Manycore Chip Data Transmission Redundancy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Manycore processors, which host multiple functions on a single chip to reduce weight and size while increasing computational capabilities, face reliability issues due to common failure modes, particularly in safety-critical applications like aircraft control, where a chip failure can lead to simultaneous loss of multiple functions.
Innovation Solution
Implementing data transmission between functions on multiple clusters using at least two distinct communication systems, such as network on a chip (NoC) and parallel bus (AXI), to ensure redundancy and maintain system functionality in case of failure, with data being transmitted on both communication links to ensure continued operation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Volume of moving object
If multiple functions are hosted on the same chip to reduce weight and size, then integration is improved, but reliability deteriorates due to common failure modes
Solution Approach 1:
The patent segments the system into multiple independent clusters (first cluster, second cluster, third cluster) distributed across different chips. Each cluster hosts critical functions independently, so that a failure in one cluster does not propagate to other clusters. This spatial segmentation across multiple chips reduces common failure modes while maintaining integration benefits.
Solution Approach 2:
The patent applies local quality by implementing the same critical function (second function) in multiple locations (second cluster on second chip, third cluster on third chip) with different reliability characteristics. This redundancy at the local level ensures that if one instance fails, another can take over, improving overall system reliability while maintaining compact integration.
2Device complexity
If data is transmitted through a single communication system between clusters, then device complexity is reduced, but reliability deteriorates due to single point of failure
Solution Approach 1:
The patent extracts the communication function into multiple independent communication systems (first communication system, second communication system) that operate in parallel. By separating the data transmission path into multiple independent channels, the system eliminates single points of failure in communication while maintaining manageable complexity through standardized interfaces.
Solution Approach 2:
The patent changes the parameter of communication redundancy by implementing data transmission through multiple communication systems simultaneously. The same data is transmitted through different communication channels (first communication system and second communication system), providing fault tolerance without significantly increasing complexity through systematic design.
3Reliability
If critical functions are implemented on multiple clusters across different chips, then reliability is improved through redundancy, but device complexity increases
Solution Approach 1:
The patent applies universality by implementing the same critical function (second function) across multiple clusters (second cluster, third cluster) that can perform identical operations. This multi-functionality allows any of the redundant clusters to take over if another fails, improving reliability while managing complexity through standardized function implementation across different chips.
Solution Approach 2:
The patent implements preliminary action by pre-configuring redundant clusters (second cluster on second chip, third cluster on third chip) with the same critical function before failure occurs. This advance preparation ensures that if a failure happens, the redundant clusters are already ready to take over immediately, improving reliability without requiring complex real-time reconfiguration.
Data Source
AI summary
A method for transmitting data between functions implemented on a first electronic chip of the manycore type. The first electronic chip includes a plurality of execution cores, the execution cores being grouped in clusters, the clusters being interconnected by at least two communication systems. The data transmission method includes the steps of: implementing a first function on a first cluster; implementing a second function on a second cluster, characterised in that the second function is also implemented on a third cluster distinct from the first and second clusters; and transmitting at least one data item between the first function and the second function.