Hash-Based Memory Read Disturbance Detection for Data Integrity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing memory devices face issues with read disturbance errors, particularly in 3D stacked flash memory, leading to data integrity problems due to over-reads of single logical block addresses, which current solutions either introduce performance degradation or require external storage, failing to effectively address single page read disturbances.
Innovation Solution
A method using hashing-based address mapping and a read count table to monitor read counts, converting addresses to maintain data integrity by identifying and addressing read-disturbance-affected pages, and transferring data to new pages to prevent further damage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If data block refresh is performed on a super block when reading operations reach a threshold, then read disturbance errors are detected, but the SSD frequently performs garbage collection causing performance degradation
Solution Approach 1:
The patent segments the address space into multiple hash tables with different bucket counts (first hash table with first bucket count, second hash table with second bucket count). This segmentation allows fine-grained monitoring of read counts at different resolution levels, enabling the system to detect read disturbances without triggering excessive garbage collection operations, thus resolving the contradiction between detection accuracy and performance.
2Reliability
If data block refresh is performed on a block when reading operations reach a threshold, then read disturbance errors are detected, but the controller IC requires external memory for storing records since internal storage is insufficient
Solution Approach 1:
The patent introduces a hashing dimension to address the storage capacity problem. By using hash tables with multiple buckets, the system can store read count records for many blocks in a compact manner within the controller's internal memory. The hashing mechanism distributes addresses across multiple buckets, allowing efficient use of limited internal storage space without requiring external memory attachments.
3Reliability
If the read threshold is set lower to prevent single page read disturbance, then data integrity is protected, but a certain proportion of useless data refresh occurs
Solution Approach 1:
The patent implements dynamic threshold management where the read count threshold is not fixed but adapts based on actual read patterns. The system monitors read counts across different hash buckets and only triggers refresh operations when genuine read disturbances are detected. This dynamic approach prevents useless refresh operations while maintaining data integrity, resolving the contradiction between protection effectiveness and energy efficiency.
Data Source
AI summary
A method for performing read-disturbance detection of memory device with aid of hashing-based address mapping for data integrity control and associated apparatus are provided. The method may include: establishing a read count table for recording multiple first read counts with respect to multiple first addresses; performing the hashing-based address mapping on a set of second addresses at which reading operations are detected in order to convert the set of second addresses into a set of first addresses, for updating a set of first read counts; monitoring at least a highest first read count to determine whether the highest first read count reaches a first read count threshold; performing reverse mapping on a first address at which the highest first read count is detected in order to convert the first address into a set of target second addresses; and performing a media scan procedure with respect to the target second addresses.