Read Disturb Scanning With Selective Concurrent iWL Reads
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Memory sub-systems face inefficiencies in performing read disturb scans due to high latency and throughput issues, particularly when concurrent read operations lead to read collisions and data retention sensitivity, causing unnecessary throttling and performance degradation.
Innovation Solution
The system selectively determines whether to perform read disturb scans as independent wordline (iWL) or non-independent wordline (non-iWL) operations based on data retention sensitivity, prioritizing iWL operations where possible to avoid collisions and increase throughput, while ensuring accurate scan results by using different read settings for sensitive data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If concurrent read operations are performed to increase throughput, then host read command throughput is improved, but read collisions occur causing data retention sensitivity issues
Solution Approach 1:
The patent applies local quality by differentiating between sensitive and non-sensitive data regions. The system identifies specific wordlines as sensitive based on their physical location and characteristics, then applies different read operation modes (concurrent vs. sequential) to different regions. This allows concurrent operations in non-sensitive regions while avoiding collisions in sensitive regions, thus maintaining both throughput and data integrity.
Solution Approach 2:
The patent segments the memory array into sensitive and non-sensitive wordlines based on their physical characteristics and retention properties. By dividing the memory structure into these distinct segments, the system can independently manage read operations for each segment, performing concurrent reads in non-sensitive segments while using sequential reads in sensitive segments to prevent collisions.
2Reliability
If read disturb scans are performed frequently to ensure data integrity, then reliability is improved, but latency increases and throughput decreases
Solution Approach 1:
The patent performs preliminary identification of sensitive wordlines during memory initialization or idle periods. By pre-characterizing which wordlines are sensitive and storing this information in a lookup table, the system avoids the need for exhaustive scans of all wordlines during normal operation. The read disturb scan is performed only on identified sensitive wordlines, significantly reducing the frequency and duration of scanning operations.
Solution Approach 2:
Instead of performing read disturb scans on all wordlines (excessive action), the patent applies partial action by scanning only the subset of wordlines identified as sensitive. This selective scanning approach maintains data integrity for critical regions while avoiding unnecessary scans in non-sensitive regions, thereby improving overall throughput.
3Reliability
If sequential read operations are used to avoid read collisions, then data retention sensitivity is maintained, but throughput is reduced
Solution Approach 1:
The patent dynamically selects between concurrent and sequential read operation modes based on real-time conditions. The system monitors which wordlines are currently active and determines whether concurrent or sequential operations are safe. When non-sensitive wordlines are accessed, concurrent operations are enabled to maximize throughput. When sensitive wordlines are detected, the system switches to sequential operations to prevent collisions, thus adapting the operation mode to current workload conditions.
Data Source
AI summary
A system includes a memory device having planes. A processing device is operatively coupled to the memory device. The processing device initiates a background scan in response to receiving host read command(s) directed at one or more pages across the planes. The processing devices determines, based on one or more attribute values indicating a data retention insensitivity of a first page of the one or more pages, that a first scan read operation, with respect to the first page, is allowed to be performed as an independent wordline (iWL) read operation. The processing device, in response to the determining, transmits a first scan read command to cause the memory device to perform the first scan read operation as the iWL read operation.