Neural Network Processor Pause Parameter Adjustment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In neural network processing, long-term operation pauses due to bandwidth limitations and data dependencies lead to poor balance in power consumption distribution, causing issues with power integrity.
Innovation Solution
A method and apparatus for processing neural network feature maps by adjusting pause parameters of operation units in a neural network processor, based on pause state information, to achieve a more balanced distribution of pause times and power consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If operation units process neural network feature maps continuously, then processing throughput is improved, but power consumption becomes unbalanced causing power integrity problems
Solution Approach 1:
The patent implements periodic pause actions for operation units based on data dependency analysis. The controller determines pause state information and adjusts pause parameters to create periodic idle periods, which balance power consumption across different operation units while maintaining processing throughput through pipelined operation scheduling.
2Use of energy by stationary object
If operation units pause frequently to wait for data dependencies, then power consumption balance is improved, but processing efficiency deteriorates
Solution Approach 1:
The patent dynamically adjusts pause parameters based on real-time pause state information and data dependency relationships. The controller adaptively determines optimal pause timing and duration for each operation unit, transforming static pause schedules into dynamic, condition-based pause control that balances power consumption without significantly impacting processing efficiency.
Solution Approach 2:
The controller performs preliminary analysis of data dependency relationships before execution, pre-determining pause state information and pause parameters for operation units. This preliminary planning allows operation units to pause at optimal moments that balance power consumption while minimizing impact on overall processing efficiency.
3Use of energy by stationary object
If pause parameters are adjusted to balance power consumption, then power integrity is improved, but operation control complexity increases
Solution Approach 1:
The patent implements a feedback mechanism where the controller continuously monitors pause state information from operation units and adjusts pause parameters accordingly. This closed-loop control system automatically balances power consumption across operation units without requiring complex manual configuration, improving power integrity while keeping control complexity manageable through automated feedback-driven adjustment.
Data Source
Figure 1~3
Figure 4~5
Figure 6~7
AI summary
Embodiments of the present disclosure disclose a method and an apparatus for processing a neural network feature map using a neural network processor, and a device. The method includes: determining an operation instruction for calculating a to-be-processed feature map; determining pause state information of an operation unit in the neural network processor for executing the operation instruction; adjusting, based on the pause state information, a first pause parameter of the operation unit for executing the operation instruction to determine an adjusted second pause parameter; and controlling, based on the second pause parameter, the operation unit to execute the operation instruction to obtain a processing result corresponding to the to-be-processed feature map. According to the embodiments of the present disclosure, the first pause parameter can be adjusted, so that time distribution and/or spatial distribution for the operation unit to pause operations become more balanced.