing requirements and power consumption while maintaining prediction accuracy for at-risk components
Data Source
AI summary
A method includes providing, when delivery of power to a memory sub-system comprising a plurality of non-volatile memory components from a main power supply is interrupted, power from an auxiliary power supply to a portion of a controller interfacing with the non-volatile memory components. The portion of the controller may utilize the power from the auxiliary power to supply to: responsive to a prediction, generated when the delivery of the power from the main power supply to the memory sub-system is interrupted, that a particular portion of the plurality of non-volatile memory components will experience an impending data loss event, perform a targeted refresh operation of data stored in the particular portion.