Neural Network Power Optimization via Dynamic Throughput Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Systems that employ artificial neural networks face challenges in optimizing power usage while meeting strict quality-of-service requirements, often constrained by limited resources such as battery power and hardware limitations.
Innovation Solution
A computer-implemented method that identifies and analyzes artificial neural networks to determine execution metrics, then configures processor clock speeds and throughput to minimize power consumption while ensuring quality-of-service demands are met, by using modules to identify, analyze, determine, and optimize the processing of artificial neural networks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If processor clock speed is increased to meet quality-of-service requirements, then processing throughput is improved, but power consumption increases
Solution Approach 1:
The system dynamically adjusts processor clock speed based on real-time quality-of-service metrics and execution metrics. The optimization module continuously monitors system performance and modifies processor configuration parameters to maintain required throughput while minimizing power consumption, rather than using a fixed high clock speed setting
Solution Approach 2:
The system changes physical parameters of the processor (clock speed, voltage) based on calculated optimal values derived from execution metrics and quality-of-service requirements. By precisely controlling these parameters to match actual workload demands, the system avoids unnecessary power consumption while maintaining required processing throughput
2Use of energy by moving object
If processor configuration is optimized to minimize power consumption, then energy efficiency is improved, but processing throughput may decrease
Solution Approach 1:
The system implements a feedback loop where quality-of-service metrics are continuously monitored and fed back to the optimization module. This feedback mechanism ensures that processor configuration adjustments maintain throughput requirements while achieving power savings, preventing the throughput degradation that would occur with simple power-saving modes
Solution Approach 2:
The system performs preliminary analysis of the artificial neural network to determine execution metrics (operations per input) before actual execution. This advance knowledge allows the optimization module to pre-calculate the minimum required processing throughput and configure the processor accordingly, ensuring both power efficiency and throughput requirements are met from the start
Data Source
AI summary
The disclosed computer-implemented method may include (i) identifying an artificial neural network that processes each input to the artificial neural network in a fixed number of operations, (ii) performing an analysis on the artificial neural network to determine an execution metric that represents the fixed number of operations performed by the artificial neural network to process each input, (iii) determining a quality-of-service metric for an executing system that executes the artificial neural network, and (iv) optimizing power consumption of the executing system by configuring, based on the execution metric and the quality-of-service metric, a processing throughput of at least one physical processor of the executing system, thereby causing the executing system to execute the artificial neural network at a rate that satisfies the quality-of-service metric while limiting the power consumption of the executing system. Various other methods, systems, and computer-readable media are also disclosed.