NAND Flash Read Threshold Linear Approximation Storage
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
NAND Flash storage systems face inefficiencies in storing read thresholds, leading to a poor tradeoff between memory requirements and the quality of returned values, as existing methods require laborious multiple reads to find or estimate optimal read thresholds for each page.
Innovation Solution
The solution involves storing a single slope and y-intercept for all pages in a block, allowing for the generation of read thresholds using linear approximation, which reduces memory requirements while maintaining high read value quality, and includes processes for partial updates to adapt to changes in program/erase cycles and voltage distributions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If read thresholds are stored for each page, then read value quality is improved, but memory requirements increase
Solution Approach 1:
The patent transforms the storage approach from storing complete read threshold values to storing only slope and y-intercept parameters. By representing read thresholds through linear equation parameters (y = mx + b), the system reduces the storage footprint while maintaining the ability to reconstruct accurate read threshold values for each page based on the page number and stored parameters.
Solution Approach 2:
Instead of storing actual read threshold values for each page, the patent stores a mathematical model (slope and y-intercept) that can generate read threshold values on-demand. This copying approach uses a compact representation that can be replicated to produce the full set of read thresholds when needed, significantly reducing storage requirements.
2Measurement precision
If multiple reads are performed to find optimal read thresholds, then read value quality is improved, but time consumption increases
Solution Approach 1:
The patent performs preliminary action by storing the slope and y-intercept values in advance during programming operations. These pre-stored parameters enable rapid generation of read thresholds during read operations without requiring multiple iterative reads to determine optimal threshold values, thus reducing time consumption while maintaining read accuracy.
Solution Approach 2:
The system uses the stored slope and y-intercept parameters to self-generate appropriate read thresholds for each page based on the page number. This self-service mechanism eliminates the need for external intervention or multiple read attempts to find optimal thresholds, as the system automatically computes the correct threshold using the linear approximation formula.
3Quantity of substance
If a single read threshold is used for all pages, then memory requirements are reduced, but read value quality deteriorates
Solution Approach 1:
The patent implements local quality by allowing each page to have its own optimized read threshold derived from the linear approximation. While the slope and y-intercept are stored once for the block, the generated read thresholds vary locally for each page based on its page number, ensuring that each page receives a tailored read threshold that accounts for local variations in voltage distribution.
Data Source
AI summary
A first read threshold associated with a first page in a block and a second read threshold associated with a second page in the block are received, where the first page has a first page number and the second page has a second page number. A slope and a y intercept are determined based at least in part on the first read threshold, the second read threshold, the first page number, and the second page number. The slope and the y intercept are stored with a block identifier associated with the block.