Wear Leveling in Memory Devices Excluding Unusable Units
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing start-gap wear leveling methods in memory sub-systems do not effectively exclude unusable management units (UMUs) from the address space, leading to reduced efficiency and increased latency due to the use of unreliable data storage.
Innovation Solution
The implementation of a modified start-gap wear leveling method that dynamically reduces the wear leveling pool by identifying and excluding unusable management units (UMUs) from the address space, using a mapping between logical and physical addresses and modifying registers to manage data movement effectively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If existing start-gap wear leveling methods include all management units in the address space, then the wear leveling pool size is maximized, but unusable management units reduce data reliability and increase latency
Solution Approach 1:
The patent extracts unusable management units (UMUs) from the wear leveling pool by maintaining a separate list of UMUs and excluding them from address translation operations. This allows the system to remove unreliable components without redesigning the entire address space management system.
Solution Approach 2:
The system performs preliminary identification and classification of management units as usable or unusable during initialization and ongoing operations. By pre-marking UMUs before wear leveling operations begin, the system avoids wasting time on unreliable units during normal operation.
2Productivity
If wear leveling operations include unusable management units, then the distribution of wear is more uniform, but the time required for wear leveling operations increases due to latency
Solution Approach 1:
The patent removes unusable management units from the wear leveling candidate pool, ensuring that wear leveling operations only target reliable units. This extraction eliminates time wasted accessing unreliable units while maintaining effective wear distribution across remaining units.
Solution Approach 2:
The system continuously monitors management unit status and updates the usable/unusable classification based on error rates and operational performance. This feedback mechanism ensures that wear leveling operations dynamically adapt to unit reliability, avoiding latency from failed operations.
3Reliability
If the wear leveling pool includes all management units, then the pool size is maximized for wear distribution, but system resources are wasted on unreliable units
Solution Approach 1:
The patent extracts unusable management units from the wear leveling pool, creating a refined pool that contains only reliable units. This maintains the largest possible pool of usable units for wear distribution while eliminating waste from including defective units.
Solution Approach 2:
The system dynamically adjusts the effective wear leveling pool size by excluding UMUs based on their reliability parameters. This parameter-based filtering ensures the pool contains only units meeting minimum reliability thresholds, optimizing both size and quality.
4Reliability
If unusable management units are excluded from the address space, then data reliability improves, but the address translation complexity increases
Solution Approach 1:
The patent extracts unusable management units from the addressable space by maintaining a separate exclusion list. Address translation logic checks this list and skips UMUs, adding minimal complexity compared to redesigning the entire address translation system.
Solution Approach 2:
The patent introduces an intermediary layer in address translation that mediates between logical addresses and physical management units. This intermediary checks the UMU exclusion list and routes translations appropriately, isolating the complexity from critical path operations.
Data Source
AI summary
A system includes a memory device; and a processing device coupled to the memory device, the processing device to perform operations including: identifying at least one unusable management unit (UMU) in a plurality of management units that are designated for wear leveling; storing, in a data structure, a physical address and a logical address of the identified at least one UMU; excluding, from a physical address space for wear leveling, the physical address of the identified at least one UMU; and performing a wear leveling operation using the physical address space, wherein the wear leveling operation moves data of a management unit to a neighboring management unit in a circular manner.