Memory Device Histogram Compression with Variable BEC Bins
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current metadata tables in memory sub-systems consume excessive volatile memory due to static bin widths, leading to high memory usage and wear on non-volatile memory, as they fail to efficiently handle varying bit error rates across memory regions.
Innovation Solution
Implementing compression techniques such as exponential and piecewise linearization to set variable bin widths for bit error count (BEC) data, allowing for fine resolution in low BEC data and coarse resolution in high BEC data, thereby reducing the memory required to maintain metadata tables.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If static bin widths are used in metadata tables, then data storage is simple and consistent, but memory consumption increases and non-volatile memory wear increases
Solution Approach 1:
The patent applies dynamics by transitioning from static bin widths to dynamic variable bin widths in the metadata table. The bin width is adjusted based on the bit error count (BEC) range, where lower BEC values use finer bin widths for precise tracking and higher BEC values use coarser bin widths to reduce memory consumption. This dynamic adaptation resolves the contradiction by maintaining data integrity while reducing overall memory usage.
Solution Approach 2:
The patent implements local quality by applying different bin width characteristics to different regions of the BEC data distribution. Specifically, the first portion of bins (covering lower BEC values) uses a first bin width for fine-grained monitoring, while the second portion of bins (covering higher BEC values) uses a second, larger bin width. This localized differentiation optimizes memory usage by allocating finer resolution only where needed.
2Ease of manufacture
If static bin widths are used in metadata tables, then implementation is straightforward, but wear on non-volatile memory increases
Solution Approach 1:
By implementing dynamic bin widths that adapt to the actual BEC distribution, the system reduces the number of bins required to represent the data. This dynamic approach decreases the frequency of metadata table updates to non-volatile memory, thereby reducing wear while maintaining effective error tracking capability.
Solution Approach 2:
The patent changes the parameter of bin width from a fixed value to a variable that depends on the BEC range. This parameter transformation allows the metadata table to use fewer bins to represent the same data range, reducing update operations and consequently reducing wear on non-volatile memory.
3Quantity of substance
If variable bin widths are implemented, then memory consumption is reduced, but data storage complexity increases
Solution Approach 1:
The patent segments the BEC data range into multiple portions, each associated with a specific bin width. The first portion of bins handles lower BEC values with a first bin width, while the second portion handles higher BEC values with a second bin width. This segmentation simplifies the management of variable bin widths by creating distinct, manageable regions with uniform characteristics within each region.
Solution Approach 2:
By assigning different bin widths to different portions of the BEC distribution, the patent creates a locally optimized structure where each region has appropriate resolution for its data characteristics. This local differentiation reduces overall memory consumption while maintaining necessary precision in critical regions.
4Object-generated harmful factors
If variable bin widths are implemented, then non-volatile memory wear is reduced, but implementation complexity increases
Solution Approach 1:
The dynamic bin width implementation reduces the size of the metadata table, which in turn reduces the frequency and volume of writes to non-volatile memory. This dynamic adaptation achieves wear reduction while the segmented structure keeps the implementation complexity manageable through systematic organization.
Data Source
AI summary
Systems and methods are disclosed including a memory device and a processing device operatively coupled to the memory device. The processing device can perform operations comprising performing a media scan on a portion of the memory device to obtain a metrics dataset comprising a plurality of data state metric values; generating, for the metrics dataset, a plurality of data state metric bins comprising a first set of bins having a first bin width and a second set of bins having a second bin width; associating a first data state metric value of the plurality of data state metric values with a first bin of first set of bins, and a second data state metric value of the plurality of data state metric values with a second bin of the second set of bins; and generating a histogram reflecting the data associated with the plurality of data state metric bins.