Selective Read Disturb Sampling in Memory Subsystems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional memory sub-systems face performance degradation and increased power consumption due to excessive data integrity scans triggered by read disturb, which are not selectively targeted, leading to unnecessary memory management operations and resource allocation issues.
Innovation Solution
Implementing selective read disturb sampling by scanning approximately 50% of the wordlines in a data block, alternating between sets of adjacent wordlines to identify stress levels and determine if a refresh is needed, thereby reducing the sample size and time required for data integrity scans.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional data integrity scans are performed on all wordlines, then measurement precision is improved, but productivity deteriorates
Solution Approach 1:
The patent divides the data block into multiple segments (e.g., 512 wordline groups) and performs scans on only a subset of these segments (e.g., every other segment). This segmentation approach maintains measurement precision by sampling representative portions while improving productivity by reducing the total number of scanned wordlines from 512 to approximately 256 per scan operation.
2Reliability
If conventional data integrity scans are performed on all wordlines, then reliability is improved, but use of energy worsens
Solution Approach 1:
The patent applies partial action by scanning only a portion (e.g., 50%) of the wordlines instead of all wordlines. The selection of alternating wordline groups provides statistically representative sampling that maintains reliability for detecting read disturb conditions while reducing power consumption by half compared to full-block scanning.
3Productivity
If selective read disturb sampling is implemented, then productivity is improved, but measurement precision deteriorates
Solution Approach 1:
The patent implements periodic action by alternating between different scanning patterns in successive scan operations. In one scan, even wordline groups are scanned; in the next scan, odd wordline groups are scanned. This periodic alternation ensures that all wordline groups are periodically measured, maintaining measurement precision while allowing high productivity in individual scan operations.
Solution Approach 2:
The system uses feedback from scan results to dynamically adjust scanning behavior. When read disturb conditions are detected in sampled wordlines, the system can trigger additional scans or refresh operations on affected blocks. This feedback mechanism ensures that measurement precision is maintained by focusing resources on problematic areas while allowing productivity to improve in stable regions.
4Use of energy by moving object
If selective read disturb sampling is implemented, then use of energy is reduced, but reliability worsens
Solution Approach 1:
The patent performs preliminary scanning on a subset of wordlines to identify blocks that require full inspection or refresh. By detecting read disturb conditions in the sampled portion, the system can proactively refresh affected blocks before actual data corruption occurs. This preliminary action maintains reliability by preventing errors while consuming less energy than continuous full-block scanning.
Data Source
AI summary
A processing device in a memory system maintains a counter to track a number of read operations performed on a data block of a memory device and determines that the number of read operations performed on the data block satisfies a first threshold criterion. The processing device further determines whether a number of scan operations performed on the data block satisfies a scan threshold criterion. Responsive to the number of scan operations performed on the data block satisfying the scan threshold criterion, the processing device performs a first data integrity scan to determine one or more first error rates for the data block, each of the one or more first error rates corresponding to a first set of wordlines of the data block, the first set comprising first alternating pairs of adjacent wordlines.