ANN Subgraph Resource Allocation for Adaptive Hardware Scheduling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing artificial neural network (ANN) systems face inefficiencies in processing operations due to the increasing demand for efficient hardware allocation and resource management in complex neural network models.
Innovation Solution
A method and module for dynamically allocating hardware and adjusting resource settings based on resource determination triggers, control signals, and metadata to optimize operations on subgraphs within neural networks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If hardware allocation and resource settings are fixed for neural network operations, then device complexity is reduced and ease of operation is improved, but processing efficiency and adaptability to different operation states deteriorate
Solution Approach 1:
The patent implements dynamic hardware allocation and resource setting adjustment based on operation states. The artificial neural network module monitors operation states and dynamically modifies hardware allocation and resource settings during execution, transforming the system from static to dynamic configuration to improve processing efficiency for different neural network operations.
Solution Approach 2:
The system changes hardware allocation parameters and resource setting parameters based on detected operation states. By adjusting these parameters dynamically, the system optimizes performance for different types of neural network operations without requiring complete system redesign, thus improving productivity while managing complexity through parameterization.
2Adaptability or versatility
If hardware resources are allocated for all possible operation states, then adaptability is improved, but device complexity and power consumption increase
Solution Approach 1:
The system dynamically adjusts resource allocation based on actual operation states rather than maintaining fixed high-level resource allocation for all possible states. This dynamic approach allows the system to adapt to different operation requirements while consuming power only when and where needed, reducing overall power consumption while maintaining high adaptability.
Solution Approach 2:
The artificial neural network module performs preliminary detection of operation states and pre-adjusts hardware allocation and resource settings before executing operations. This preliminary action ensures that resources are optimally configured for the specific operation at hand without maintaining excessive resources in standby, thereby reducing power consumption while preserving adaptability.
3Productivity
If dynamic hardware allocation and resource setting adjustment are implemented, then neural network operation efficiency is improved, but device complexity and control difficulty increase
Solution Approach 1:
The artificial neural network module is designed as a universal component that handles multiple functions: operation state detection, hardware allocation determination, resource setting adjustment, and operation execution. By consolidating these functions into a single multi-functional module, the system improves neural network operation efficiency while minimizing the increase in overall device complexity through functional integration.
Data Source
AI summary
A method for an artificial neural network operation on a plurality of subgraphs may include generating a resource determination trigger corresponding to a target subgraph among the plurality of subgraphs included in a target neural network model; generating a control signal for hardware allocated to the target subgraph and driving resource settings in response to the resource determination trigger; changing, based on the control signal, at least one of hardware allocated to the target subgraph and driving resource settings; and performing an operation on the target subgraph based on the changed hardware and driving resource settings.