NAND Memory Read Retry Table Segmentation for Performance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
As storage capacity in NAND memory increases, issues such as poor data retention capability and read disturbance lead to reduced fitness of read retry tables, increasing workload and reducing read performance.
Innovation Solution
The memory system organizes read retry tables into subsets corresponding to different types of pages, allowing for targeted polling based on affected pages due to threshold voltage shifts, improving efficiency and fitness of read retry operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If read retry tables are used to handle threshold voltage shifts in NAND memory, then data retention capability is improved, but device complexity increases due to the need to maintain and poll multiple read retry tables
Solution Approach 1:
The patent segments the read retry table into multiple subsets, where each subset corresponds to a specific page type (e.g., SLC pages, MLC pages, TLC pages). This segmentation allows the system to selectively poll only the relevant subset based on the page type being read, rather than maintaining and polling a single comprehensive read retry table. The segmentation reduces the complexity of read retry table management while maintaining the ability to handle threshold voltage shifts for different page types.
2Productivity
If comprehensive read retry tables are polled for all page types, then read performance is maintained, but workload increases due to unnecessary polling of irrelevant page types
Solution Approach 1:
The patent applies local quality by making the read retry table structure adaptive to the specific page type being accessed. Each page type (SLC, MLC, TLC) has its own dedicated subset of read retry tables with parameters optimized for that specific page type. When a read operation is performed, the system selectively polls only the subset corresponding to the page type, avoiding the time waste of polling subsets for other page types. This localized approach maintains read performance while reducing the overall workload.
3Reliability
If read retry operations are performed with high polling frequency, then fitness of read retry tables is improved, but energy consumption increases
Solution Approach 1:
The patent implements a dynamic read retry mechanism where the polling frequency and selection of read retry tables are adjusted based on the specific page type and read operation requirements. Instead of uniformly polling all read retry tables at high frequency, the system dynamically selects and polls only the relevant subset corresponding to the page type. This dynamic approach maintains the fitness of read retry tables for each page type while significantly reducing energy consumption by avoiding unnecessary polling of irrelevant tables.
Data Source
AI summary
In one example, the memory system includes a memory coupled to a memory controller; the memory includes a plurality of word lines and a plurality of multi-bit memory cells coupled to the plurality of word lines; the memory includes multiple types of pages, each corresponding to a one read voltage level. In one example, the operation method includes obtaining, when a read operation of the memory fails, a target read retry table from a set of read retry tables, the set of read retry tables comprises all read retry tables corresponding to the multiple types of pages; a read retry table corresponding to each type of page comprises a set of bias voltages corresponding to respective read voltage levels that are used to distinguish stored data of corresponding bit; obtaining read retry voltages through the target read retry table, and performing a read retry operation using the read retry voltages.