MPI Implementation in Embedded OS for Parallel Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current MPI implementations in embedded systems face limitations due to processor upgrades being unable to keep pace with demand, leading to inefficiencies in processing large applications, especially when the number of tasks exceeds the number of physical cores, resulting in prolonged operation times.
Innovation Solution
A system and method for MPI implementation in an embedded operating system that divides complex computing processes into segments, distributes them across multiple MPI nodes for parallel processing, and collects results automatically, utilizing a process handling module, underlying function library, and MPI API interface to optimize hardware usage and improve performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the number of tasks is increased to handle large applications, then the processing capability is improved, but the operation time increases due to time-division mode on limited physical cores
Solution Approach 1:
The patent divides a large application into multiple sub-applications, which are then distributed across multiple MPI nodes for parallel execution. This segmentation allows the system to process large applications efficiently by distributing the workload rather than handling everything sequentially on a single processor.
Solution Approach 2:
The patent transitions from single-node sequential processing to multi-node parallel processing, adding a spatial dimension to the computation. By distributing sub-applications across multiple MPI nodes that can execute simultaneously, the system overcomes the limitation of physical cores through networked parallelism.
2Speed
If more powerful hardware processors are used to handle large applications, then the processing speed is improved, but the hardware cost and complexity increase
Solution Approach 1:
The patent combines multiple standard processors into a distributed computing system using MPI. Instead of relying on a single powerful processor, the system merges the capabilities of multiple standard processors across different nodes, achieving high processing speed through coordinated parallel execution rather than through hardware upgrades.
Solution Approach 2:
The patent creates a universal computing architecture where standard processors can function as MPI nodes. This multi-functional approach allows ordinary processors to participate in high-performance computing tasks through the MPI framework, eliminating the need for specialized or high-end hardware while maintaining processing speed.
3Loss of time
If multi-thread parallel processing is used to utilize hardware capacity, then the computing time is reduced, but the performance improvement is limited by the number of physical cores
Solution Approach 1:
The patent moves beyond intra-node multi-threading to inter-node parallel processing. By distributing sub-applications across multiple MPI nodes that can execute independently and simultaneously, the system achieves performance improvement that is not constrained by the physical core count of a single processor.
Data Source
AI summary
The disclosure discloses a system and method for Message Passing Interface (MPI) implementation in an embedded operating system, applicable to a communication system including an application layer, an MPI layer, and an operating system layer, where an MPI application at the application layer is an application allocated to one or more MPI nodes; the Mil layer includes a process handling module, an underlying function library, and an MPI Application Programming Interface (API) to be invoked by the MPI application; and the process handling module receives a command of a user to invoke the MPI application, starts the application in response to the command, and invokes a service function of the embedded real-time operating system at the operating system layer through the underlying function library to control the one or more MPI node to execute and finish sub-tasks corresponding to the MPI application on the respective nodes.