Adaptive Data Shaping in Nonvolatile Memory Blocks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing nonvolatile memory systems face inefficiencies in data storage and retrieval due to unbalanced data distribution, which affects power consumption, programming time, and error rates, as well as inaccurate block health monitoring when using shaped data for characterization.
Innovation Solution
Implementing a method that maintains write-erase cycle counts for blocks, performs data shaping during non-polling cycles to balance logic states, and initiates polling cycles to collect accurate characterization data by disabling shaping, storing unshaped data, and using Error Correction Code (ECC) to correct errors, while transferring data between on-chip and off-chip locations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If data shaping is applied during write-erase cycles, then power consumption and programming time are reduced, but block health monitoring accuracy deteriorates
Solution Approach 1:
The system implements periodic polling cycles at predetermined intervals during which data shaping is temporarily disabled to collect accurate block health characterization data. Between polling cycles, data shaping is re-enabled for normal write operations. This periodic alternation between shaped and unshaped data writing resolves the contradiction by accepting temporary reduction in power efficiency during polling to ensure accurate long-term health monitoring.
2Loss of time
If data shaping is applied during write-erase cycles, then programming time is reduced, but characterization data accuracy deteriorates
Solution Approach 1:
The system implements periodic polling cycles at predetermined intervals during which data shaping is temporarily disabled to collect accurate block health characterization data. Between polling cycles, data shaping is re-enabled for normal write operations. This periodic alternation between shaped and unshaped data writing resolves the contradiction by accepting temporary reduction in programming speed during polling to ensure accurate long-term health monitoring.
3Loss of energy
If data shaping is applied to balance logic states, then power consumption is reduced, but data distribution uniformity worsens during polling cycles
Solution Approach 1:
The system implements periodic polling cycles at predetermined intervals during which data shaping is temporarily disabled to collect accurate block health characterization data. Between polling cycles, data shaping is re-enabled for normal write operations. This periodic alternation between shaped and unshaped data writing resolves the contradiction by accepting temporary non-uniform data distribution during polling to ensure accurate long-term health monitoring.
Data Source
AI summary
A nonvolatile memory block experiences multiple write-erase cycles during which data is subject to a shaping operation prior to storage. In response to a write-erase cycle count reaching a predetermined number, a polling cycle occurs during which shaping is disabled and data is collected that indicates a condition of the block. Subsequently, shaping is reenabled for subsequent cycles.