Neural Network Task Scheduling for Computing Resource Load Balance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing solutions fail to effectively utilize the processing capacity of remote computing resources in a resource pool, leading to idleness and workload unbalance, particularly in handling computing tasks like neural network model processing.
Innovation Solution
A method that involves obtaining status information of multiple computing resources, determining configuration information for neural network model layers, and selecting suitable resources based on this information and parameter data to efficiently process computing tasks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If existing technical solutions are used to process computing tasks, then the processing can be completed, but the processing capacity of remote computing resources is not effectively utilized, leading to resource idleness and workload unbalance
Solution Approach 1:
The patent implements dynamic resource allocation by continuously monitoring the status information (workload, resource availability) of multiple computing resources in the pool and adapting the task assignment in real-time. The scheduling system dynamically selects which computing resource receives which neural network layer processing based on current conditions, ensuring optimal utilization and preventing resource idleness.
Solution Approach 2:
The system changes the parameter of resource assignment by considering multiple factors including computing resource status information, layer parameter data (such as computational requirements of different neural network layers), and matching them optimally. This parameter-based matching approach transforms static resource allocation into a flexible, condition-based assignment system that maximizes resource utilization.
2Productivity
If computing tasks are distributed to multiple computing resources, then processing capacity is increased, but workload unbalance occurs among resources
Solution Approach 1:
The patent applies local quality by assigning different neural network layers to different computing resources based on their specific capabilities and current status. Each computing resource is optimally matched with layers that match its computational characteristics and available capacity, creating a specialized division of labor that balances the workload across heterogeneous resources while maximizing overall processing capacity.
3Productivity
If neural network model processing is distributed across multiple layers and resources, then processing efficiency is improved, but the complexity of resource selection and management increases
Solution Approach 1:
The system implements self-service by automatically monitoring computing resource status information, analyzing layer parameter data, and making intelligent matching decisions without manual intervention. The scheduling mechanism autonomously selects optimal computing resources for each neural network layer based on current system state, reducing management complexity while maintaining high processing efficiency through automated resource orchestration.
Data Source
AI summary
Implementations of the present disclosure relate to a method, apparatus and computer program product for processing a computing task. The method comprises: obtaining status information of multiple computing resources; in response to receiving a neural network model-based computing task, determining configuration information of multiple layers associated with the neural network model; obtaining parameter data associated with at least one part of the multiple layers on the basis of the configuration information; and based on the status information and the parameter data, selecting from the multiple computing resources a group of computing resources for processing the computing task. According to the example implementations of the present disclosure, multiple computing resources may be utilized sufficiently, and it may be guaranteed that a load balance may be stricken between the multiple computing resources.