Coordinated 3D ECC for Stacked Memory Vulnerabilities
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional memory systems treating stacked memories as '2D' structures impose limitations on bandwidth and speed, leading to inefficiencies in power consumption and error correction, with conventional error correction code (ECC) approaches incurring significant overhead and missing correctable errors.
Innovation Solution
Implementing coordinated 3D ECC mechanisms between tiers of memory, where error correction code engines communicate to detect and correct vulnerabilities across different memory dies, reducing unnecessary refreshes and optimizing performance and power efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional 2D ECC approaches are used for each tier of stacked memory, then error correction is performed independently per die, but performance and power overhead increase significantly and correctable errors may be missed
Solution Approach 1:
The patent merges error correction operations across multiple memory tiers by having ECC engines on different tiers coordinate and share vulnerability information. Instead of each tier operating independently, the system combines error detection and correction resources across tiers, allowing a single coordinated ECC operation to cover multiple tiers simultaneously, thereby reducing redundant overhead while maintaining or improving error correction reliability
Solution Approach 2:
The ECC engines are designed to perform multiple functions: they can operate independently on their local tier or coordinate with other tiers to perform unified error correction across the stacked memory structure. This multi-functionality allows the system to adapt error correction scope based on detected vulnerabilities, optimizing both performance and reliability
2Reliability
If conventional 2D ECC approaches are used for each tier of stacked memory, then error correction is performed independently per die, but power consumption increases due to redundant operations
Solution Approach 1:
The patent merges error correction operations across multiple memory tiers by having ECC engines on different tiers coordinate and share vulnerability information. Instead of each tier operating independently, the system combines error detection and correction resources across tiers, allowing a single coordinated ECC operation to cover multiple tiers simultaneously, thereby reducing redundant overhead while maintaining or improving error correction reliability
Solution Approach 2:
The system performs error correction only when and where vulnerabilities are detected through coordination between tiers. Rather than continuously performing full ECC operations on every tier independently, the coordinated approach enables partial action—applying error correction selectively to affected regions and tiers, thus reducing unnecessary power consumption while maintaining reliability
3Ease of manufacture
If stacked memory is treated as 2D structure, then conventional ECC can be applied, but bandwidth and speed are limited
Solution Approach 1:
The patent transitions from treating stacked memory as separate 2D tiers to utilizing the third dimension (vertical stacking) for coordinated error correction. By having ECC engines communicate across tiers and share vulnerability information, the system exploits the 3D architecture to enable parallel error detection and correction operations that increase bandwidth and speed while maintaining implementation feasibility
Data Source
AI summary
Error correction for stacked memory is described. In accordance with the described techniques, a system includes a plurality of error correction code engines to detect vulnerabilities in a stacked memory and coordinate at least one vulnerability detected for a portion of the stacked memory to at least one other portion of the stacked memory.