Memory Refresh Coordination Across Victim Rows for Disturbance Mitigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
As processors and memories operate more quickly together, increasing chip density in memory devices leads to electromagnetic coupling between adjacent rows, causing voltage fluctuations that result in memory errors and data loss due to usage-based disturbance, which existing mitigation methods struggle to address efficiently without slowing down performance or consuming excessive power.
Innovation Solution
Implementing multiple-row refresh techniques that refresh at least one victim row for usage-based disturbance mitigation during another refresh operation, allowing parallel refreshing of victim rows coupled to different components, thereby mitigating interference without conflicting with regular memory access operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If chip density is increased in memory devices, then memory capacity and integration are improved, but electromagnetic coupling between adjacent rows increases causing voltage fluctuations and memory errors
Solution Approach 1:
The patent applies preliminary action by performing refresh operations on victim rows before usage-based disturbance can cause errors. The system proactively identifies rows that will be affected by electromagnetic coupling and refreshes them in advance during designated refresh cycles, preventing voltage fluctuations from leading to memory errors before they occur.
Solution Approach 2:
The patent converts the harmful electromagnetic coupling effect into a beneficial pattern by using the timing and electrical characteristics of the disturbance to trigger targeted refresh operations. The usage-based disturbance itself provides timing information that guides the refresh mechanism, transforming the harmful voltage fluctuations into a signal for corrective action.
2Reliability
If existing mitigation methods are applied to address usage-based disturbance, then memory reliability is improved, but performance is slowed down and power consumption increases
Solution Approach 1:
The patent applies local quality by providing targeted refresh operations only to specific victim rows that are affected by usage-based disturbance, rather than refreshing all rows uniformly. The system identifies which rows are vulnerable based on their access patterns and electromagnetic coupling characteristics, and applies refresh operations locally to those rows, leaving other rows unaffected and maintaining overall performance.
Solution Approach 2:
The patent implements dynamics by making the refresh strategy adaptive and responsive to actual usage patterns. The system continuously monitors row access patterns and dynamically adjusts which rows require refresh operations based on real-time data, allowing the mitigation strategy to evolve with changing workload characteristics rather than using a static approach.
3Reliability
If multiple rows are refreshed simultaneously for usage-based disturbance mitigation, then memory reliability is improved, but device complexity increases
Solution Approach 1:
The patent applies merging by combining multiple refresh operations into a single integrated refresh cycle. Instead of managing separate refresh operations for different rows independently, the system merges the refresh of multiple victim rows into one coordinated operation, reducing the complexity of control logic while maintaining comprehensive coverage of all affected rows.
Data Source
AI summary
Apparatuses and techniques for implementing multiple-row refresh for usage-based disturbance mitigation are described. Multiple-row refresh for usage-based disturbance mitigation involves refreshing at least one victim row for usage-based disturbance mitigation during a refresh operation that involves another row. The victim row and the other row are effectively refreshed in parallel. Possible victim rows that can be refreshed can be coupled to components that are not shared by the other row associated with the refresh operation. Any count update procedures associated with refreshing the victim row and the other row can be performed at different times to avoid potential conflict on global input/output lines. Refreshing a victim row during a refresh operation of another row allows for refreshing extra victim rows.