Die-to-Die Main Band Link Width for Asymmetric Lane Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing interconnect standards like UCIe 1.0 do not efficiently manage power consumption and data rate requirements in bidirectional chip-to-chip connections by disabling both directions when a fault occurs in one, leading to unnecessary power usage and potential underutilization of data lanes.
Innovation Solution
Implementing a variable link width mechanism in the main band connection that allows independent configuration of data lanes in each direction through a sideband request and response, enabling dynamic adjustment based on traffic demands and fault conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If both directions of the main band are disabled when a fault occurs in one direction, then system reliability is maintained, but power consumption increases and data lane utilization decreases
Solution Approach 1:
The main band connection is segmented into independent bidirectional channels, allowing each direction to be independently enabled or disabled based on fault conditions. When a fault is detected in one direction, only that specific direction is disabled while the other direction continues to operate, thereby maintaining reliability for unaffected communications while reducing power consumption compared to disabling both directions.
Solution Approach 2:
The link width configuration is made dynamic and adaptive rather than static. The system continuously monitors for faults and adjusts the operational state of each direction independently. This dynamic adjustment allows the connection to maintain optimal reliability while minimizing power consumption by activating only the necessary data lanes in each direction based on real-time conditions.
2Reliability
If both directions are disabled when one direction has a fault, then fault propagation is prevented, but data lane utilization decreases
Solution Approach 1:
The connection is divided into independent transmit and receive directions that can be independently managed. When a fault is detected in one direction, the system segments the fault isolation to only that direction, allowing the other direction to continue utilizing its data lanes for communication, thereby maintaining fault isolation while maximizing data lane utilization.
Solution Approach 2:
The system dynamically adjusts the operational status of each direction based on real-time fault detection. This dynamic approach ensures that fault isolation is applied only where necessary, preventing unnecessary disablement of functional data lanes and thereby maintaining high productivity while still ensuring reliable fault isolation.
3Device complexity
If a fixed link width is used in the main band, then implementation simplicity is maintained, but adaptability to different traffic demands and fault conditions decreases
Solution Approach 1:
The link width is configured dynamically based on traffic demands and detected fault conditions rather than being fixed. The system can adjust the number of active data lanes in each direction independently, providing adaptability to different operational scenarios while maintaining a relatively simple implementation by building upon existing interconnect standards and using straightforward fault detection and configuration mechanisms.
Solution Approach 2:
The system changes the operational parameters (link width, number of active data lanes) based on traffic requirements and fault conditions. This parameter adaptation allows the main band to be versatile and responsive to different scenarios while maintaining implementation simplicity by using configurable parameters rather than requiring complex hardware redesigns.
Data Source
AI summary
Aspects relate to variable link width in two directions for a main band chip module connection. In one aspect, a different set of data lines is active for transmit and receive data lines. In one example a method includes sending an enable request from a module of a first die to a module partner of a second die through a sideband to operate a main band of a die-to-die connection that connects the first die module to the second die module partner at a specified link width, the specified link width having a specified set of data lines of the main band. An enable response is received from the module partner through the sideband to operate the main band at the specified link width and data is communicated with the module partner through the main band using the specified link width in response to receiving the enable response.