Row Hammer Refresh Address Detection With Hierarchical Frequency Counting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for obtaining row hammer refresh addresses are either inaccurate or require complex circuit structures and large chip areas, with the row hammer effect leading to data loss due to excessive charge loss in adjacent addresses.
Innovation Solution
A method involving high and low registers to store and count access frequencies of row addresses, where high addresses meeting a frequency threshold are locked first, and their corresponding low addresses are recorded, allowing for accurate identification of row hammer refresh addresses with a simpler circuit design.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If random selection method is used to obtain row hammer address, then the circuit structure is simple, but the accuracy of identifying row hammer addresses is not high
Solution Approach 1:
The patent changes the parameter of address selection from random to frequency-based, using access frequency as a new parameter to identify row hammer addresses. This allows the system to maintain relatively simple circuit structure while significantly improving identification accuracy by selecting addresses that are accessed frequently rather than randomly.
Solution Approach 2:
The patent replaces the mechanical/random selection process with a counting and comparison mechanism. Instead of randomly selecting addresses, the system uses counters to track access frequencies and comparators to identify addresses exceeding thresholds, substituting physical randomness with a deterministic frequency-based approach.
2Measurement precision
If counting method is used to obtain row hammer address, then the accuracy of identifying row hammer addresses is improved, but the circuit structure becomes complex and chip area increases
Solution Approach 1:
The patent segments the address counting process into two hierarchical levels: high-order address counting and low-order address counting. This segmentation allows the system to identify row hammer addresses more accurately while keeping individual counter units relatively simple, as each counter only needs to track a portion of the address space rather than the entire space.
Solution Approach 2:
The patent introduces a hierarchical dimension to the counting process by dividing addresses into high-order and low-order parts. This dimensional change allows the system to manage complexity by organizing counters in a tree-like structure, where high-order counters first identify candidate regions, and low-order counters then identify specific addresses within those regions.
3Measurement precision
If counting method is used to obtain row hammer address, then the accuracy of identifying row hammer addresses is improved, but the chip area consumption increases
Solution Approach 1:
The patent segments the chip area usage by organizing counters hierarchically. Instead of allocating one large counter for the entire address space, the system divides the area into multiple smaller counter units organized in a tree structure, reducing the area required for each individual counter while maintaining overall counting accuracy.
Solution Approach 2:
The patent implements a nested counter structure where low-order counters are nested within high-order counter regions. This nesting allows the system to share hardware resources between different counting levels, reducing total chip area consumption while maintaining the ability to accurately identify row hammer addresses through the hierarchical counting process.
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
The disclosure provides a method and device for obtaining a row hammer refresh address. The method includes: after a last row hammer refresh signal arrives, obtaining a current sampling address, and determining whether a high address is locked in a current row hammer refresh cycle; in response to the high address being locked, determining whether a high address of the current sampling address is identical to the high address locked in the current row hammer refresh cycle; in response to the high address of the current sampling address being identical to the high address locked in the current row hammer refresh cycle, updating an access frequency of the locked high address, and updating access frequencies of low addresses with a low address of the current sampling address; and when a next row hammer refresh signal arrives, using a low address with a highest access frequency as a low address of the row hammer refresh address, and using the locked high address as a high address of the row hammer refresh address. According to such configuration, it is unnecessary to count all sampling addresses, the method is simpler, the required circuit is simple, and the row hammer refresh address may also be accurately obtained.