Dynamic Row Hammer Risk-Based Refresh for Semiconductor Memory
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Semiconductor memory devices face data loss due to the row hammering phenomenon, where frequent activations of specific word lines lead to data damage, and existing refresh operations are inefficient in preventing this, especially in terms of power consumption and refresh period optimization.
Innovation Solution
A memory system that determines the row hammer risk level per bank by counting target refresh commands and adjusts the target refresh period accordingly, incorporating a hidden refresh operation during normal refresh periods to prevent row hammering, while also considering thermal information for optimal refresh efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If target refresh operation is performed on specific word lines to prevent row hammering, then data reliability is improved, but power consumption increases due to additional refresh operations
Solution Approach 1:
The patent changes the refresh period parameter dynamically based on row hammer risk levels. Banks are classified into different risk levels (first level for high risk, second level for low risk), and the refresh period is adjusted accordingly - shorter for high-risk banks and longer for low-risk banks. This parameter adaptation allows the system to maintain data reliability while reducing unnecessary refresh operations, thereby lowering power consumption.
2Reliability
If refresh period is shortened to prevent row hammering, then data protection is improved, but refresh efficiency deteriorates due to increased operation frequency
Solution Approach 1:
The patent applies different refresh strategies to different banks based on their individual row hammer risk levels. Instead of uniformly shortening the refresh period for all banks, the system identifies specific banks at high risk (first level) and applies shorter refresh periods only to those banks, while maintaining longer refresh periods for low-risk banks (second level). This localized approach ensures data protection for vulnerable banks without sacrificing overall refresh efficiency.
3Device complexity
If uniform refresh operation is applied to all banks, then implementation simplicity is maintained, but refresh optimization deteriorates due to inability to address specific row hammer risks
Solution Approach 1:
The patent transitions from a static, uniform refresh approach to a dynamic, adaptive refresh strategy. The system continuously monitors active command counts for each bank and dynamically adjusts the refresh period based on detected row hammer risk levels. This dynamic adaptation enables the system to respond to changing workloads and threat patterns, optimizing refresh operations for each bank's specific needs while maintaining manageable implementation complexity through systematic classification and control.
Data Source
AI summary
A memory system includes: a memory controller suitable for: generating a normal refresh command and a target refresh command when a number of inputs of an active command reaches a certain number, and providing the active command, the normal refresh command, the target refresh command, and an address; and a memory device including a plurality of banks and suitable for: performing a target refresh operation on one or more word lines of at least one bank in response to the target refresh command, determining a row hammer risk level per bank by counting, within a periodic interval, a number of inputs of the target refresh command per bank based on the address, and performing a hidden refresh operation corresponding to the row hammer risk level per bank in response to the normal refresh command.