DRAM Controller Target Refresh Row Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Dynamic Random Access Memory (DRAM) refreshing processes face challenges of energy consumption and data access delay due to unnecessary refreshing and long data storing times, leading to inefficiencies and delays in data access.
Innovation Solution
A method where a controller determines a target refreshing row based on data storage duration, refreshing only the data rows with storage durations shorter than a preset duration while allowing longer duration rows to remain unrefreshed, thereby reducing energy consumption and minimizing data access delays.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all basic storage units are refreshed with the lowest refreshing period to ensure data accuracy for units with short storage time, then data accuracy is improved, but energy consumption increases hugely and data access delay occurs
Solution Approach 1:
The patent segments the basic storage units into different groups based on their data storage times. Instead of uniformly refreshing all storage units, it divides them into first basic storage units (with storage time ≤ first threshold) and second basic storage units (with storage time > first threshold), applying different refreshing periods to each group. This segmentation allows the system to refresh only the necessary portions, reducing energy consumption while maintaining data accuracy for units that require it.
Solution Approach 2:
The patent applies local quality by assigning different refreshing characteristics to different regions of the memory based on their specific needs. Storage units with shorter data retention requirements receive more frequent refreshing (lower refreshing period), while units with longer retention times receive less frequent refreshing (higher refreshing period). This localized approach ensures that each storage unit receives the appropriate level of maintenance without unnecessarily consuming energy across the entire memory array.
2Reliability
If DRAM refreshing is executed before data access to ensure data accuracy, then data accuracy is improved, but data access delay occurs
Solution Approach 1:
The patent implements dynamic refreshing by adjusting the refreshing period based on the data storage time characteristics of different storage units. Rather than using a fixed, conservative refreshing schedule that causes delays, the system dynamically determines appropriate refreshing intervals for different groups of storage units. This allows data access to proceed without unnecessary waiting periods while still maintaining data accuracy through targeted refreshing of units that actually require it.
Solution Approach 2:
The patent applies partial action by refreshing only the necessary portion of storage units at any given time. Instead of refreshing all storage units with the lowest (most frequent) period, it refreshes only the first basic storage units that have shorter data storage times. This partial refreshing approach eliminates unnecessary delays in data access while maintaining data accuracy for the units that require frequent refreshing.
3Device complexity
If a uniform refreshing period is applied to all basic storage units, then implementation simplicity is maintained, but energy consumption increases due to unnecessary refreshing of units with long storage time
Solution Approach 1:
The patent segments storage units into different categories based on their data storage time characteristics, creating distinct groups (first basic storage units and second basic storage units) that can be managed with different refreshing periods. This segmentation enables the system to avoid unnecessary refreshing of storage units with long retention times, significantly reducing energy consumption while adding manageable complexity through category-based control.
4Reliability
If refreshing frequency is increased to accommodate storage units with short data storage time, then data accuracy for short-duration units is improved, but overall system energy consumption increases hugely
Solution Approach 1:
The patent applies local quality by providing different refreshing intensities to different storage units based on their specific requirements. Storage units with short data storage times receive high-frequency refreshing (lower refreshing period) to maintain data accuracy, while storage units with long data storage times receive low-frequency or no refreshing (higher refreshing period). This localized approach eliminates energy waste on units that don't require frequent refreshing while maintaining data accuracy for units that do.
Data Source
AI summary
The disclosure provides a data refreshing method of a memory, a controller of a memory, and a memory. The method includes that a target refreshing row in a data row of the memory is determined according to a running state of the controller, the target refreshing row including at least one data row; and in response to determining that a first refreshing interval expires and the target refreshing row includes a first reserved row, the first reserved row in the target refreshing row is refreshed. The first reserved row is a data row with data storage duration smaller than preset duration in the memory. The first refreshing interval is a refreshing interval of the first reserved row.