Autonomous Link Management for Multi-Die Assemblies
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
As system-on-chip (SoC) die sizes increase to support advanced functionalities and communication speeds, manufacturing costs rise, and yields decline, with reliable and fast communication channels between multiple dice in a multi-die assembly becoming challenging due to noise interference and frequent link failures.
Innovation Solution
Implementing a link management system with detection units and controllers on each die to autonomously detect failing SERDES communication links, redirect traffic to operational links during maintenance, and re-establish communication once the failing link is repaired, ensuring continuous operation without software intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If the SoC die size is increased to support advanced functionalities and communication speeds, then the performance and communication capabilities are improved, but the manufacturing costs increase and yields decline
Solution Approach 1:
The patent divides a large SoC die into multiple smaller dice (first die, second die, etc.) that are fabricated separately and then assembled into a multi-die assembly. This segmentation allows each smaller die to be manufactured with higher yield and lower cost, while the overall system achieves the required communication performance through multiple parallel communication links between the dice.
2Ease of manufacture
If multiple dice are assembled in a multi-chip module to reduce manufacturing costs, then the manufacturing yield is improved, but the reliability of communication channels among dice becomes challenging
Solution Approach 1:
The patent implements dynamic parameter changes in communication link management by continuously monitoring link quality metrics (such as error rates) and adjusting system behavior accordingly. When link quality degrades below a threshold, the system automatically triggers link failure detection and initiates traffic redirection to alternative links, thereby maintaining communication reliability despite the presence of multiple dice.
Solution Approach 2:
The patent establishes a feedback mechanism where the system continuously monitors the quality of communication links between dice and uses this information to make real-time decisions about link maintenance and traffic routing. The feedback loop includes detection of link failures, notification to controllers, and automatic redirection of communication traffic, ensuring high reliability in multi-die assemblies.
3Duration of action of stationary object
If communication traffic is redirected to operational links during link maintenance, then the continuity of operation is maintained, but the system complexity increases
Solution Approach 1:
The patent implements self-service mechanisms where the multi-die assembly automatically detects link failures and redirects traffic without requiring external software intervention. The controllers on each die autonomously monitor link quality, detect failures, and manage traffic redirection, reducing the need for complex external control systems while maintaining continuous operation.
Solution Approach 2:
The patent prepares alternative communication paths in advance by establishing multiple parallel communication links between dice. When a link failure is detected, the system can immediately redirect traffic to pre-configured alternative links, minimizing disruption and maintaining continuous operation without requiring complex real-time routing decisions.
Data Source
AI summary
Systems and methods are disclosed to provide an autonomous management of communication links between dice on a multi-die assembly. Each die can include a detection unit and a controller to detect a failing communication link and perform link maintenance by directing the communication traffic on the failing link to an operational link before the link fails. Once the failing link has been repaired, the controller can re-direct the traffic back to the repaired link. The controllers on each die can negotiate through a handshake process to provide the continuous operation by switching the communication traffic from the failing link to the operational link, and then from the operational link to the repaired link.