Memory Fragmentation Assessment via Address Mapping Tables
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Memory systems face performance degradation due to fragmentation, where data files become distributed across non-continuous pages and dies, reducing parallelism and efficiency, and existing methods to address this often require reading the entire file to determine fragmentation level, consuming power and time.
Innovation Solution
The use of address mapping information, such as an address mapping table, to determine the fragmentation level and performance impact, allowing for targeted maintenance operations like defragmentation without reading the entire file, thereby optimizing power and time usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the entire data file is read to determine fragmentation level, then measurement precision is improved, but use of energy and time are worsened
Solution Approach 1:
The patent extracts only the necessary address mapping information from the address mapping table to determine fragmentation level, rather than reading the entire data file. This selective extraction of relevant information (physical addresses and their distribution across dies) enables accurate fragmentation measurement while minimizing energy consumption and access time.
Solution Approach 2:
The address mapping table is maintained and updated in advance, storing the relationship between logical addresses and physical addresses. This preliminary action allows the system to quickly query fragmentation information without performing full file reads when fragmentation assessment is needed, thereby reducing real-time energy consumption and latency.
2Measurement precision
If the entire data file is read to determine fragmentation level, then measurement precision is improved, but time consumption is worsened
Solution Approach 1:
The patent extracts only the necessary address mapping information from the address mapping table to determine fragmentation level, rather than reading the entire data file. This selective extraction of relevant information (physical addresses and their distribution across dies) enables accurate fragmentation measurement while minimizing energy consumption and access time.
Solution Approach 2:
The address mapping table is maintained and updated in advance, storing the relationship between logical addresses and physical addresses. This preliminary action allows the system to quickly query fragmentation information without performing full file reads when fragmentation assessment is needed, thereby reducing real-time energy consumption and latency.
3Quantity of substance
If data files are distributed across non-continuous pages and dies, then storage capacity utilization is improved, but memory system performance is worsened
Solution Approach 1:
The patent implements a feedback mechanism that continuously monitors fragmentation levels by querying the address mapping table. When fragmentation exceeds a threshold, the system triggers defragmentation operations to restore performance. This feedback loop allows the system to maintain optimal performance while maximizing storage utilization, as defragmentation is performed only when necessary rather than continuously.
Solution Approach 2:
The patent changes the organizational parameter of data files from fragmented distribution to contiguous distribution across memory dies when defragmentation is performed. This parameter change (from non-continuous to continuous physical addressing) improves parallelism and memory system performance while preserving the ability to utilize storage capacity efficiently through dynamic remapping.
Data Source
AI summary
Methods, systems, and devices for memory performance evaluation using address mapping information are described. A memory system may write a data file associated with a range of logical block addresses to a first set of blocks in a set of memory dies. The memory system may then determine, for a memory die of the set of memory dies that stores a portion of the data file, a quantity of read operations for reading the portion of the data file. The memory system may determine the quantity of read operations based on address mapping information that maps logical block addresses associated with the datafile to physical addresses. After determining the quantity of read operations, the memory device may use the quantity of read operations to determine a performance metric associated with the data file and the first set of blocks.