Phase-Change Memory Bad Row Management via Self-Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Phase-change memory devices reach the end of their life prematurely due to a small number of worn-out rows, leading to reduced capacity and functionality, as existing technologies do not effectively manage or reuse bad rows.
Innovation Solution
A memory system with a nonvolatile memory comprising a main region and a spare region, where the controller writes data to a selected row, determines if it's bad, and if so, writes the data to a spare address in the spare region and updates the bad row with the spare address, allowing data access without requiring additional memory for mapping.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Duration of action of stationary object
If traditional bad row management is used where worn-out rows are simply marked as unusable, then the memory device reaches end of life prematurely, but the capacity utilization is reduced and manufacturing costs increase due to needing external mapping memory
Solution Approach 1:
The patent recovers and reuses bad rows that would traditionally be discarded. When a row is detected as bad during write operations, instead of marking it as permanently unusable, the system writes the data to a spare location and stores the mapping address in the bad row itself, thereby recovering the bad row's utility for storage purposes
Solution Approach 2:
The memory system performs self-service by using its own internal resources (spare regions within the NVM) to manage bad rows, eliminating the need for external mapping memory. The bad rows are converted into self-contained storage units that include their own mapping information
2Ease of operation
If external memory is used to store mapping tables for bad row management, then mapping functionality is provided, but device complexity and manufacturing costs increase
Solution Approach 1:
The patent merges the mapping table functionality with the NVM's existing spare regions. The mapping information is combined with data storage in the same physical structure, eliminating the need for separate external mapping memory and reducing overall system complexity
Solution Approach 2:
The spare regions of the NVM are given multiple functions: they serve both as backup storage for data from bad rows and as storage for mapping addresses. This multi-functionality eliminates the need for dedicated mapping memory structures
3Device complexity
If spare regions are used to store mapping addresses instead of external memory, then mapping functionality is achieved with simpler architecture, but the spare region capacity is consumed
Solution Approach 1:
The system recovers capacity by converting previously unusable bad rows into functional storage units. Each bad row becomes a self-contained storage location that includes its own mapping address, effectively recovering the capacity of rows that would have been completely discarded
Solution Approach 2:
The spare region is segmented into multiple small storage units distributed throughout the memory array. Each segment (spare location) is associated with specific bad rows, allowing efficient use of spare capacity without requiring a large centralized mapping table
Data Source
AI summary
A memory system including a nonvolatile memory (NVM) device and a controller is provided. The NVM device includes a main region and a spare region. The controller writes write data to a selected row of the main region, determines whether the written row is bad, and writes the write data to a spare address in the spare region and writes the spare address to the bad row, when the written row is determined to be bad.