SoC DVFS Algorithm Adjusting Frequency by Network Throughput
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Dynamic voltage frequency scaling (DVFS) methods in mobile devices do not effectively consider data transmission rates, leading to inefficient power management and performance optimization.
Innovation Solution
A system-on-chip with a network input/output device, central processing unit, performance monitoring device, and dynamic voltage frequency scaling unit that learns correlations between network throughput and processing performance to estimate data transmission rates and dynamically adjust the DVFS algorithm settings, thereby controlling the operation frequency of the central processing unit.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If dynamic voltage frequency scaling is applied to each IP block independently, then power consumption is reduced, but data transmission rate requirements are not considered leading to performance degradation
Solution Approach 1:
The patent implements a feedback mechanism where the performance monitoring device continuously monitors network throughput and processing performance, learns their correlation, and uses this information to dynamically adjust DVFS settings. The controller receives feedback about actual system performance and adjusts voltage/frequency settings accordingly, ensuring both power efficiency and data transmission rate requirements are met
Solution Approach 2:
The patent dynamically changes the parameters of the DVFS algorithm based on learned correlations between network throughput and processing performance. By adjusting voltage and frequency parameters according to actual data transmission rate requirements rather than using fixed IP-block-level settings, the system optimizes both power consumption and performance
2Productivity
If processing speed is increased to meet standardized performance requirements, then data throughput improves, but power consumption increases
Solution Approach 1:
The patent transitions from static DVFS settings to dynamic adjustment by continuously monitoring network throughput and processing performance. The system adapts processing speed dynamically based on actual workload requirements, increasing speed only when necessary to meet performance standards while reducing speed during lower-demand periods to conserve power
Solution Approach 2:
The performance monitoring device provides continuous feedback about processing performance and network throughput, enabling the controller to adjust power consumption in real-time based on actual needs rather than using fixed high-performance settings that would continuously consume maximum power
3Device complexity
If traditional DVFS methods are used without considering network throughput, then device complexity is reduced, but power management efficiency deteriorates
Solution Approach 1:
The system performs self-optimization by automatically learning the correlation between network throughput and processing performance through the performance monitoring device. This self-learning mechanism eliminates the need for complex manual configuration while achieving superior power management efficiency compared to traditional DVFS methods
Solution Approach 2:
The performance monitoring device establishes a feedback loop that automatically adjusts DVFS settings based on learned correlations, improving power management efficiency without requiring complex external control systems or manual intervention
Data Source
AI summary
Inventive concepts relates to a system-on-chip using a dynamic voltage frequency scaling and a method of operating the same. The method of operating the system-on-chip may include learning a correlation between network throughput of a network input/output device receiving data packets and processing performance of a central processing unit processing the data packets, estimating a data transmission rate of the data packets based on a learning result of the correlation, dynamically changing setting information of a dynamic voltage frequency scaling algorithm based on the estimated data transmission rate, and controlling an operation frequency of the central processing unit according to the dynamic voltage frequency scaling algorithm. According to example embodiments of inventive concepts, the dynamic voltage frequency scaling algorithm may dynamically be applied considering the data transmission rate of the data packets being received.