DRAM Victim Data Adjustment via Reverse Scrambler
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In semiconductor memory devices, particularly DRAM, the repeated access to certain word lines causes an increased rate of data degradation in nearby memory cells, known as the 'row hammer' effect, leading to the need for higher refresh rates, which can increase power consumption and delay memory access operations.
Innovation Solution
The implementation of a 'reverse scrambler' that calculates row addresses of potential aggressor word lines and adjusts victim word line data to reduce unnecessary targeted refresh operations by identifying and 'healing' victim word lines through targeted or auto-refresh mechanisms, thereby minimizing the impact on auto-refresh rates and power consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the refresh rate of victim word lines is increased to preserve data integrity, then data reliability is improved, but power consumption increases
Solution Approach 1:
The system performs preliminary identification of victim word lines by detecting aggressor row access patterns before data degradation occurs. By pre-calculating and marking victim word lines associated with detected aggressor rows, the system prepares a targeted refresh list in advance, allowing refresh operations to be performed only when necessary rather than continuously refreshing all word lines.
Solution Approach 2:
Instead of uniformly refreshing all word lines across the memory array, the system applies differentiated refresh strategies to specific local regions. Victim word lines identified through aggressor row detection receive targeted refresh operations, while non-victim word lines follow the standard auto-refresh schedule. This localized approach concentrates refresh resources only where data integrity is at risk.
2Reliability
If targeted refresh operations are performed on victim word lines, then data reliability is improved, but memory access delay increases
Solution Approach 1:
The system pre-identifies and marks victim word lines by calculating their association with aggressor rows before refresh operations are needed. This preliminary marking process allows the refresh controller to quickly determine which word lines require refresh without performing complex calculations during the actual refresh operation, reducing memory access delay.
Solution Approach 2:
The system performs refresh operations only on the subset of word lines identified as victims of row hammer attacks, rather than refreshing all word lines in the memory array. This partial action approach reduces the total number of refresh operations required, thereby minimizing the impact on memory access performance while still protecting vulnerable data.
3Reliability
If the number of refresh operations is increased to protect victim word lines, then data reliability is improved, but productivity decreases
Solution Approach 1:
The system implements localized refresh operations only on victim word lines identified through aggressor row detection, rather than performing blanket refresh operations across the entire memory array. This selective approach maintains data reliability for vulnerable regions while preserving memory access efficiency for the majority of non-victim word lines.
Solution Approach 2:
By performing refresh operations on only the necessary subset of victim word lines rather than all word lines, the system reduces the total refresh overhead. This partial action strategy protects data integrity where needed while minimizing the impact on overall memory productivity and access efficiency.
Data Source
AI summary
Addresses of accessed word lines are stored. Data related to victim word lines associated with the accessed word line are also stored. The victim word lines may have data stored in relation to multiple accessed word lines. The data related to the victim word lines is adjusted when the victim word line is refreshed during a targeted refresh operation or an auto-refresh operation. The data related to the victim word lines is adjusted when the victim word line is accessed during a memory access operation.