Configurable Memory Column Planes for Metadata and Capacity Recovery
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing semiconductor memory devices face challenges in efficiently storing metadata while maintaining single-pass access and minimizing the loss of addressable memory space, leading to performance reductions in some applications.
Innovation Solution
A memory device with selectively configurable column planes that can store metadata, allowing single-pass access to data, metadata, and ECC data, and enabling flexible operation modes to utilize metadata space for data when needed, with the option to form virtual column planes when metadata is not stored.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If metadata is stored in the memory array in dedicated column planes, then metadata can be retrieved along with data and ECC data in a single pass, but addressable memory space is lost
Solution Approach 1:
The memory device dynamically reconfigures column planes based on whether metadata storage is enabled. When metadata storage is disabled, column planes that would be dedicated to metadata are dynamically reallocated to store data, maximizing addressable memory space. When metadata storage is enabled, these same column planes are configured to store metadata, enabling single-pass retrieval. This dynamic reconfiguration resolves the contradiction by allowing the system to adapt its structure based on operational requirements.
Solution Approach 2:
Column planes in the memory array are designed to serve multiple functions: they can store data, metadata, or be reallocated between these purposes based on configuration. This multi-functionality allows the same physical memory resources to support both single-pass access (when configured for metadata) and maximum storage capacity (when reallocated to data), eliminating the need for dedicated metadata planes that would permanently reduce addressable space.
2Adaptability or versatility
If metadata is stored in the memory array, then metadata can supplement data with additional information, but array density for data storage is reduced
Solution Approach 1:
The memory device employs dynamic reconfiguration of column planes based on the operational mode. When metadata functionality is required, specific column planes are allocated to metadata storage. When maximum data storage capacity is needed, these same column planes are dynamically reallocated to extend the data storage array. This dynamic allocation allows the system to adapt its functionality and capacity based on real-time requirements, resolving the contradiction between metadata versatility and data storage density.
Solution Approach 2:
The system changes the operational parameters of column planes based on configuration modes. By adjusting whether column planes operate in data-storage mode or metadata-storage mode, the system can toggle between maximizing data capacity and enabling metadata functionality. This parameter change approach allows the same hardware resources to support both high data density and metadata supplementation without permanent compromise.
3Quantity of substance
If multiple passes are used to retrieve metadata and data, then more efficient storage of metadata is possible, but access time increases
Solution Approach 1:
The invention merges the retrieval of metadata, data, and ECC data into a single simultaneous access operation. By organizing metadata and data in specific column plane configurations and using unified address decoding, the system retrieves all required information in one pass through the memory array, eliminating the time penalty associated with multiple sequential access passes while maintaining efficient metadata storage.
Solution Approach 2:
The system resolves the access time issue by organizing metadata and data in different dimensional spaces within the memory array. Metadata is stored in specific column planes that can be simultaneously accessed with data column planes through parallel decoding paths. This spatial organization in different dimensions allows both metadata and data to be retrieved concurrently in a single access operation, rather than requiring sequential passes.
Data Source
AI summary
A bank of a memory device may be divided into column planes. Each column plane may be associated with column selects. In some examples, one or more physical column planes may be selectively configured to store metadata. When the memory device is configured not to store metadata, the column selects may be arranged into virtual column planes to allow data to be stored in physical column planes used for metadata. The physical column planes may be arranged into virtual planes to store the data. Different mapping of the virtual planes to the physical planes may be used.