Autonomous Memory Refresh Manager for DRAM Bandwidth Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Dynamic random access memory (DRAM) cells require frequent refresh operations due to charge leakage, leading to data loss and memory bandwidth bottlenecks, especially in high-data processing applications like AI, which degrades performance.
Innovation Solution
A memory system with a sharing manager that autonomously manages refresh operations based on monitored information, allowing the system to perform refresh operations independently in auto mode, reducing the need for frequent host intervention and optimizing memory bandwidth.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If frequent refresh operations are performed on DRAM cells, then data loss is prevented, but memory bandwidth is reduced and data processing performance is degraded
Solution Approach 1:
The memory system performs autonomous refresh operations by monitoring its own refresh status and automatically initiating refresh when needed, eliminating the need for continuous host intervention and reducing bandwidth consumption while maintaining data integrity
Solution Approach 2:
The system monitors refresh status and operational parameters, using this feedback information to dynamically adjust refresh timing and duration, optimizing the balance between data retention and bandwidth availability
2Measurement precision
If the host manually manages refresh operations, then refresh timing can be precisely controlled, but system complexity increases and bandwidth is consumed for refresh commands
Solution Approach 1:
The memory system autonomously monitors its own refresh status and automatically initiates refresh operations when needed, eliminating the need for continuous host intervention and reducing bandwidth consumption while maintaining data integrity
Solution Approach 2:
The refresh management function is separated into autonomous monitoring and execution components within the memory system, allowing precise control to be maintained while reducing overall system complexity and host burden
Data Source
AI summary
The present disclosure relates to a memory system and an operating method thereof. The memory system may include a shared memory device to store data, a sharing manager to store operation policy information and to autonomously generate a first internal command by using the operation policy information during an auto mode started in response to receiving an auto mode start command from a host, and a memory controller to generate a second internal command for controlling the shared memory device in response to the first internal command.


