Memory System Parallel Translation Table Access
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
As data capacity in nonvolatile memory increases, the size of the translation table also grows, leading to a proportional increase in memory requirements for the cache memory, which can result in higher costs and potential decreases in read operation throughput if not managed efficiently.
Innovation Solution
A memory system with a nonvolatile first memory and a volatile second memory, where only a part of the translation table is copied into the second memory, utilizing a management table to track stored data and a scheduler to optimize command execution, allowing parallel reading of data and translation tables, thereby reducing memory capacity needs and maintaining throughput.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the entire translation table is copied into the volatile second memory, then read operation throughput is improved, but memory capacity requirements and system cost increase
Solution Approach 1:
The translation table is divided into multiple segments that are distributed across different memory locations. The volatile second memory stores only the currently needed translation table segments, while the nonvolatile first memory stores the complete translation table. This segmentation allows the system to maintain high read throughput by having ready access to needed segments without requiring the entire translation table in fast memory.
Solution Approach 2:
The system performs preliminary actions by pre-loading translation table segments into the volatile second memory before they are needed for read operations. The controller anticipates read requests and loads the corresponding translation table segments in advance, ensuring that when read operations occur, the necessary translation data is already available in fast memory, thus maintaining high throughput without permanently allocating large memory capacity.
2Quantity of substance
If only a part of the translation table is copied into the volatile second memory, then memory capacity requirements are reduced, but read operation throughput may decrease
Solution Approach 1:
The system dynamically manages the translation table segments in the volatile second memory, loading and unloading segments based on current read operation needs. The controller monitors access patterns and adjusts which translation table segments are kept in fast memory, ensuring that the most frequently accessed segments are available while minimizing the total memory capacity required. This dynamic approach maintains throughput by ensuring needed data is available when required.
Solution Approach 2:
The controller acts as an intermediary between the host and the memory system, managing the translation table segments in the volatile second memory. It intercepts read requests, determines which translation table segments are needed, loads them from the nonvolatile first memory if necessary, and coordinates their use. This intermediary management ensures that read operation throughput is maintained through efficient coordination without requiring the entire translation table to be permanently stored in fast memory.
3Quantity of substance
If the translation table is stored entirely in nonvolatile memory, then memory capacity requirements are reduced, but access speed decreases
Solution Approach 1:
The system applies local quality by storing translation table segments with different access frequency characteristics in different memory types. Frequently accessed translation table segments are kept in the volatile second memory for fast access, while less frequently accessed segments remain in the nonvolatile first memory. This localized optimization ensures that the critical path for read operations uses fast memory, maintaining high access speed for common operations while using capacity-efficient nonvolatile memory for the complete translation table.
Data Source
AI summary
According to one embodiment, a memory system includes a first memory as a nonvolatile memory storing first data, second data as a translation table for accessing the first data, third data, fourth data as a translation table for accessing the third data, and including two memory cell arrays which are accessible in parallel, a second memory in which the second and fourth data is storable, and which stores a management table for managing information about whether the second and fourth data is stored in the second memory, a controller checking whether the second and fourth data is stored in the second memory based on the management table, a third memory storing an order of executing commands to be issued to the first memory, and a scheduler scheduling the order based on a result of the checking, two of a first command for reading the first data, a second command for reading the second data, a third command for reading the third data and a fourth command for reading the fourth data being executed in parallel in the order.