Memory Controller Row-Hammer Tracking Without Per-Row Counting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing memory systems face issues with row hammering, where continuous access to memory cells leads to distortion of data in adjacent cells, and current methods to mitigate this often require counting access numbers for each row, which is inefficient in terms of area usage.
Innovation Solution
A memory system with a memory controller that analyzes access patterns to select vulnerable banks and tracks rows using average address values, allowing for row hammer mitigation without counting individual row accesses, thereby optimizing area usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If row hammer mitigation is implemented by counting access numbers for each row, then row hammering can be effectively prevented, but the area usage increases significantly
Solution Approach 1:
The patent merges the row hammer tracking function with the existing refresh operation. The memory controller performs row hammer tracking by analyzing access patterns during normal refresh operations, combining two functions (row hammer detection and memory refresh) into a single integrated process, thereby preventing row hammering without requiring separate dedicated hardware circuits for counting
Solution Approach 2:
The refresh operation is given multi-functionality: it simultaneously serves as both a memory refresh operation and a row hammer tracking operation. By analyzing access patterns during the refresh process, the system can identify vulnerable rows and apply mitigation without adding separate specialized hardware, thus reducing area usage while maintaining reliability
2Measurement precision
If access patterns are analyzed for each row to identify vulnerable rows, then row hammer vulnerability can be detected, but the complexity of the system increases
Solution Approach 1:
The patent segments the analysis process by dividing rows into groups and analyzing access patterns at the group level rather than tracking each individual row. This segmentation approach reduces the complexity of tracking while still enabling effective identification of vulnerable rows through pattern analysis of grouped access data
Data Source
AI summary
A memory system includes a memory device including banks each including a plurality of rows, and configured to select, within a selected bank of the banks, at least one row from the plurality of rows based on a target address; and a row-hammer tracking module configured to: select the selected bank based on a pattern of accesses to each of the banks, and select the target address according to an access number for each of a plurality of groups of the rows within the selected bank.