Flash Memory Read Threshold Tracking for Row-Level Error Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional flash memory systems face challenges in maintaining high read performance due to errors caused by varying stress conditions during programming and reading, leading to increased failure probabilities and reduced decoding capabilities.
Innovation Solution
Implementing a system that dynamically adapts read thresholds based on per-row optimal characterization using a machine learning model, such as a deep neural network (DNN), to predict and adjust voltage thresholds for each read operation, thereby minimizing retry rates and improving read throughput.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional fixed read thresholds are used in flash memory systems, then the system is simple to operate, but read performance deteriorates under varying stress conditions leading to higher error rates
Solution Approach 1:
The patent implements dynamic read threshold adjustment by transitioning from fixed thresholds to adaptive thresholds that change based on stress conditions. The system continuously monitors stress indicators (temperature, humidity, operational cycles) and adjusts read thresholds accordingly, enabling the memory system to maintain optimal performance across varying environmental and operational conditions.
Solution Approach 2:
The patent employs feedback mechanisms where the system monitors read error rates and stress conditions, then uses this information to adjust read thresholds. The feedback loop includes collecting stress condition data, comparing it against threshold criteria, and modifying read parameters to minimize errors, thereby improving reliability through continuous adaptation.
2Reliability
If dynamic threshold adjustment is implemented to improve read performance, then decoding capabilities improve, but the complexity of the read operation increases
Solution Approach 1:
The patent performs preliminary actions by pre-characterizing the memory device under various stress conditions and pre-calculating optimal read thresholds for different operational scenarios. This characterization data is stored and retrieved during read operations, allowing the system to adapt thresholds quickly without performing complex real-time analysis, thus maintaining ease of operation while improving decoding capability.
Solution Approach 2:
The patent changes read operation parameters (voltage thresholds, timing parameters) based on monitored stress conditions. By adjusting these parameters dynamically, the system optimizes decoding capability for current operational conditions while maintaining a streamlined read operation process that automatically adapts without requiring manual intervention.
3Reliability
If per-row optimal thresholds characterization is performed, then read failure probability decreases, but the time required for threshold determination increases
Solution Approach 1:
The patent performs threshold characterization and determination in advance during device manufacturing and initial programming phases. Optimal thresholds for different rows and stress conditions are pre-calculated and stored in lookup tables or databases. During actual read operations, the system simply retrieves the appropriate pre-determined thresholds based on current conditions, significantly reducing the time required for threshold determination while maintaining low read failure probability.
Solution Approach 2:
The patent uses copying by creating replicated threshold datasets that can be quickly retrieved during operations. Instead of performing complex threshold determination algorithms in real-time, the system copies and retrieves pre-computed threshold values from storage structures, enabling fast threshold selection that reduces operational time while maintaining high reliability through accurate, pre-optimized threshold values.
Data Source
AI summary
The present disclosure relates to a flash memory system may include a non-volatile memory and a circuit. The non-volatile memory may include one or more blocks, each block including a plurality of rows of cells. The circuit for performing operations on the non-volatile memory, may obtain a row identifier identifying a row of a target page, among the plurality of rows. The circuit may generate, by a machine learning model, one or more voltage thresholds for a read operation, based on the row identifier. The circuit may perform the read operation on the target page of the non-volatile memory with the one or more voltage thresholds.