Computing Control Device Accuracy Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In OpenVX computing control devices, unnecessary calculations occur due to high processing accuracy specified for preceding nodes, leading to performance degradation, as users are not aware of using processing functions with higher accuracy than required for subsequent nodes.
Innovation Solution
A computing control device analyzes the connection relationship between nodes in a graph, calculates the processing accuracy of preceding nodes based on the required accuracy of subsequent nodes, and replaces nodes with processing functions having sufficient accuracy to execute the graph, thereby optimizing processing efficiency without requiring users to consider accuracy levels.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If processing functions with high accuracy are used for preceding nodes, then processing reliability is improved, but processing time increases due to unnecessary calculations
Solution Approach 1:
The patent dynamically changes the accuracy parameter of processing functions based on the specific requirements of subsequent nodes. The graph manager analyzes the graph structure to determine the minimum required accuracy for each node's output, then selects or adjusts processing functions to match that requirement rather than always using high-accuracy functions. This resolves the contradiction by adapting the accuracy parameter to the actual needs of the computation graph.
Solution Approach 2:
The patent applies the principle of partial action by using only the necessary amount of processing accuracy required by subsequent nodes, rather than applying excessive high-accuracy processing uniformly across all preceding nodes. The system calculates the sufficient accuracy level needed for each node based on downstream requirements, avoiding unnecessary computational precision where it is not needed.
2Manufacturing precision
If processing functions with high accuracy are specified, then manufacturing precision is improved, but productivity decreases due to increased computational load
Solution Approach 1:
The system dynamically adjusts the accuracy parameter of processing functions based on the specific requirements of subsequent nodes in the graph. By analyzing the graph structure and determining the minimum required accuracy for each node's output, the system selects processing functions that provide sufficient but not excessive precision, thereby improving processing efficiency while maintaining necessary accuracy levels.
Solution Approach 2:
The patent applies partial action by using only the necessary amount of processing accuracy required by subsequent nodes. Instead of uniformly applying high-accuracy processing to all nodes, the system calculates the sufficient accuracy level needed for each node based on downstream requirements, avoiding unnecessary computational precision where it is not needed.
3Productivity
If users manually optimize processing accuracy for each node, then processing efficiency is improved, but device complexity increases due to user burden
Solution Approach 1:
The graph manager automatically performs the optimization of processing accuracy levels by analyzing the graph structure and determining appropriate accuracy requirements for each node. This self-service mechanism eliminates the need for users to manually optimize accuracy settings, as the system autonomously identifies the minimum required accuracy based on subsequent node requirements and selects appropriate processing functions accordingly.
Solution Approach 2:
The system uses feedback from the graph structure analysis to automatically determine optimal processing accuracy levels. By examining the connection relationships between nodes and the requirements of subsequent nodes, the graph manager receives feedback about what accuracy level is sufficient, then automatically configures processing functions to match those requirements without user intervention.
Data Source
AI summary
A computing control device includes extracting a node having a plurality of processing functions having accuracies different from one another from a graph; calculating a required accuracy of a node subsequent to the node that has been extracted; selecting a processing function having a minimum accuracy equal to or higher than the required accuracy that has been calculated from among the plurality of processing functions included in the node that has been extracted; and setting the processing function that has been selected as the processing function of the node that has been extracted.