DRAM Refresh Management for Row Hammer Mitigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Repeated row activations in dynamic random access memory (DRAM) devices cause neighboring cells to lose stored values due to the 'row hammer' effect, leading to uncorrectable and undetectable errors across multiple DRAM devices.
Innovation Solution
Implementing refresh management (RFM) commands that refresh rows adjacent to actively used rows, alternating the number of activate commands between RFM commands to mitigate the row hammer effect, and using algorithmic or randomized selection techniques to determine which rows to refresh.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If refresh management commands are issued to mitigate row hammer effects, then data reliability is improved, but memory access time increases
Solution Approach 1:
The system proactively identifies rows that are likely to be hammered based on access patterns and refreshes them before actual data corruption occurs. The refresh management command pre-refreshes adjacent rows based on predicted hammering targets, preventing errors before they happen rather than detecting and correcting them after
Solution Approach 2:
The refresh interval varies dynamically based on the observed access patterns and hammering risk. The system adjusts the timing and frequency of refresh management commands adaptively - issuing refreshes more frequently for high-risk rows and less frequently for low-risk rows, optimizing the balance between reliability and access time
2Reliability
If refresh management commands are issued for every activate command, then data integrity is improved, but system performance deteriorates
Solution Approach 1:
Instead of uniformly refreshing all rows or using a fixed refresh rate for all memory operations, the system applies refresh management selectively to specific rows that are identified as hammering targets. The refresh management command targets only the adjacent rows that are at risk, rather than applying a blanket refresh approach to the entire memory array
Solution Approach 2:
The system issues refresh management commands at a rate that is sufficient to mitigate row hammer effects but not excessively frequent. Rather than refreshing after every single activate command, the system uses a threshold-based approach where refreshes are issued when the number of activates between refreshes reaches a configured threshold, providing adequate protection while avoiding unnecessary overhead
Data Source
AI summary
Refresh management commands are issued to a memory device in order to cause the refresh of rows in the vicinity of one or more rows being “hammered.” These refresh management commands are each associated with respective row addresses that indicates the row(s) to be refreshed in order to mitigate the likelihood of data being corrupted by “row hammer.” In an embodiment, the refresh management commands are issued in response to a varying number of activate (ACT) commands having been issued since the last refresh management command. The row selected for a given refresh management command may be selected based on rows that have recently been activated. The selection may be based on “pools” of recently activated rows where these pools are of unequal size. The selection from a given pool may be based on algorithmic and/or random techniques.