Memory Controller Dynamic Cluster Mode Switching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing storage capacity of SSDs leads to a proportional increase in the size of logical-to-physical conversion tables, necessitating a reduction in table size to enhance memory efficiency, while maintaining efficient page usage and avoiding increased Write Amplification Factor (WAF) that can decrease SSD performance and lifetime.
Innovation Solution
A memory controller that dynamically manages logical-to-physical conversion data using two cluster modes (large and normal cluster modes) and compresses mapping data, allowing for efficient storage and reduced memory requirements by switching between modes based on data usage ratios and compression thresholds.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If the storage capacity of SSD is increased, then the storage capacity is improved, but the size of logical-to-physical conversion table increases proportionally
Solution Approach 1:
The patent divides the conversion table into multiple segments or partitions, allowing the table to be managed in smaller units. This segmentation enables more efficient memory utilization and reduces the impact of table size increases on overall system performance as storage capacity scales.
Solution Approach 2:
The patent introduces a hierarchical structure with multiple levels of conversion tables (e.g., first conversion table and second conversion table), effectively adding a dimensional layer to the address mapping system. This allows the system to manage large storage capacities without requiring a single monolithic conversion table to grow proportionally.
2Volume of stationary object
If the conversion table size is reduced, then memory efficiency is improved, but page usage efficiency may deteriorate
Solution Approach 1:
The patent applies different cluster modes (first cluster mode and second cluster mode) to different regions or portions of the conversion table based on local characteristics. This allows highly accessed areas to use finer-grained normal clusters for optimal page usage efficiency, while less accessed areas use coarser large clusters to minimize table size, achieving both goals simultaneously.
Solution Approach 2:
The patent dynamically switches between different cluster modes based on access patterns and usage characteristics. The system can transition from large cluster mode to normal cluster mode and vice versa, allowing the conversion table structure to adapt to changing workloads and maintain optimal performance while keeping table size manageable.
3Productivity
If normal cluster mode is used instead of large cluster mode, then page usage efficiency is improved, but Write Amplification Factor increases
Solution Approach 1:
The patent applies normal cluster mode only to specific portions of the conversion table where fine-grained page management is necessary for high page usage efficiency, while using large cluster mode for other portions where such efficiency is less critical. This localized application minimizes the overall Write Amplification Factor while maintaining necessary performance in key areas.
Solution Approach 2:
The patent uses normal cluster mode partially rather than universally, applying it only where needed based on access patterns and performance requirements. This partial application of the finer-grained normal cluster mode achieves sufficient page usage efficiency without incurring the full penalty of increased Write Amplification Factor that would result from using normal clusters throughout the entire conversion table.
Data Source
AI summary
A memory controller controls first and second memory, and includes a control unit. In response to a first write command from a host, which designates a logical address for first data to be written to the first memory, the control unit determines whether mapping of the logical address is presently being managed in a first mode with a first cluster size or a second mode with a second cluster size that is smaller than the first cluster size, changes first mapping data for the logical address stored in a first table in the second memory, from the first cluster size to the second cluster size, if the mapping of the logical address is being managed in the first mode and the first mapping data can be compressed at a ratio lower than a first compression ratio, and writes the first data to a physical address of the first memory.